Pediatric Pathology Hyperkalemia Study: Risk Factors associated with Transfusion Associated Hyperkalemia in Pediatric Population
End-to-end electronic R-markdown notebook
Chisa Yamada, Maureen Edelson, Angela Lee, Nabiha Huq Saifee, Ivo D. Dinov, and SOCR
09/02/22 September 2022
A rare, yet potentially life-threatening, disease associated with red blood cell (RBC) transfusion. A prior Transfusion-associated hyperkalemia (TAH) study reported total 35 pediatric patients who had hyperkalemia during or within 12 hours after the RBC transfusion from four facilities in 1-year period (2015-2016). The researchers examined the overall prevalence of TAH in pediatric population is 0.93%, 1-days mortality after TAH is as high as 20% in such patients, all patients who experienced TAH were severely ill with multiple comorbidities, transfused volume within 12 hours before the TAH occurrence was 24%, but infusion rate was high (19.57 ml/Kg/hour).
In this new research, we collected data from the pediatric patients with RBC transfusion without TAH serving as a control group to contrast patients with TAH and without TAH. We performed model-based and model-free multivariable analyses to predict TAH.
1 Background
This is a pediatric study of Transfusion Associated Hyperkalemia (TAH). The multi-institutional and retrospective study is trying to examine the risk factors associated with fatal hyperkalemia following RBC transfusion in pediatric population (<18 years old). A prior study is already published in 2021.
This follow up study aims to investigate the differences between a control group (\(n=60\)) and the TAH group (\(n=35\)). We would like to:
- Compare the two cohorts and quantify their differences determine which factors play a key role in discriminating the populations,
- Identify risk factors associated with TAH (e.g., age, body weight, transfusion volume, infusion rate, irradiation of the unit, etc).
2 Data Import
Using the latest dataset:
rev8_Data_N_95_TransfusionHyperkalemia.csv.
setwd(file.path("C:/Users/dinov/Desktop/Ivo.dir/Research/UMichigan/Collaborations/2021/Chisa_Yamada_Pediatric_Pathology_Hyperkalemia_Study/TAH_Paper_2022/R3/"))
TAH_data <- read.csv("rev8_Data_N_95_TransfusionHyperkalemia.csv", head=T)
factorVarNames <- c('Case' ,'Group', 'TAH_occurrence', 'Mortality1Day', 'Sex', 'Race', 'Access', 'Dialysis', 'Cardiac_dysf',
'Respiratory_dysf', 'Kidney_dysf', 'Liver_dysf', 'Intracranial_diso', 'Orthopedics', 'Anemia',
'Hemoglobinopathy', 'Sepsis', 'ECMO', 'CPB', 'Prematurity', 'RBC_Unit_type',
'Leukoreduction', 'Irradiation', 'Washed', 'Plasma_reduction')
TAH_data[ , factorVarNames] <- lapply(TAH_data[ , factorVarNames] , factor)
# Impute the missing data
summary(TAH_data)## Case Group TAH_occurrence Mortality1Day Sex Race
## 1 : 2 1:22 0:60 0 :87 1:43 1 :33
## 2 : 2 2:17 1:35 1 : 7 2:52 2 :21
## 3 : 2 3:24 missing : 1 3 : 6
## 4 : 2 4:13 4 : 1
## 5 : 2 5:19 5 :27
## 6 : 2 Missing: 7
## (Other):83
## Access Age_days Weight eTBV
## 1 :26 Min. : 1 Min. : 0.63 Min. : 63.0
## 2 :36 1st Qu.: 43 1st Qu.: 3.48 1st Qu.: 284.5
## Missing:33 Median : 723 Median :11.00 Median : 735.0
## Mean :1784 Mean :18.22 Mean :1254.6
## 3rd Qu.:3264 3rd Qu.:27.25 3rd Qu.:1848.0
## Max. :6346 Max. :74.40 Max. :4836.0
##
## Pre_transfusion_K Cr_at_transfusion Cardiac_dysf Respiratory_dysf
## Length:95 Length:95 0:74 0:60
## Class :character Class :character 1:21 1:35
## Mode :character Mode :character
##
##
##
##
## Kidney_dysf Liver_dysf Intracranial_diso Orthopedics Anemia Hemoglobinopathy
## 0:71 0:82 0:84 0:91 0:47 0:86
## 1:24 1:13 1:11 1: 4 1:48 1: 9
##
##
##
##
##
## Sepsis ECMO CPB Prematurity Dialysis RBC_Unit_type Age_of_RBC_unit
## 0:87 0:85 0:91 0:71 0:89 1 :29 Min. : 1.00
## 1: 8 1:10 1: 4 1:24 1: 6 2 :62 1st Qu.: 6.00
## 3 : 3 Median :10.00
## Missing: 1 Mean :12.97
## 3rd Qu.:17.00
## Max. :42.00
##
## Leukoreduction Irradiation Washed Plasma_reduction tV_12_weight_ml_Kg
## 0: 3 0:29 0:93 0:94 Min. : 4.39
## 1:92 1:66 1: 2 1: 1 1st Qu.: 10.54
## Median : 15.22
## Mean : 27.58
## 3rd Qu.: 21.81
## Max. :283.65
##
## tV_12_eTBV Infusion_rate_hour_Pt_weight Infusion_rate_hour_eTBV
## Min. :0.0600 Length:95 Length:95
## 1st Qu.:0.1450 Class :character Class :character
## Median :0.2100 Mode :character Mode :character
## Mean :0.3802
## 3rd Qu.:0.2800
## Max. :4.3600
##
## Time_from_irradition_to_transfusion_min
## Length:95
## Class :character
## Mode :character
##
##
##
## # Quick code to replace missing values with the mean
df_TAH_data_NA <- data.frame(sapply(TAH_data[ , -1], function(x) ifelse(x=="Missing", "NA",x)))
df_TAH_data_numeric <- as.data.frame(sapply(df_TAH_data_NA, as.numeric))
df_TAH_data_impute_mean <- data.frame(sapply(df_TAH_data_numeric, function(x) ifelse(is.na(x), as.integer(median(x, na.rm = TRUE)), x)))
df_TAH_data_impute_mean <- subset(df_TAH_data_impute_mean,
# select = -c(tV_12_weight_ml_Kg, Infusion_rate_hour_Pt_weight, eTBV))
select = -c(eTBV))Note: Make a separate chi-square test for
access, to address the significant missingness…
3 Summary Statystics (Collective and Grouping by Age and Cohort)
library(DT)
# Round a single data-frame
round_df <- function(x, digits) {
# round all numeric variables
# x: data frame
# digits: number of digits to round
numeric_columns <- sapply(x, mode) == 'numeric'
x[numeric_columns] <- round(x[numeric_columns], digits)
x
}
# Round a list of data-frames
round_list_df <- function(x, digits) {
# round all numeric variables
# x: data frame
# digits: number of digits to round
n <- length(x)
for (i in 1:n) {
numeric_columns <- sapply(x[[i]], mode) == 'numeric'
x[[i]][numeric_columns] <- round(x[[i]][numeric_columns], digits)
}
x
}3.1 Multivariate Relations and Variable Inflation Factors (VIF)
VIF of \(1.5\) indicates that the variance of a particular coefficient is \(50\%\) bigger than the expected variance when there is no multicollinearity, i.e., when there is no correlation with other predictors. Interpreting the (raw) variance inflation factor (VIF) values is based on:
- \(VIF\sim 1\), not correlated.
- \(1\leq VIF\leq 5\), moderately correlated.
- \(VIF > 5\), highly correlated.
The adjusted general VIF estimates account for the dimension of the confidence ellipsoid, \(GVIF^{\frac{1}{2*df}}\), where \(df\) is the degrees of freedom associated with the variable.
car::vif(glm(as.factor(TAH_occurrence) ~ ., data=df_TAH_data_impute_mean, family = "binomial")) ## Group Mortality1Day
## 7016.936412 15.415438
## Sex Race
## 23.687356 46.961407
## Access Age_days
## 226.662073 1202.468822
## Weight Pre_transfusion_K
## 637.752284 7.334058
## Cr_at_transfusion Cardiac_dysf
## 364.329292 121.928097
## Respiratory_dysf Kidney_dysf
## 277.336190 97.775334
## Liver_dysf Intracranial_diso
## 3.950198 91.769367
## Orthopedics Anemia
## 180.362740 240.307225
## Hemoglobinopathy Sepsis
## 2.299605 52.395145
## ECMO CPB
## 9.779380 293.258070
## Prematurity Dialysis
## 545.239945 157.676999
## RBC_Unit_type Age_of_RBC_unit
## 458.436895 329.331761
## Leukoreduction Irradiation
## 11.196305 228.718812
## Washed Plasma_reduction
## 2.547648 1.160800
## tV_12_weight_ml_Kg tV_12_eTBV
## 3090.332264 2113.421848
## Infusion_rate_hour_Pt_weight Infusion_rate_hour_eTBV
## 4058.700623 5543.985431
## Time_from_irradition_to_transfusion_min
## 15.4301413.2 Complete Sample Statistics
data_tbl <- round_df(psych::describe(df_TAH_data_impute_mean), 3)
datatable(data_tbl, caption="Overall Study Descriptive Statistics")3.3 Grouping by Age and Diagnosis
3.3.1 Grouping Solely by Diagnosis
# Group by df_TAH_data_impute_mean ~ Group
# Compute the descriptive statistics for each of the 2 TAH_occurrence groups (Dx)
data_tbls <- round_list_df(psych::describe(df_TAH_data_impute_mean ~ TAH_occurrence), 3)
titleDF <- list()
df_list <- list()
for (trait in 1:2) {
if (trait==1) {
titleDF[[trait]] <- paste0("Descriptive Statistics (Dx Cohort=Controls)")
} else {
titleDF[[trait]] <- paste0("Descriptive Statistics (Dx Cohort=TAH Patients)")
}
df_list[[ trait]] <- datatable(data_tbls[[trait]], caption=titleDF[[trait]])
}
df_list[[1]]df_list[[2]]3.3.2 Grouping by \(Age\times Diagnosis\)
# Group by df_TAH_data_impute_mean ~ Group + TAH_occurrence
# Compute the descriptive statistics for each of the 10 = 5*2 (Group * TAH_occurrence) data-frames
data_tbls <- round_list_df(psych::describe(df_TAH_data_impute_mean ~ Group + TAH_occurrence), 3)
titleDF <- list()
df_list <- list()
for (gr in 1:5) {
for (trait in 1:2) {
if (trait==1) {
titleDF[[(trait-1)*5 + gr]] <- paste0("Descriptive Statistics (Age Group=", gr, ", Dx Cohort=Controls)")
} else {
titleDF[[(trait-1)*5 + gr]] <- paste0("Descriptive Statistics (Age Group=", gr, ", Dx Cohort=TAH Patients)")
}
df_list[[(trait-1)*5 + gr]] <-
datatable(data_tbls[[(trait-1)*5 + gr]], caption=titleDF[[(trait-1)*5 + gr]])
# data_tbls[[(trait-1)*5 + gr]]
}
}
df_list[[1]]df_list[[2]]df_list[[3]]df_list[[4]]df_list[[5]]df_list[[6]]df_list[[7]]df_list[[8]]df_list[[9]]df_list[[10]]3.4 Other Statistics
aggregate(cbind(Group,TAH_occurrence) ~ Sex + Race, df_TAH_data_impute_mean, sum)## Sex Race Group TAH_occurrence
## 1 1 1 54 26
## 2 2 1 41 20
## 3 1 2 26 12
## 4 2 2 57 27
## 5 1 3 6 2
## 6 2 3 10 5
## 7 1 4 2 2
## 8 1 5 39 21
## 9 2 5 40 15with(df_TAH_data_impute_mean, tapply(Sex, list(group=Group,TAH=TAH_occurrence), sum))## TAH
## group 1 2
## 1 19 15
## 2 21 7
## 3 20 16
## 4 19 1
## 5 18 11library(dplyr)
# df_TAH_data_impute_mean %>% group_by(as.factor(Group), as.factor(TAH_occurrence)) %>% summarize(sum)
library(reshape2)
dcast(df_TAH_data_impute_mean, Group ~ TAH_occurrence, fun.aggregate = length)## Group 1 2
## 1 1 12 10
## 2 2 12 5
## 3 3 12 12
## 4 4 12 1
## 5 5 12 7# dplyr::count(df_TAH_data_impute_mean, c(Group,TAH_occurrence))
cross <- table(Group=df_TAH_data_impute_mean$Group, TAH=df_TAH_data_impute_mean$TAH_occurrence)
addmargins(cross)## TAH
## Group 1 2 Sum
## 1 12 10 22
## 2 12 5 17
## 3 12 12 24
## 4 12 1 13
## 5 12 7 19
## Sum 60 35 95cross <- prop.table(cross, 2)
addmargins(cross)## TAH
## Group 1 2 Sum
## 1 0.20000000 0.28571429 0.48571429
## 2 0.20000000 0.14285714 0.34285714
## 3 0.20000000 0.34285714 0.54285714
## 4 0.20000000 0.02857143 0.22857143
## 5 0.20000000 0.20000000 0.40000000
## Sum 1.00000000 1.00000000 2.00000000group_by(df_TAH_data_impute_mean, TAH_occurrence) %>%
summarise(
count = n(),
Irradiation_mean = mean(Irradiation, na.rm = TRUE),
Irradiation_sd = sd(Irradiation, na.rm = TRUE)
)## # A tibble: 2 × 4
## TAH_occurrence count Irradiation_mean Irradiation_sd
## <dbl> <int> <dbl> <dbl>
## 1 1 60 1.68 0.469
## 2 2 35 1.71 0.458group_by(df_TAH_data_impute_mean, Group) %>%
summarise(
count = n(),
Irradiation_mean = mean(Irradiation, na.rm = TRUE),
Irradiation_sd = sd(Irradiation, na.rm = TRUE),
tV_12_weight_ml_Kg_mean = mean(tV_12_weight_ml_Kg, na.rm = TRUE),
tV_12_weight_ml_Kg_sd = sd(tV_12_weight_ml_Kg, na.rm = TRUE)
)## # A tibble: 5 × 6
## Group count Irradiation_mean Irradiation_sd tV_12_weight_ml_Kg_mean tV_12_we…¹
## <dbl> <int> <dbl> <dbl> <dbl> <dbl>
## 1 1 22 1.64 0.492 33.5 30.3
## 2 2 17 1.82 0.393 33.5 57.3
## 3 3 24 1.75 0.442 36.1 59.0
## 4 4 13 1.54 0.519 12.0 4.53
## 5 5 19 1.68 0.478 15.4 16.9
## # … with abbreviated variable name ¹​tV_12_weight_ml_Kg_sd4 TAH Analysis
4.1 Table 2
4.1.1 Factor Variables (Chi-Square Test)
library(tidyverse)
for (i in 1:length(colnames(df_TAH_data_impute_mean))){
print (colnames(df_TAH_data_impute_mean)[i])
ifelse (colnames(df_TAH_data_impute_mean)[i] %in%
c("Age_days",
"Weight",
"Pre_transfusion_K",
"Cr_at_transfusion",
"Age_of_RBC_unit",
"tV_12_weight_ml_Kg",
"tV_12_eTBV",
"Infusion_rate_hour_Pt_weight",
"Infusion_rate_hour_eTBV",
"Time_from_irradition_to_transfusion_min"),
df_TAH_data_impute_mean[,i] <- as.numeric(df_TAH_data_impute_mean[,i]),
df_TAH_data_impute_mean[,i] <- as.factor(df_TAH_data_impute_mean[,i]))
}## [1] "Group"
## [1] "TAH_occurrence"
## [1] "Mortality1Day"
## [1] "Sex"
## [1] "Race"
## [1] "Access"
## [1] "Age_days"
## [1] "Weight"
## [1] "Pre_transfusion_K"
## [1] "Cr_at_transfusion"
## [1] "Cardiac_dysf"
## [1] "Respiratory_dysf"
## [1] "Kidney_dysf"
## [1] "Liver_dysf"
## [1] "Intracranial_diso"
## [1] "Orthopedics"
## [1] "Anemia"
## [1] "Hemoglobinopathy"
## [1] "Sepsis"
## [1] "ECMO"
## [1] "CPB"
## [1] "Prematurity"
## [1] "Dialysis"
## [1] "RBC_Unit_type"
## [1] "Age_of_RBC_unit"
## [1] "Leukoreduction"
## [1] "Irradiation"
## [1] "Washed"
## [1] "Plasma_reduction"
## [1] "tV_12_weight_ml_Kg"
## [1] "tV_12_eTBV"
## [1] "Infusion_rate_hour_Pt_weight"
## [1] "Infusion_rate_hour_eTBV"
## [1] "Time_from_irradition_to_transfusion_min"indFactor <- sapply(df_TAH_data_impute_mean, is.factor)
table(indFactor)## indFactor
## FALSE TRUE
## 10 24# Contingency Table - TAH vs. Kidney-dysfunction, e.g.,
# table(df_TAH_data_impute_mean$TAH_occurrence, df_TAH_data_impute_mean$Kidney_dysf)
contTable <- list()
for (var in c(4:length(factorVarNames))) {
print(paste0("TAH_occurrence vs ", factorVarNames[var]))
contTable[[var-3]] <- table(df_TAH_data_impute_mean$TAH_occurrence,
df_TAH_data_impute_mean[[factorVarNames[var]]],
dnn = list("TAH_occurrence", factorVarNames[var]))
}## [1] "TAH_occurrence vs Mortality1Day"
## [1] "TAH_occurrence vs Sex"
## [1] "TAH_occurrence vs Race"
## [1] "TAH_occurrence vs Access"
## [1] "TAH_occurrence vs Dialysis"
## [1] "TAH_occurrence vs Cardiac_dysf"
## [1] "TAH_occurrence vs Respiratory_dysf"
## [1] "TAH_occurrence vs Kidney_dysf"
## [1] "TAH_occurrence vs Liver_dysf"
## [1] "TAH_occurrence vs Intracranial_diso"
## [1] "TAH_occurrence vs Orthopedics"
## [1] "TAH_occurrence vs Anemia"
## [1] "TAH_occurrence vs Hemoglobinopathy"
## [1] "TAH_occurrence vs Sepsis"
## [1] "TAH_occurrence vs ECMO"
## [1] "TAH_occurrence vs CPB"
## [1] "TAH_occurrence vs Prematurity"
## [1] "TAH_occurrence vs RBC_Unit_type"
## [1] "TAH_occurrence vs Leukoreduction"
## [1] "TAH_occurrence vs Irradiation"
## [1] "TAH_occurrence vs Washed"
## [1] "TAH_occurrence vs Plasma_reduction"# addmargins(contTable[[8]]) # "Kidney_dysf"
# prop.table(contTable[[8]]) # "Kidney_dysf"
# chisq.test(contTable[[8]]) # "Kidney_dysf"
# fisher.test(contTable[[8]])
for (var in c(4:length(factorVarNames))) {
print(addmargins(contTable[[var-3]]))
print(prop.table(contTable[[var-3]]))
print(chisq.test(contTable[[var-3]]))
### Adjust p-values for multiple comparison
factorPvalueResults_adj <-
p.adjust(chisq.test(contTable[[var-3]])$p.value, method = 'fdr')
print(paste0("Adjusted p-value: ", factorPvalueResults_adj))
}## Mortality1Day
## TAH_occurrence 1 2 3 Sum
## 1 59 0 1 60
## 2 28 7 0 35
## Sum 87 7 1 95
## Mortality1Day
## TAH_occurrence 1 2 3
## 1 0.62105263 0.00000000 0.01052632
## 2 0.29473684 0.07368421 0.00000000
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 3]]
## X-squared = 13.395, df = 2, p-value = 0.001234
##
## [1] "Adjusted p-value: 0.00123421762995026"
## Sex
## TAH_occurrence 1 2 Sum
## 1 23 37 60
## 2 20 15 35
## Sum 43 52 95
## Sex
## TAH_occurrence 1 2
## 1 0.2421053 0.3894737
## 2 0.2105263 0.1578947
##
## Pearson's Chi-squared test with Yates' continuity correction
##
## data: contTable[[var - 3]]
## X-squared = 2.4431, df = 1, p-value = 0.118
##
## [1] "Adjusted p-value: 0.118042363651031"
## Race
## TAH_occurrence 1 2 3 4 5 Sum
## 1 20 17 5 0 18 60
## 2 13 11 1 1 9 35
## Sum 33 28 6 1 27 95
## Race
## TAH_occurrence 1 2 3 4 5
## 1 0.21052632 0.17894737 0.05263158 0.00000000 0.18947368
## 2 0.13684211 0.11578947 0.01052632 0.01052632 0.09473684
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 3]]
## X-squared = 3.071, df = 4, p-value = 0.546
##
## [1] "Adjusted p-value: 0.546022787542228"
## Access
## TAH_occurrence 1 2 Sum
## 1 18 42 60
## 2 8 27 35
## Sum 26 69 95
## Access
## TAH_occurrence 1 2
## 1 0.18947368 0.44210526
## 2 0.08421053 0.28421053
##
## Pearson's Chi-squared test with Yates' continuity correction
##
## data: contTable[[var - 3]]
## X-squared = 0.26493, df = 1, p-value = 0.6068
##
## [1] "Adjusted p-value: 0.606753866064084"
## Dialysis
## TAH_occurrence 1 2 Sum
## 1 57 3 60
## 2 32 3 35
## Sum 89 6 95
## Dialysis
## TAH_occurrence 1 2
## 1 0.60000000 0.03157895
## 2 0.33684211 0.03157895
##
## Pearson's Chi-squared test with Yates' continuity correction
##
## data: contTable[[var - 3]]
## X-squared = 0.064066, df = 1, p-value = 0.8002
##
## [1] "Adjusted p-value: 0.800180997258303"
## Cardiac_dysf
## TAH_occurrence 1 2 Sum
## 1 47 13 60
## 2 27 8 35
## Sum 74 21 95
## Cardiac_dysf
## TAH_occurrence 1 2
## 1 0.49473684 0.13684211
## 2 0.28421053 0.08421053
##
## Pearson's Chi-squared test with Yates' continuity correction
##
## data: contTable[[var - 3]]
## X-squared = 0, df = 1, p-value = 1
##
## [1] "Adjusted p-value: 1"
## Respiratory_dysf
## TAH_occurrence 1 2 Sum
## 1 35 25 60
## 2 25 10 35
## Sum 60 35 95
## Respiratory_dysf
## TAH_occurrence 1 2
## 1 0.3684211 0.2631579
## 2 0.2631579 0.1052632
##
## Pearson's Chi-squared test with Yates' continuity correction
##
## data: contTable[[var - 3]]
## X-squared = 1.1149, df = 1, p-value = 0.291
##
## [1] "Adjusted p-value: 0.291012661013251"
## Kidney_dysf
## TAH_occurrence 1 2 Sum
## 1 53 7 60
## 2 18 17 35
## Sum 71 24 95
## Kidney_dysf
## TAH_occurrence 1 2
## 1 0.55789474 0.07368421
## 2 0.18947368 0.17894737
##
## Pearson's Chi-squared test with Yates' continuity correction
##
## data: contTable[[var - 3]]
## X-squared = 14.051, df = 1, p-value = 0.0001779
##
## [1] "Adjusted p-value: 0.000177939153222725"
## Liver_dysf
## TAH_occurrence 1 2 Sum
## 1 56 4 60
## 2 26 9 35
## Sum 82 13 95
## Liver_dysf
## TAH_occurrence 1 2
## 1 0.58947368 0.04210526
## 2 0.27368421 0.09473684
##
## Pearson's Chi-squared test with Yates' continuity correction
##
## data: contTable[[var - 3]]
## X-squared = 5.2731, df = 1, p-value = 0.02166
##
## [1] "Adjusted p-value: 0.0216575110494646"
## Intracranial_diso
## TAH_occurrence 1 2 Sum
## 1 55 5 60
## 2 29 6 35
## Sum 84 11 95
## Intracranial_diso
## TAH_occurrence 1 2
## 1 0.57894737 0.05263158
## 2 0.30526316 0.06315789
##
## Pearson's Chi-squared test with Yates' continuity correction
##
## data: contTable[[var - 3]]
## X-squared = 0.92563, df = 1, p-value = 0.336
##
## [1] "Adjusted p-value: 0.336000872717627"
## Orthopedics
## TAH_occurrence 1 2 Sum
## 1 57 3 60
## 2 34 1 35
## Sum 91 4 95
## Orthopedics
## TAH_occurrence 1 2
## 1 0.60000000 0.03157895
## 2 0.35789474 0.01052632
##
## Pearson's Chi-squared test with Yates' continuity correction
##
## data: contTable[[var - 3]]
## X-squared = 6.1336e-31, df = 1, p-value = 1
##
## [1] "Adjusted p-value: 0.999999999999999"
## Anemia
## TAH_occurrence 1 2 Sum
## 1 24 36 60
## 2 23 12 35
## Sum 47 48 95
## Anemia
## TAH_occurrence 1 2
## 1 0.2526316 0.3789474
## 2 0.2421053 0.1263158
##
## Pearson's Chi-squared test with Yates' continuity correction
##
## data: contTable[[var - 3]]
## X-squared = 4.8638, df = 1, p-value = 0.02743
##
## [1] "Adjusted p-value: 0.0274255139026991"
## Hemoglobinopathy
## TAH_occurrence 1 2 Sum
## 1 52 8 60
## 2 34 1 35
## Sum 86 9 95
## Hemoglobinopathy
## TAH_occurrence 1 2
## 1 0.54736842 0.08421053
## 2 0.35789474 0.01052632
##
## Pearson's Chi-squared test with Yates' continuity correction
##
## data: contTable[[var - 3]]
## X-squared = 1.7392, df = 1, p-value = 0.1872
##
## [1] "Adjusted p-value: 0.187244274713748"
## Sepsis
## TAH_occurrence 1 2 Sum
## 1 57 3 60
## 2 30 5 35
## Sum 87 8 95
## Sepsis
## TAH_occurrence 1 2
## 1 0.60000000 0.03157895
## 2 0.31578947 0.05263158
##
## Pearson's Chi-squared test with Yates' continuity correction
##
## data: contTable[[var - 3]]
## X-squared = 1.4141, df = 1, p-value = 0.2344
##
## [1] "Adjusted p-value: 0.23437750497731"
## ECMO
## TAH_occurrence 1 2 Sum
## 1 56 4 60
## 2 29 6 35
## Sum 85 10 95
## ECMO
## TAH_occurrence 1 2
## 1 0.58947368 0.04210526
## 2 0.30526316 0.06315789
##
## Pearson's Chi-squared test with Yates' continuity correction
##
## data: contTable[[var - 3]]
## X-squared = 1.5837, df = 1, p-value = 0.2082
##
## [1] "Adjusted p-value: 0.208233429915851"
## CPB
## TAH_occurrence 1 2 Sum
## 1 58 2 60
## 2 33 2 35
## Sum 91 4 95
## CPB
## TAH_occurrence 1 2
## 1 0.61052632 0.02105263
## 2 0.34736842 0.02105263
##
## Pearson's Chi-squared test with Yates' continuity correction
##
## data: contTable[[var - 3]]
## X-squared = 0.00077675, df = 1, p-value = 0.9778
##
## [1] "Adjusted p-value: 0.977765605178571"
## Prematurity
## TAH_occurrence 1 2 Sum
## 1 46 14 60
## 2 25 10 35
## Sum 71 24 95
## Prematurity
## TAH_occurrence 1 2
## 1 0.4842105 0.1473684
## 2 0.2631579 0.1052632
##
## Pearson's Chi-squared test with Yates' continuity correction
##
## data: contTable[[var - 3]]
## X-squared = 0.1037, df = 1, p-value = 0.7474
##
## [1] "Adjusted p-value: 0.747429380160965"
## RBC_Unit_type
## TAH_occurrence 1 2 3 Sum
## 1 17 40 3 60
## 2 12 23 0 35
## Sum 29 63 3 95
## RBC_Unit_type
## TAH_occurrence 1 2 3
## 1 0.17894737 0.42105263 0.03157895
## 2 0.12631579 0.24210526 0.00000000
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 3]]
## X-squared = 2.0096, df = 2, p-value = 0.3661
##
## [1] "Adjusted p-value: 0.366119393126778"
## Leukoreduction
## TAH_occurrence 1 2 Sum
## 1 3 57 60
## 2 0 35 35
## Sum 3 92 95
## Leukoreduction
## TAH_occurrence 1 2
## 1 0.03157895 0.60000000
## 2 0.00000000 0.36842105
##
## Pearson's Chi-squared test with Yates' continuity correction
##
## data: contTable[[var - 3]]
## X-squared = 0.54191, df = 1, p-value = 0.4616
##
## [1] "Adjusted p-value: 0.461640301416673"
## Irradiation
## TAH_occurrence 1 2 Sum
## 1 19 41 60
## 2 10 25 35
## Sum 29 66 95
## Irradiation
## TAH_occurrence 1 2
## 1 0.2000000 0.4315789
## 2 0.1052632 0.2631579
##
## Pearson's Chi-squared test with Yates' continuity correction
##
## data: contTable[[var - 3]]
## X-squared = 0.0072383, df = 1, p-value = 0.9322
##
## [1] "Adjusted p-value: 0.93219907427448"
## Washed
## TAH_occurrence 1 2 Sum
## 1 58 2 60
## 2 35 0 35
## Sum 93 2 95
## Washed
## TAH_occurrence 1 2
## 1 0.61052632 0.02105263
## 2 0.36842105 0.00000000
##
## Pearson's Chi-squared test with Yates' continuity correction
##
## data: contTable[[var - 3]]
## X-squared = 0.12313, df = 1, p-value = 0.7257
##
## [1] "Adjusted p-value: 0.725666503561585"
## Plasma_reduction
## TAH_occurrence 1 2 Sum
## 1 59 1 60
## 2 35 0 35
## Sum 94 1 95
## Plasma_reduction
## TAH_occurrence 1 2
## 1 0.62105263 0.01052632
## 2 0.36842105 0.00000000
##
## Pearson's Chi-squared test with Yates' continuity correction
##
## data: contTable[[var - 3]]
## X-squared = 5.2911e-31, df = 1, p-value = 1
##
## [1] "Adjusted p-value: 0.999999999999999"# ### Adjust p-values for multiple comparison
# df_TAH_data_impute_mean_factor <- df_TAH_data_impute_mean[, indFactor]
# factorPvalueResults <- df_TAH_data_impute_mean_factor[-1] %>%
# summarise_each(funs(chisq.test(., df_TAH_data_impute_mean_factor[-1]$TAH_occurrence)$p.value),
# -one_of("TAH_occurrence"))
# factorPvalueResults_adj <- p.adjust(factorPvalueResults, method = 'fdr')
# factorPvalueResults_adj4.1.2 Special case for
Access
Account for the significant missingness in the variable
access.
print("TAH_occurrence vs Access")## [1] "TAH_occurrence vs Access"ctTable <- table(TAH_data$TAH_occurrence, TAH_data$Access,
dnn = list("TAH_occurrence", "Access"))
print(addmargins(ctTable)) ## Access
## TAH_occurrence 1 2 Missing Sum
## 0 18 29 13 60
## 1 8 7 20 35
## Sum 26 36 33 95print(prop.table(ctTable)) ## Access
## TAH_occurrence 1 2 Missing
## 0 0.18947368 0.30526316 0.13684211
## 1 0.08421053 0.07368421 0.21052632print(chisq.test(ctTable))##
## Pearson's Chi-squared test
##
## data: ctTable
## X-squared = 13.104, df = 2, p-value = 0.001427### Adjust p-values for multiple comparison
factorPvalueResults_adj <- p.adjust(chisq.test(ctTable)$p.value, method = 'fdr')
print(paste0("TAH vs. Access Adjusted p-value: ", factorPvalueResults_adj))## [1] "TAH vs. Access Adjusted p-value: 0.00142727460940099"4.1.3 Quantitative numerical variables (Wilcoxon test)
############################################################
# If need to adjust numeric and factor p-values separately #
############################################################
df_TAH_data_impute_mean_numeric <- df_TAH_data_impute_mean[, !indFactor]
tests_list <- lapply(colnames(df_TAH_data_impute_mean_numeric),
function(x) wilcox.test(
as.formula(paste0(x, "~ TAH_occurrence")),
data = cbind(df_TAH_data_impute_mean_numeric,
TAH_occurrence=df_TAH_data_impute_mean$TAH_occurrence)))
numericPvalueResults <- sapply(tests_list, `[[`, "p.value")
numericPvalueResults_adj <- p.adjust(numericPvalueResults, method = 'fdr')
result <- cbind(QuantVariables=colnames(df_TAH_data_impute_mean[!indFactor]),
numeric_TAH_Pvalues=as.numeric(numericPvalueResults),
numeric_TAH_Pvalues_adj=as.numeric(numericPvalueResults_adj))
result## QuantVariables numeric_TAH_Pvalues
## [1,] "Age_days" "0.204330730300978"
## [2,] "Weight" "0.451902378599295"
## [3,] "Pre_transfusion_K" "0.0325746993155222"
## [4,] "Cr_at_transfusion" "0.000736976834655347"
## [5,] "Age_of_RBC_unit" "0.0200204860912063"
## [6,] "tV_12_weight_ml_Kg" "0.00286239354727626"
## [7,] "tV_12_eTBV" "0.00680924098372903"
## [8,] "Infusion_rate_hour_Pt_weight" "0.184471972128806"
## [9,] "Infusion_rate_hour_eTBV" "1"
## [10,] "Time_from_irradition_to_transfusion_min" "0.875201386003872"
## numeric_TAH_Pvalues_adj
## [1,] "0.291901043287112"
## [2,] "0.564877973249119"
## [3,] "0.0651493986310444"
## [4,] "0.00736976834655347"
## [5,] "0.0500512152280158"
## [6,] "0.0143119677363813"
## [7,] "0.0226974699457634"
## [8,] "0.291901043287112"
## [9,] "1"
## [10,] "0.972445984448747"4.2 Other results
library(tidyverse)
for (i in 1:length(colnames(df_TAH_data_impute_mean))) { # Separate Factor & Numeric Var's
print (colnames(df_TAH_data_impute_mean)[i])
ifelse (colnames(df_TAH_data_impute_mean)[i] %in%
c("Age_days",
"Weight",
"Pre_transfusion_K",
"Cr_at_transfusion",
"Age_of_RBC_unit",
"tV_12_weight_ml_Kg",
"tV_12_eTBV",
"Infusion_rate_hour_Pt_weight",
"Infusion_rate_hour_eTBV",
"Time_from_irradition_to_transfusion_min"),
df_TAH_data_impute_mean[,i] <- as.numeric(df_TAH_data_impute_mean[,i]),
df_TAH_data_impute_mean[,i] <- as.factor(df_TAH_data_impute_mean[,i]))
}## [1] "Group"
## [1] "TAH_occurrence"
## [1] "Mortality1Day"
## [1] "Sex"
## [1] "Race"
## [1] "Access"
## [1] "Age_days"
## [1] "Weight"
## [1] "Pre_transfusion_K"
## [1] "Cr_at_transfusion"
## [1] "Cardiac_dysf"
## [1] "Respiratory_dysf"
## [1] "Kidney_dysf"
## [1] "Liver_dysf"
## [1] "Intracranial_diso"
## [1] "Orthopedics"
## [1] "Anemia"
## [1] "Hemoglobinopathy"
## [1] "Sepsis"
## [1] "ECMO"
## [1] "CPB"
## [1] "Prematurity"
## [1] "Dialysis"
## [1] "RBC_Unit_type"
## [1] "Age_of_RBC_unit"
## [1] "Leukoreduction"
## [1] "Irradiation"
## [1] "Washed"
## [1] "Plasma_reduction"
## [1] "tV_12_weight_ml_Kg"
## [1] "tV_12_eTBV"
## [1] "Infusion_rate_hour_Pt_weight"
## [1] "Infusion_rate_hour_eTBV"
## [1] "Time_from_irradition_to_transfusion_min"indFactor <- sapply(df_TAH_data_impute_mean, is.factor)
table(indFactor)## indFactor
## FALSE TRUE
## 10 24##################################################################
# If don't need to adjust numeric and factor p-values separately #
##################################################################
Pvalue = function(x) {
if (is.factor(x)) {
##print(paste0("Variable: ", x))
return(chisq.test(x, df_TAH_data_impute_mean$TAH_occurrence)$p.value)
} else {
return(wilcox.test(x ~ df_TAH_data_impute_mean$TAH_occurrence)$p.value)
}
}
PvalueResults = sapply(df_TAH_data_impute_mean[, 3:dim(df_TAH_data_impute_mean)[2]], Pvalue)
PvalueResults_adj = p.adjust(PvalueResults, method="fdr")
result = cbind(p.value.fdr.adj = unname(PvalueResults_adj))
rownames(result) = names(PvalueResults_adj)
result## p.value.fdr.adj
## Mortality1Day 0.013164988
## Sex 0.377735564
## Race 0.832034724
## Access 0.882551078
## Age_days 0.475962126
## Weight 0.738624482
## Pre_transfusion_K 0.115821153
## Cr_at_transfusion 0.011791629
## Cardiac_dysf 1.000000000
## Respiratory_dysf 0.582025322
## Kidney_dysf 0.005694053
## Liver_dysf 0.099005765
## Intracranial_diso 0.632472231
## Orthopedics 1.000000000
## Anemia 0.109702056
## Hemoglobinopathy 0.475962126
## Sepsis 0.500005344
## ECMO 0.475962126
## CPB 1.000000000
## Prematurity 0.996572507
## Dialysis 1.000000000
## RBC_Unit_type 0.650878921
## Age_of_RBC_unit 0.099005765
## Leukoreduction 0.738624482
## Irradiation 1.000000000
## Washed 0.996572507
## Plasma_reduction 1.000000000
## tV_12_weight_ml_Kg 0.022899148
## tV_12_eTBV 0.043579142
## Infusion_rate_hour_Pt_weight 0.475962126
## Infusion_rate_hour_eTBV 1.000000000
## Time_from_irradition_to_transfusion_min 1.000000000############################################################
# If need to adjust numeric and factor p-values separately #
############################################################
df_TAH_data_impute_mean_factor <- df_TAH_data_impute_mean[, indFactor]
factorPvalueResults <- df_TAH_data_impute_mean_factor[-1] %>%
summarise_each(funs(chisq.test(., df_TAH_data_impute_mean_factor[-1]$TAH_occurrence)$p.value),
-one_of("TAH_occurrence"))
factorPvalueResults_adj <- p.adjust(factorPvalueResults, method = 'fdr')
df_TAH_data_impute_mean_numeric <- df_TAH_data_impute_mean[, !indFactor]
tests_list <- lapply(colnames(df_TAH_data_impute_mean_numeric),
function(x) wilcox.test(
as.formula(paste0(x, "~ TAH_occurrence")),
data = cbind(df_TAH_data_impute_mean_numeric,
TAH_occurrence=df_TAH_data_impute_mean$TAH_occurrence)))
numericPvalueResults <- sapply(tests_list, `[[`, "p.value")
numericPvalueResults_adj <- p.adjust(numericPvalueResults, method = 'fdr')
PvalueResults_adj = c(unname(factorPvalueResults_adj), numericPvalueResults_adj)
result <- cbind(p.value.fdr.adj = PvalueResults_adj)
rownames(result) <- c(names(factorPvalueResults_adj),
colnames(df_TAH_data_impute_mean[!indFactor]))
result## p.value.fdr.adj
## Mortality1Day 0.013576394
## Sex 0.519386400
## Race 0.924038564
## Access 0.953470361
## Cardiac_dysf 1.000000000
## Respiratory_dysf 0.711364282
## Kidney_dysf 0.003914661
## Liver_dysf 0.150840326
## Intracranial_diso 0.732238786
## Orthopedics 1.000000000
## Anemia 0.150840326
## Hemoglobinopathy 0.644538139
## Sepsis 0.644538139
## ECMO 0.644538139
## CPB 1.000000000
## Prematurity 1.000000000
## Dialysis 1.000000000
## RBC_Unit_type 0.732238786
## Leukoreduction 0.846340553
## Irradiation 1.000000000
## Washed 1.000000000
## Plasma_reduction 1.000000000
## Age_days 0.291901043
## Weight 0.564877973
## Pre_transfusion_K 0.065149399
## Cr_at_transfusion 0.007369768
## Age_of_RBC_unit 0.050051215
## tV_12_weight_ml_Kg 0.014311968
## tV_12_eTBV 0.022697470
## Infusion_rate_hour_Pt_weight 0.291901043
## Infusion_rate_hour_eTBV 1.000000000
## Time_from_irradition_to_transfusion_min 0.972445984# Paired TAH t-tests comparing patients vs. controls
# Which test to use: http://www.bristol.ac.uk/medical-school/media/rms/red/which_test.html
library(tidyverse)
######### Factor variables ===============
# df_TAH_data_impute_mean_wo_TAH <- subset(df_TAH_data_impute_mean, select = -c(TAH_occurrence))
#
# for (i in 1:length(colnames(df_TAH_data_impute_mean_wo_TAH))){
# print (colnames(df_TAH_data_impute_mean_wo_TAH)[i])
# ifelse (colnames(df_TAH_data_impute_mean_wo_TAH)[i] %in% c("Age_days", "Weight", "Pre_transfusion_K", "Cr_at_transfusion",
# "Age_of_RBC_unit", "tV_12_weight_ml_Kg", "tV_12_eTBV",
# "Infusion_rate_hour_Pt_weight", "Infusion_rate_hour_eTBV",
# "Time_from_irradition_to_transfusion_min"),
# df_TAH_data_impute_mean_wo_TAH[,i] <- as.numeric(df_TAH_data_impute_mean_wo_TAH[,i]),
# df_TAH_data_impute_mean_wo_TAH[,i] <- as.factor(df_TAH_data_impute_mean_wo_TAH[,i]))
# }
# indFactor <- sapply(df_TAH_data_impute_mean_wo_TAH, is.factor)
# table(indFactor)
for (i in 1:length(colnames(df_TAH_data_impute_mean))){
print (colnames(df_TAH_data_impute_mean)[i])
ifelse (colnames(df_TAH_data_impute_mean)[i] %in% c("Age_days", "Weight", "Pre_transfusion_K", "Cr_at_transfusion",
"Age_of_RBC_unit", "tV_12_weight_ml_Kg", "tV_12_eTBV",
"Infusion_rate_hour_Pt_weight", "Infusion_rate_hour_eTBV",
"Time_from_irradition_to_transfusion_min"),
df_TAH_data_impute_mean[,i] <- as.numeric(df_TAH_data_impute_mean[,i]),
df_TAH_data_impute_mean[,i] <- as.factor(df_TAH_data_impute_mean[,i]))
}## [1] "Group"
## [1] "TAH_occurrence"
## [1] "Mortality1Day"
## [1] "Sex"
## [1] "Race"
## [1] "Access"
## [1] "Age_days"
## [1] "Weight"
## [1] "Pre_transfusion_K"
## [1] "Cr_at_transfusion"
## [1] "Cardiac_dysf"
## [1] "Respiratory_dysf"
## [1] "Kidney_dysf"
## [1] "Liver_dysf"
## [1] "Intracranial_diso"
## [1] "Orthopedics"
## [1] "Anemia"
## [1] "Hemoglobinopathy"
## [1] "Sepsis"
## [1] "ECMO"
## [1] "CPB"
## [1] "Prematurity"
## [1] "Dialysis"
## [1] "RBC_Unit_type"
## [1] "Age_of_RBC_unit"
## [1] "Leukoreduction"
## [1] "Irradiation"
## [1] "Washed"
## [1] "Plasma_reduction"
## [1] "tV_12_weight_ml_Kg"
## [1] "tV_12_eTBV"
## [1] "Infusion_rate_hour_Pt_weight"
## [1] "Infusion_rate_hour_eTBV"
## [1] "Time_from_irradition_to_transfusion_min"indFactor <- sapply(df_TAH_data_impute_mean, is.factor)
table(indFactor)## indFactor
## FALSE TRUE
## 10 24# dat[indFactor] <- lapply(dat[indx], as.numeric.factor)
# str(df_TAH_data_impute_mean_wo_TAH)
df_TAH_data_impute_mean_factor <- df_TAH_data_impute_mean[, indFactor]
factorPvalueResults <- df_TAH_data_impute_mean_factor[-1] %>%
summarise_each(funs(chisq.test(., df_TAH_data_impute_mean_factor[-1]$TAH_occurrence)$p.value),
-one_of("TAH_occurrence"))
factorPvalueResults_adj <- p.adjust(factorPvalueResults, method = 'fdr')
# For Kidney dysfunction
p.adjust(chisq.test(cbind(x=c(17,18), y=c(7,53)))$p.value, method = 'fdr', n=18) # Pearson's Chi^2 test / Yates' continuity correction## [1] 0.003202905# data: cbind(x = c(17, 18), y = c(7, 53))
# X-squared = 14.051, df = 1, p-value = 0.0001779, adj.p.value = 0.003202905
######### Numerical Variables ===============
# df_TAH_data_impute_mean_wo_TAH <- subset(df_TAH_data_impute_mean_wo_TAH, select = -c(TAH_occurrence, indFactor))
#
# tests_list <- lapply(colnames(df_TAH_data_impute_mean_wo_TAH[-indFactor]),
# # function(x) t.test(as.formula(paste0(x, "~ TAH_occurrence")), data = df_TAH_data_impute_mean)) # Parametric t-test
# # Wilcoxon rank sum test with continuity correction
# function(x) wilcox.test(as.formula(paste0(x, "~ TAH_occurrence")), data = df_TAH_data_impute_mean)) # Non-parametric Wilcox-test
#
# result <- do.call(rbind, lapply(tests_list, `[[`, "estimate"))
# pval <- sapply(tests_list, `[[`, "p.value")
# pval_adj <- p.adjust(pval, method = 'fdr')
# est <- sapply(tests_list, `[[`, "statistic")
# result <- cbind(result, Wilcoxon.p.value.fdr.adj = pval_adj)
# rownames(result) <- colnames(df_TAH_data_impute_mean_wo_TAH[-indFactor])
# result
# colnames(df_TAH_data_impute_clean)
# colnames(TAH_data)
# df_TAH_data_impute_mean$Mortality1Day <- ifelse(df_TAH_data_impute_mean$Mortality1Day<0.5, "0", "1")
# rownames(df_TAH_data_impute_mean) <- df_TAH_data_impute_mean[ , 1]
# df_TAH_data_impute_mean <- df_TAH_data_impute_mean[ , -1]
# EDA
library(tidyverse)
df_TAH_data_Mortality0 <- filter(df_TAH_data_impute_mean, Mortality1Day==1)
dim(df_TAH_data_Mortality0) # [1] 87 41## [1] 87 34df_TAH_data_Mortality1 <- filter(df_TAH_data_impute_mean, Mortality1Day==2)
dim(df_TAH_data_Mortality1) # 7 41## [1] 7 34# note that some missing Mortality1Day values are estimated to be between 0 and 1!
#### Cohort separation
df_TAH_data_TAH_Controls<- filter(df_TAH_data_impute_mean, TAH_occurrence==1)
dim(df_TAH_data_TAH_Controls) # [1] 60 41## [1] 60 34df_TAH_data_TAH_Patients <- filter(df_TAH_data_impute_mean, TAH_occurrence==2)
dim(df_TAH_data_TAH_Patients) # 35 41## [1] 35 34select_Group_Sex_Mortality_TAH <- select(df_TAH_data_impute_mean, Group, Sex, Mortality1Day)
library(ggplot2)
ggplot(data = df_TAH_data_impute_mean, mapping = aes(x = as.numeric(Mortality1Day), colour = as.factor(Group))) +
geom_freqpoly(binwidth = 0.1)
ggplot(data = df_TAH_data_impute_mean, mapping = aes(x = as.factor(Group), y = tV_12_eTBV)) +
geom_boxplot()
library(plotly)
plot_ly(df_TAH_data_impute_mean) %>%
add_trace(type = 'splom', dimensions = list( list(label='Infusion_rate_hour_eTBV', values=~Infusion_rate_hour_eTBV),
list(label='Mortality1Day', values=~Mortality1Day),
list(label='Sex', values=~Sex),
list(label='Age_days', values=~Age_days),
list(label='Time_from_iradition_to_transfusion_min',
values=~Irradiation),
list(label='tV_12_eTBV', values=~tV_12_eTBV)),
text=~Group,
marker = list(color = as.integer(df_TAH_data_impute_mean$Group), size=7,
line=list(width=1, color = 'rgb(230,230,230)')
)
) # %>%
# layout(
# title= 'Random Data Pairs Plot', hovermode='closest', dragmode= 'select',
# plot_bgcolor='rgba(240,240,240, 0.95)',
# xaxis=list(domain=NULL, showline=F, zeroline=F, gridcolor='#ffff', ticklen=4),
# yaxis=list(domain=NULL, showline=F, zeroline=F, gridcolor='#ffff', ticklen=4),
# xaxis2=axis, xaxis3=axis, xaxis4=axis,yaxis2=axis, yaxis3=axis, yaxis4=axis)
# manually re-ran the Wilcoxon rank-sum test comparing Infusion_rate_hour_Pt_weight across TAH and CN:
# tah_Infusion_rate_hour_Pt_weight <-
# df_TAH_data_impute_mean[which(df_TAH_data_impute_mean$TAH_occurrence == 2), "Infusion_rate_hour_Pt_weight"]
# cn_Infusion_rate_hour_Pt_weight <-
# df_TAH_data_impute_mean[which(df_TAH_data_impute_mean$TAH_occurrence == 1), "Infusion_rate_hour_Pt_weight"]
#
# summary(tah_Infusion_rate_hour_Pt_weight)
# summary(cn_Infusion_rate_hour_Pt_weight)
# # summary(tah_Infusion_rate_hour_Pt_weight)
# # Min. 1st Qu. Median Mean 3rd Qu. Max.
# # 1.810 3.765 6.890 25.630 18.670 193.400
# # summary(cn_Infusion_rate_hour_Pt_weight)
# # Min. 1st Qu. Median Mean 3rd Qu. Max.
# # 0.040 3.865 5.270 8.719 7.862 120.480
#
# wilcox.test(tah_Infusion_rate_hour_Pt_weight, cn_Infusion_rate_hour_Pt_weight)
# =============================
# Use Chi-square test for categorical (factor variables)
# str(TAH_data): str(df_TAH_data_impute_mean_wo_TAH)
# dim(df_TAH_data_impute_mean_wo_TAH); dim(TAH_data)
# Sex : Factor w/ 2 levels "1","2": 1 2 2 1 2 1 1 2 2 1 ...
# Race : Factor w/ 6 levels "1","2","3","4",..: 5 5 5 5 3 1 1 1 2 1 ...
# Access : Factor w/ 3 levels "1","2","Missing": 2 2 2 2 2 1 2 2 2 2 ...
tah1 <- subset(df_TAH_data_impute_mean, TAH_occurrence == 1, select = c("Sex", "Race", "Access"))
tah2 <- subset(df_TAH_data_impute_mean, TAH_occurrence == 2, select = c("Sex", "Race", "Access"))
prop.test(x = table(tah1$Sex), n=table(tah2$Sex), correct=TRUE)##
## 1-sample proportions test with continuity correction
##
## data: table(tah1$Sex), null probability 0.5
## X-squared = 2.8167, df = 1, p-value = 0.09329
## alternative hypothesis: true p is not equal to 0.5
## 95 percent confidence interval:
## 0.2635284 0.5180989
## sample estimates:
## p
## 0.3833333prop.test(x = as.vector(table(tah2$Race)[-4]), n=as.vector(table(tah1$Race)[-4]), correct=TRUE)##
## 4-sample test for equality of proportions without continuity correction
##
## data: as.vector(table(tah2$Race)[-4]) out of as.vector(table(tah1$Race)[-4])
## X-squared = 4.0764, df = 3, p-value = 0.2533
## alternative hypothesis: two.sided
## sample estimates:
## prop 1 prop 2 prop 3 prop 4
## 0.6500000 0.6470588 0.2000000 0.5000000prop.test(x = table(tah1$Access), n=table(tah2$Access), correct=TRUE)##
## 1-sample proportions test with continuity correction
##
## data: table(tah1$Access), null probability 0.5
## X-squared = 8.8167, df = 1, p-value = 0.002985
## alternative hypothesis: true p is not equal to 0.5
## 95 percent confidence interval:
## 0.1920041 0.4337324
## sample estimates:
## p
## 0.34.3 Interpretation of a simple logistic model
In general, we can have multiple predictor variables in a logistic regression model (to quantify the probability, \(p\), of a participant to be control or patient).
\[logit(p)= \log\left(\frac{p}{1-p}\right ) = \beta_o + \beta_1*x_1 + \cdots + \beta_k*x_k.\] Assume, we already calculated/estimated the coefficients (effect-sizes, \(\beta_i,\ i=1,2,\cdots, k\)). For simplicity, assume all other features, but \(x_1\), are fixed, so we have a single predictor (\(x_1\)).
\[logit(p) = \beta_o + \beta_1*x_1.\] ————-|————|———–|——-|———|———–|——— Outcome | Coef. | Std. Err.| z| P>|z|| [95% Conf. Interval] ————-|————|———–|——-|———|———–|——— \(x_1\) | .1563404 | .0256095 | 6.10 | 0.000 | .1061467 | .206534 intercept | -9.793942 | 1.481745 | -6.61 | 0.000 | -12.69811 | -6.889775 —————————————————————————— In this case, \(p = prob(TAH=1)\) the estimated coefficient for the intercept is the log odds of a patient with a \(x_1\)=0 being a control (\(TAH=1\)). In other words, the odds of being a control patient when \(x_1\)=0 is \(e^{-9.793942} = .00005579\). These odds are very low. So the intercept in this model corresponds to the log odds of when \(x_1\) is at a hypothetical value of zero – this is of course, not particularly interesting (in a statistical sense).
Let’s now interpret the value of the \(x_1\) coefficient \(\beta_1\). The coefficient and intercept estimates give us the following equation:
\[\log\left(\frac{p}{1-p}\right) = logit(p) = –9.793942 + 0.1563404* x_1.\] Let’s fix \(x_1\) at some value, e.g., \(x_1=54\). Then the conditional logit of a control class label (\(TAH=1\)) when \(x_1=54\) is
\[\log\left(\frac{p}{1-p}\right)( x_1 =54) = – 9.793942 + .1563404 *54.\]
Let’s examine the effect of a one-unit increase in \(x_1\). When \(x_1\) is held at 55, the conditional logit of the Control class ((\(TAH=1\))) is
\[\log\left(\frac{p}{1-p}\right)(x_1 =55) = – 9.793942 + .1563404*55.\]
Taking the difference of the two equations corresponding to \(x_1 =55\) and \(x_1 =55\), we get:
\[\underbrace{\underbrace{\log\left(\frac{p}{1-p}\right)(x_1=55)}\ –\ \underbrace{\log\left(\frac{p}{1-p}\right)(x_1= 54)}}_{LHS} = \underbrace{ 0.1563404\equiv \beta_1}_{RHS}.\]
Note that the coefficient for \(x_1\), \(\beta_1\), is just the difference in these two log odds. Thus, for a one-unit increase of \(x_1\), the expected change in log odds is \(0.1563404\).
To interpret the \(x_1\) coefficient (and determine the effect of \(x_1\) on diagnosis, TAH vs. control), we translate this change in log odds to a change in odds. Recall that the logarithm, \(y=\log(x)\), converts multiplication and division to addition and subtraction., and conversely, its inverse, the exponential function, \(x=e^y\), converts addition and subtraction back to multiplication and division.
Exponentiating both sides (\(LHS\) and \(RHS\)) of the last equation yields:
\[e^{\log\left(\frac{p}{1-p}\right)(x_1=55) – \log\left(\frac{p}{1-p}\right)(x_1 =54)}= \frac{e^{\log\left(\frac{p}{1-p}\right)(x_1=55)}}{e^{\log\left(\frac{p}{1-p}\right)(x_1=54)}}= \frac{odds(x_1=55)}{odds(x_1=54)} = e^{0.1563404} = 1.1692241\sim 1.0+0.17.\]
To summarize, a single-unit increase in \(x_1\) is expected to yield \(\sim 17\%\) increase in the odds of a participant being a control (\(TAH=1\)). This \(17\%\) increase is independent of the actual value of the variable \(x_1\).
5 One-Day Mortality Analysis
# Paired 1-Day Mortality-Group t-tests comparing patients vs. controls
df_TAH_data_impute_mean_2Way_Mortality <- df_TAH_data_impute_mean[!(df_TAH_data_impute_mean$Mortality1Day==3), ] # remove case with Mortality1Day==3
df_TAH_data_impute_mean_wo_Mortality <- subset(df_TAH_data_impute_mean_2Way_Mortality, select = -c(Mortality1Day))
####### Fix the numeric & categorical variable group-diff tests (Wilcoxon vs. Chi^2) #####
# tests_list <- lapply(colnames(df_TAH_data_impute_mean_wo_Mortality),
# # function(x) t.test(as.formula(paste0(x, "~ Mortality1Day")), data = df_TAH_data_impute_mean_2Way_Mortality)) # Parametric t-test
# # Wilcoxon rank sum test with continuity correction
# function(x) wilcox.test(as.formula(paste0(x, "~ Mortality1Day")), data = df_TAH_data_impute_mean_2Way_Mortality)) # Non-parametric Wilcox-test
#
# result <- do.call(rbind, lapply(tests_list, `[[`, "estimate"))
# pval <- sapply(tests_list, `[[`, "p.value")
# pval_adj <- p.adjust(pval, method = 'fdr')
# result <- cbind(result, Wilcoxon.p.value.fdr.adj = pval_adj, signif = ifelse(pval_adj<0.05, 1, 0))
# rownames(result) <- colnames(df_TAH_data_impute_mean_wo_Mortality)
# print("P-values Mortality1Day ~ . ")
# result
# colnames(df_TAH_data_impute_clean)
# colnames(TAH_data)
# df_TAH_data_impute_mean$Mortality1Day <- ifelse(df_TAH_data_impute_mean$Mortality1Day<0.5, "0", "1")
# rownames(df_TAH_data_impute_mean) <- df_TAH_data_impute_mean[ , 1]
# df_TAH_data_impute_mean <- df_TAH_data_impute_mean[ , -1]
# EDA
# library(tidyverse)
# df_TAH_data_Mortality0 <- filter(df_TAH_data_impute_mean, Mortality1Day==1)
# dim(df_TAH_data_Mortality0) # [1] 87 41
# df_TAH_data_Mortality1 <- filter(df_TAH_data_impute_mean, Mortality1Day==2)
# dim(df_TAH_data_Mortality1) # 7 41
# # note that some missing Mortality1Day values are estimated to be between 0 and 1!
tah1 <- subset(df_TAH_data_impute_mean, Mortality1Day == 1, select = c("Sex", "Race", "Access"))
tah2 <- subset(df_TAH_data_impute_mean, Mortality1Day != 1, select = c("Sex", "Race", "Access"))
prop.test(x = table(tah1$Sex), n=table(tah2$Sex), correct=TRUE)##
## 1-sample proportions test with continuity correction
##
## data: table(tah1$Sex), null probability 0.5
## X-squared = 1.1494, df = 1, p-value = 0.2837
## alternative hypothesis: true p is not equal to 0.5
## 95 percent confidence interval:
## 0.3320722 0.5471261
## sample estimates:
## p
## 0.4367816prop.test(x = as.vector(table(tah2$Race)[-4]), n=as.vector(table(tah1$Race)[-4]), correct=TRUE)##
## 4-sample test for equality of proportions without continuity correction
##
## data: as.vector(table(tah2$Race)[-4]) out of as.vector(table(tah1$Race)[-4])
## X-squared = 2.632, df = 3, p-value = 0.4519
## alternative hypothesis: two.sided
## sample estimates:
## prop 1 prop 2 prop 3 prop 4
## 0.03125 0.12000 0.20000 0.12500prop.test(x = table(tah1$Access), n=table(tah2$Access), correct=TRUE)##
## 1-sample proportions test with continuity correction
##
## data: table(tah1$Access), null probability 0.5
## X-squared = 14.897, df = 1, p-value = 0.0001136
## alternative hypothesis: true p is not equal to 0.5
## 95 percent confidence interval:
## 0.1978913 0.3958646
## sample estimates:
## p
## 0.28735635.1 Table 5
for (i in 1:length(colnames(df_TAH_data_impute_mean))){
print (colnames(df_TAH_data_impute_mean)[i])
ifelse (colnames(df_TAH_data_impute_mean)[i] %in%
c("Age_days",
"Weight",
"Pre_transfusion_K",
"Cr_at_transfusion",
"Age_of_RBC_unit",
"tV_12_weight_ml_Kg",
"tV_12_eTBV",
"Infusion_rate_hour_Pt_weight",
"Infusion_rate_hour_eTBV",
"Time_from_irradition_to_transfusion_min"),
df_TAH_data_impute_mean[,i] <- as.numeric(df_TAH_data_impute_mean[,i]),
df_TAH_data_impute_mean[,i] <- as.factor(df_TAH_data_impute_mean[,i]))
}## [1] "Group"
## [1] "TAH_occurrence"
## [1] "Mortality1Day"
## [1] "Sex"
## [1] "Race"
## [1] "Access"
## [1] "Age_days"
## [1] "Weight"
## [1] "Pre_transfusion_K"
## [1] "Cr_at_transfusion"
## [1] "Cardiac_dysf"
## [1] "Respiratory_dysf"
## [1] "Kidney_dysf"
## [1] "Liver_dysf"
## [1] "Intracranial_diso"
## [1] "Orthopedics"
## [1] "Anemia"
## [1] "Hemoglobinopathy"
## [1] "Sepsis"
## [1] "ECMO"
## [1] "CPB"
## [1] "Prematurity"
## [1] "Dialysis"
## [1] "RBC_Unit_type"
## [1] "Age_of_RBC_unit"
## [1] "Leukoreduction"
## [1] "Irradiation"
## [1] "Washed"
## [1] "Plasma_reduction"
## [1] "tV_12_weight_ml_Kg"
## [1] "tV_12_eTBV"
## [1] "Infusion_rate_hour_Pt_weight"
## [1] "Infusion_rate_hour_eTBV"
## [1] "Time_from_irradition_to_transfusion_min"indFactor <- sapply(df_TAH_data_impute_mean, is.factor)
table(indFactor)## indFactor
## FALSE TRUE
## 10 24# Contingency Table - Mortality1Day vs. Kidney-dysfunction, e.g.,
# table(df_TAH_data_impute_mean$Mortality1Day, df_TAH_data_impute_mean$Kidney_dysf)
contTable <- list()
for (var in c(2:length(factorVarNames))) {
print(paste0("Mortality1Day vs ", factorVarNames[var]))
contTable[[var-1]] <- table(df_TAH_data_impute_mean$Mortality1Day,
df_TAH_data_impute_mean[[factorVarNames[var]]],
dnn = list("Mortality1Day", factorVarNames[var]))
}## [1] "Mortality1Day vs Group"
## [1] "Mortality1Day vs TAH_occurrence"
## [1] "Mortality1Day vs Mortality1Day"
## [1] "Mortality1Day vs Sex"
## [1] "Mortality1Day vs Race"
## [1] "Mortality1Day vs Access"
## [1] "Mortality1Day vs Dialysis"
## [1] "Mortality1Day vs Cardiac_dysf"
## [1] "Mortality1Day vs Respiratory_dysf"
## [1] "Mortality1Day vs Kidney_dysf"
## [1] "Mortality1Day vs Liver_dysf"
## [1] "Mortality1Day vs Intracranial_diso"
## [1] "Mortality1Day vs Orthopedics"
## [1] "Mortality1Day vs Anemia"
## [1] "Mortality1Day vs Hemoglobinopathy"
## [1] "Mortality1Day vs Sepsis"
## [1] "Mortality1Day vs ECMO"
## [1] "Mortality1Day vs CPB"
## [1] "Mortality1Day vs Prematurity"
## [1] "Mortality1Day vs RBC_Unit_type"
## [1] "Mortality1Day vs Leukoreduction"
## [1] "Mortality1Day vs Irradiation"
## [1] "Mortality1Day vs Washed"
## [1] "Mortality1Day vs Plasma_reduction"# addmargins(contTable[[10]]) # "Kidney_dysf"
# prop.table(contTable[[10]]) # "Kidney_dysf"
# chisq.test(contTable[[10]]) # "Kidney_dysf"
# fisher.test(contTable[[10]])
for (var in c(2:length(factorVarNames))) {
print(addmargins(contTable[[var-1]]))
print(prop.table(contTable[[var-1]]))
print(chisq.test(contTable[[var-1]]))
### Adjust p-values for multiple comparison
factorPvalueResults_adj <- p.adjust(chisq.test(contTable[[var-1]])$p.value,
method = 'fdr')
print(paste0("Adjusted p-value: ", factorPvalueResults_adj))
}## Group
## Mortality1Day 1 2 3 4 5 Sum
## 1 19 16 21 13 18 87
## 2 2 1 3 0 1 7
## 3 1 0 0 0 0 1
## Sum 22 17 24 13 19 95
## Group
## Mortality1Day 1 2 3 4 5
## 1 0.20000000 0.16842105 0.22105263 0.13684211 0.18947368
## 2 0.02105263 0.01052632 0.03157895 0.00000000 0.01052632
## 3 0.01052632 0.00000000 0.00000000 0.00000000 0.00000000
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 1]]
## X-squared = 5.6299, df = 8, p-value = 0.6886
##
## [1] "Adjusted p-value: 0.688613141982977"
## TAH_occurrence
## Mortality1Day 1 2 Sum
## 1 59 28 87
## 2 0 7 7
## 3 1 0 1
## Sum 60 35 95
## TAH_occurrence
## Mortality1Day 1 2
## 1 0.62105263 0.29473684
## 2 0.00000000 0.07368421
## 3 0.01052632 0.00000000
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 1]]
## X-squared = 13.395, df = 2, p-value = 0.001234
##
## [1] "Adjusted p-value: 0.00123421762995026"
## Mortality1Day
## Mortality1Day 1 2 3 Sum
## 1 87 0 0 87
## 2 0 7 0 7
## 3 0 0 1 1
## Sum 87 7 1 95
## Mortality1Day
## Mortality1Day 1 2 3
## 1 0.91578947 0.00000000 0.00000000
## 2 0.00000000 0.07368421 0.00000000
## 3 0.00000000 0.00000000 0.01052632
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 1]]
## X-squared = 190, df = 4, p-value < 2.2e-16
##
## [1] "Adjusted p-value: 5.30023898594739e-40"
## Sex
## Mortality1Day 1 2 Sum
## 1 38 49 87
## 2 5 2 7
## 3 0 1 1
## Sum 43 52 95
## Sex
## Mortality1Day 1 2
## 1 0.40000000 0.51578947
## 2 0.05263158 0.02105263
## 3 0.00000000 0.01052632
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 1]]
## X-squared = 2.8495, df = 2, p-value = 0.2406
##
## [1] "Adjusted p-value: 0.24057323869112"
## Race
## Mortality1Day 1 2 3 4 5 Sum
## 1 32 25 5 1 24 87
## 2 1 3 0 0 3 7
## 3 0 0 1 0 0 1
## Sum 33 28 6 1 27 95
## Race
## Mortality1Day 1 2 3 4 5
## 1 0.33684211 0.26315789 0.05263158 0.01052632 0.25263158
## 2 0.01052632 0.03157895 0.00000000 0.00000000 0.03157895
## 3 0.00000000 0.00000000 0.01052632 0.00000000 0.00000000
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 1]]
## X-squared = 17.325, df = 8, p-value = 0.0269
##
## [1] "Adjusted p-value: 0.0268967038251054"
## Access
## Mortality1Day 1 2 Sum
## 1 25 62 87
## 2 1 6 7
## 3 0 1 1
## Sum 26 69 95
## Access
## Mortality1Day 1 2
## 1 0.26315789 0.65263158
## 2 0.01052632 0.06315789
## 3 0.00000000 0.01052632
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 1]]
## X-squared = 1.0613, df = 2, p-value = 0.5882
##
## [1] "Adjusted p-value: 0.58820879061329"
## Dialysis
## Mortality1Day 1 2 Sum
## 1 82 5 87
## 2 6 1 7
## 3 1 0 1
## Sum 89 6 95
## Dialysis
## Mortality1Day 1 2
## 1 0.86315789 0.05263158
## 2 0.06315789 0.01052632
## 3 0.01052632 0.00000000
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 1]]
## X-squared = 0.86644, df = 2, p-value = 0.6484
##
## [1] "Adjusted p-value: 0.648419306274419"
## Cardiac_dysf
## Mortality1Day 1 2 Sum
## 1 68 19 87
## 2 6 1 7
## 3 0 1 1
## Sum 74 21 95
## Cardiac_dysf
## Mortality1Day 1 2
## 1 0.71578947 0.20000000
## 2 0.06315789 0.01052632
## 3 0.00000000 0.01052632
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 1]]
## X-squared = 3.776, df = 2, p-value = 0.1514
##
## [1] "Adjusted p-value: 0.151376936861258"
## Respiratory_dysf
## Mortality1Day 1 2 Sum
## 1 57 30 87
## 2 3 4 7
## 3 0 1 1
## Sum 60 35 95
## Respiratory_dysf
## Mortality1Day 1 2
## 1 0.60000000 0.31578947
## 2 0.03157895 0.04210526
## 3 0.00000000 0.01052632
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 1]]
## X-squared = 3.1622, df = 2, p-value = 0.2057
##
## [1] "Adjusted p-value: 0.205747651107535"
## Kidney_dysf
## Mortality1Day 1 2 Sum
## 1 67 20 87
## 2 3 4 7
## 3 1 0 1
## Sum 71 24 95
## Kidney_dysf
## Mortality1Day 1 2
## 1 0.70526316 0.21052632
## 2 0.03157895 0.04210526
## 3 0.01052632 0.00000000
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 1]]
## X-squared = 4.3444, df = 2, p-value = 0.1139
##
## [1] "Adjusted p-value: 0.113927815387277"
## Liver_dysf
## Mortality1Day 1 2 Sum
## 1 77 10 87
## 2 4 3 7
## 3 1 0 1
## Sum 82 13 95
## Liver_dysf
## Mortality1Day 1 2
## 1 0.81052632 0.10526316
## 2 0.04210526 0.03157895
## 3 0.01052632 0.00000000
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 1]]
## X-squared = 5.5555, df = 2, p-value = 0.06218
##
## [1] "Adjusted p-value: 0.0621790780732849"
## Intracranial_diso
## Mortality1Day 1 2 Sum
## 1 80 7 87
## 2 4 3 7
## 3 0 1 1
## Sum 84 11 95
## Intracranial_diso
## Mortality1Day 1 2
## 1 0.84210526 0.07368421
## 2 0.04210526 0.03157895
## 3 0.00000000 0.01052632
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 1]]
## X-squared = 15.386, df = 2, p-value = 0.000456
##
## [1] "Adjusted p-value: 0.000456019591037674"
## Orthopedics
## Mortality1Day 1 2 Sum
## 1 83 4 87
## 2 7 0 7
## 3 1 0 1
## Sum 91 4 95
## Orthopedics
## Mortality1Day 1 2
## 1 0.87368421 0.04210526
## 2 0.07368421 0.00000000
## 3 0.01052632 0.00000000
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 1]]
## X-squared = 0.38398, df = 2, p-value = 0.8253
##
## [1] "Adjusted p-value: 0.82531354019221"
## Anemia
## Mortality1Day 1 2 Sum
## 1 42 45 87
## 2 5 2 7
## 3 0 1 1
## Sum 47 48 95
## Anemia
## Mortality1Day 1 2
## 1 0.44210526 0.47368421
## 2 0.05263158 0.02105263
## 3 0.00000000 0.01052632
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 1]]
## X-squared = 2.3789, df = 2, p-value = 0.3044
##
## [1] "Adjusted p-value: 0.304388656774702"
## Hemoglobinopathy
## Mortality1Day 1 2 Sum
## 1 78 9 87
## 2 7 0 7
## 3 1 0 1
## Sum 86 9 95
## Hemoglobinopathy
## Mortality1Day 1 2
## 1 0.82105263 0.09473684
## 2 0.07368421 0.00000000
## 3 0.01052632 0.00000000
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 1]]
## X-squared = 0.91419, df = 2, p-value = 0.6331
##
## [1] "Adjusted p-value: 0.633118903732215"
## Sepsis
## Mortality1Day 1 2 Sum
## 1 81 6 87
## 2 5 2 7
## 3 1 0 1
## Sum 87 8 95
## Sepsis
## Mortality1Day 1 2
## 1 0.85263158 0.06315789
## 2 0.05263158 0.02105263
## 3 0.01052632 0.00000000
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 1]]
## X-squared = 4.0397, df = 2, p-value = 0.1327
##
## [1] "Adjusted p-value: 0.132675897142401"
## ECMO
## Mortality1Day 1 2 Sum
## 1 78 9 87
## 2 6 1 7
## 3 1 0 1
## Sum 85 10 95
## ECMO
## Mortality1Day 1 2
## 1 0.82105263 0.09473684
## 2 0.06315789 0.01052632
## 3 0.01052632 0.00000000
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 1]]
## X-squared = 0.22573, df = 2, p-value = 0.8933
##
## [1] "Adjusted p-value: 0.893270496829204"
## CPB
## Mortality1Day 1 2 Sum
## 1 83 4 87
## 2 7 0 7
## 3 1 0 1
## Sum 91 4 95
## CPB
## Mortality1Day 1 2
## 1 0.87368421 0.04210526
## 2 0.07368421 0.00000000
## 3 0.01052632 0.00000000
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 1]]
## X-squared = 0.38398, df = 2, p-value = 0.8253
##
## [1] "Adjusted p-value: 0.82531354019221"
## Prematurity
## Mortality1Day 1 2 Sum
## 1 67 20 87
## 2 4 3 7
## 3 0 1 1
## Sum 71 24 95
## Prematurity
## Mortality1Day 1 2
## 1 0.70526316 0.21052632
## 2 0.04210526 0.03157895
## 3 0.00000000 0.01052632
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 1]]
## X-squared = 4.3444, df = 2, p-value = 0.1139
##
## [1] "Adjusted p-value: 0.113927815387277"
## RBC_Unit_type
## Mortality1Day 1 2 3 Sum
## 1 26 58 3 87
## 2 3 4 0 7
## 3 0 1 0 1
## Sum 29 63 3 95
## RBC_Unit_type
## Mortality1Day 1 2 3
## 1 0.27368421 0.61052632 0.03157895
## 2 0.03157895 0.04210526 0.00000000
## 3 0.00000000 0.01052632 0.00000000
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 1]]
## X-squared = 1.203, df = 4, p-value = 0.8776
##
## [1] "Adjusted p-value: 0.877598446034101"
## Leukoreduction
## Mortality1Day 1 2 Sum
## 1 3 84 87
## 2 0 7 7
## 3 0 1 1
## Sum 3 92 95
## Leukoreduction
## Mortality1Day 1 2
## 1 0.03157895 0.88421053
## 2 0.00000000 0.07368421
## 3 0.00000000 0.01052632
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 1]]
## X-squared = 0.28486, df = 2, p-value = 0.8672
##
## [1] "Adjusted p-value: 0.867249312726399"
## Irradiation
## Mortality1Day 1 2 Sum
## 1 26 61 87
## 2 2 5 7
## 3 1 0 1
## Sum 29 66 95
## Irradiation
## Mortality1Day 1 2
## 1 0.27368421 0.64210526
## 2 0.02105263 0.05263158
## 3 0.01052632 0.00000000
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 1]]
## X-squared = 2.3053, df = 2, p-value = 0.3158
##
## [1] "Adjusted p-value: 0.315791696970328"
## Washed
## Mortality1Day 1 2 Sum
## 1 85 2 87
## 2 7 0 7
## 3 1 0 1
## Sum 93 2 95
## Washed
## Mortality1Day 1 2
## 1 0.89473684 0.02105263
## 2 0.07368421 0.00000000
## 3 0.01052632 0.00000000
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 1]]
## X-squared = 0.18786, df = 2, p-value = 0.9103
##
## [1] "Adjusted p-value: 0.910345092473811"
## Plasma_reduction
## Mortality1Day 1 2 Sum
## 1 86 1 87
## 2 7 0 7
## 3 1 0 1
## Sum 94 1 95
## Plasma_reduction
## Mortality1Day 1 2
## 1 0.90526316 0.01052632
## 2 0.07368421 0.00000000
## 3 0.01052632 0.00000000
##
## Pearson's Chi-squared test
##
## data: contTable[[var - 1]]
## X-squared = 0.092932, df = 2, p-value = 0.9546
##
## [1] "Adjusted p-value: 0.954596893519514"5.1.1 Quantitative numerical variables (Wilcoxon test)
# For Wilcoxon test, binarize Mortality1Day ....
df_TAH_data_impute_mean[ which(df_TAH_data_impute_mean$Mortality1Day==3), ] <- 2
############################################################
# If need to adjust numeric and factor p-values separately #
############################################################
df_TAH_data_impute_mean_numeric <- df_TAH_data_impute_mean[, !indFactor]
tests_list <- lapply(colnames(df_TAH_data_impute_mean_numeric),
function(x) wilcox.test(
as.formula(paste0(x, "~ Mortality1Day")),
data = cbind(df_TAH_data_impute_mean_numeric,
Mortality1Day=df_TAH_data_impute_mean$Mortality1Day)))
numericPvalueResults <- sapply(tests_list, `[[`, "p.value")
numericPvalueResults_adj <- p.adjust(numericPvalueResults, method = 'fdr')
result <- cbind(QuantVariables=colnames(df_TAH_data_impute_mean[!indFactor]),
numeric_Mortality1Day_Pvalues=as.numeric(numericPvalueResults),
numeric_Mortality1Day_Pvalues_adj=
as.numeric(numericPvalueResults_adj))
result## QuantVariables numeric_Mortality1Day_Pvalues
## [1,] "Age_days" "0.151553445361102"
## [2,] "Weight" "0.054461343885384"
## [3,] "Pre_transfusion_K" "0.706953701787112"
## [4,] "Cr_at_transfusion" "0.103067633535562"
## [5,] "Age_of_RBC_unit" "0.295116461303598"
## [6,] "tV_12_weight_ml_Kg" "0.0314689872073928"
## [7,] "tV_12_eTBV" "0.00324443217529795"
## [8,] "Infusion_rate_hour_Pt_weight" "0.498518170274654"
## [9,] "Infusion_rate_hour_eTBV" "0.590997387053384"
## [10,] "Time_from_irradition_to_transfusion_min" "0.114192523260935"
## numeric_Mortality1Day_Pvalues_adj
## [1,] "0.252589075601837"
## [2,] "0.18153781295128"
## [3,] "0.706953701787112"
## [4,] "0.228385046521869"
## [5,] "0.421594944719426"
## [6,] "0.157344936036964"
## [7,] "0.0324443217529795"
## [8,] "0.623147712843317"
## [9,] "0.656663763392648"
## [10,] "0.228385046521869"6 Dimensionality Reduction
6.1 PCA
df_TAH_data_impute_mean <- as.data.frame(sapply(df_TAH_data_impute_mean, as.numeric))
pca1 <- prcomp(as.matrix(df_TAH_data_impute_mean), center = T)
summary(pca1)## Importance of components:
## PC1 PC2 PC3 PC4 PC5 PC6 PC7
## Standard deviation 3546.3724 1998.4663 50.01536 15.45642 8.9 5.157 1.628
## Proportion of Variance 0.7589 0.2410 0.00015 0.00001 0.0 0.000 0.000
## Cumulative Proportion 0.7589 0.9998 0.99998 0.99999 1.0 1.000 1.000
## PC8 PC9 PC10 PC11 PC12 PC13 PC14 PC15
## Standard deviation 1.008 0.8314 0.5916 0.564 0.5108 0.4832 0.4624 0.4464
## Proportion of Variance 0.000 0.0000 0.0000 0.000 0.0000 0.0000 0.0000 0.0000
## Cumulative Proportion 1.000 1.0000 1.0000 1.000 1.0000 1.0000 1.0000 1.0000
## PC16 PC17 PC18 PC19 PC20 PC21 PC22 PC23
## Standard deviation 0.4185 0.3865 0.3564 0.3232 0.302 0.2817 0.2749 0.2397
## Proportion of Variance 0.0000 0.0000 0.0000 0.0000 0.000 0.0000 0.0000 0.0000
## Cumulative Proportion 1.0000 1.0000 1.0000 1.0000 1.000 1.0000 1.0000 1.0000
## PC24 PC25 PC26 PC27 PC28 PC29 PC30 PC31
## Standard deviation 0.2312 0.2003 0.1953 0.1807 0.1628 0.1494 0.1394 0.1182
## Proportion of Variance 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
## Cumulative Proportion 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000
## PC32 PC33 PC34
## Standard deviation 0.09406 0.07904 0.0347
## Proportion of Variance 0.00000 0.00000 0.0000
## Cumulative Proportion 1.00000 1.00000 1.0000library(plotly)
plot_ly(x = c(1:length(pca1$sdev)), y = pca1$sdev*pca1$sdev, name = "Scree", type = "bar") %>%
layout(title="Scree Plot", xaxis = list(title="PC's"), yaxis = list(title="Variances (SD^2)"))scores <- pca1$x
# Loadings
loadings <- pca1$rotation
# Visualization scale factor for loadings
scaleLoad <- 10
# p <- plot_ly() %>%
# add_trace(x=scores[,1], y=scores[,2], z=scores[,3], type="scatter3d", mode="markers", name="",
# marker = list(color=scores[,2], colorscale = c("#FFE1A1", "#683531"), opacity = 0.7))
# for (k in 1:ncol(loadings)) {
# x <- c(0, loadings[1, k])*scaleLoad # Project PCAs only on the last 3 original data dimensions (6,5,4)
# y <- c(0, loadings[2, k])*scaleLoad
# z <- c(0, loadings[3, k])*scaleLoad
# p <- p %>% add_trace(x=x, y=y, z=z, type="scatter3d", mode="lines",
# name=paste0("Loading PC ", k, " ", colnames(df_TAH_data_impute_mean)[k]), line=list(width=8), opacity=1)
# # %>%
# # add_annotations( x = 0, y = 0, z = 0,
# # xref = "x", yref = "y", zref = "z",
# # # axref = "x", ayref = "y", azref = "z",
# # text = "", showarrow = T,
# # ax = c(0, loadings[k,1])*scaleLoad, ay = c(0, loadings[k,2])*scaleLoad, az = c(0,
# # loadings[k,3])*scaleLoad)
# }
#
# p <- p %>%
# layout(legend = list(orientation = 'h'), title="3D Projection of 6D Data along First 3 PCs",
# scene = list ( xaxis = list(title = rownames(loadings)[1]),
# yaxis = list(title = rownames(loadings)[2]),
# zaxis = list(title = rownames(loadings)[3])))
# p
loadings## PC1 PC2
## Group 1.432597e-04 -6.051592e-04
## TAH_occurrence -6.115045e-06 2.424250e-05
## Mortality1Day -6.643445e-06 1.401652e-05
## Sex -4.606691e-08 2.593938e-06
## Race 1.409822e-05 8.628286e-06
## Access 5.000452e-06 1.730294e-05
## Age_days 1.947683e-01 -9.807995e-01
## Weight 1.665779e-03 -8.342779e-03
## Pre_transfusion_K -9.876751e-06 4.746697e-05
## Cr_at_transfusion 1.644493e-06 -3.418279e-05
## Cardiac_dysf -1.093803e-05 3.643566e-05
## Respiratory_dysf -2.657478e-05 9.535280e-05
## Kidney_dysf -9.955980e-06 2.440058e-05
## Liver_dysf -1.146675e-05 1.429035e-05
## Intracranial_diso -9.649700e-06 1.498090e-05
## Orthopedics 1.669145e-06 -1.815357e-05
## Anemia 1.949051e-05 -8.737288e-05
## Hemoglobinopathy -3.573062e-06 -4.125291e-05
## Sepsis 1.343524e-06 3.342350e-05
## ECMO -9.302558e-06 1.740889e-05
## CPB -4.710171e-06 2.422605e-06
## Prematurity -3.164081e-05 8.911268e-05
## Dialysis 5.017988e-06 8.046405e-06
## RBC_Unit_type 2.146804e-06 -5.167423e-05
## Age_of_RBC_unit 4.907380e-04 -1.170085e-03
## Leukoreduction -9.141377e-06 -1.111745e-05
## Irradiation 2.715703e-05 4.224997e-05
## Washed 1.355439e-05 1.270107e-05
## Plasma_reduction -2.713212e-06 7.715669e-06
## tV_12_weight_ml_Kg -1.741111e-03 3.653784e-03
## tV_12_eTBV -2.490642e-05 5.137836e-05
## Infusion_rate_hour_Pt_weight -9.977001e-04 2.671995e-03
## Infusion_rate_hour_eTBV -1.438915e-05 3.935351e-05
## Time_from_irradition_to_transfusion_min 9.808457e-01 1.947832e-01
## PC3 PC4
## Group -7.542679e-04 -1.693229e-03
## TAH_occurrence -3.233289e-03 6.932950e-03
## Mortality1Day -1.231427e-03 2.588660e-03
## Sex 2.136534e-03 -2.747610e-04
## Race -6.324362e-03 -2.913385e-05
## Access 1.659270e-04 8.342737e-04
## Age_days -4.889595e-03 2.736328e-04
## Weight 6.680159e-03 -1.540786e-02
## Pre_transfusion_K 2.327459e-03 4.544406e-03
## Cr_at_transfusion 5.871514e-04 6.498611e-04
## Cardiac_dysf -2.773133e-03 -9.496110e-06
## Respiratory_dysf 5.628565e-04 7.922445e-04
## Kidney_dysf 5.332355e-05 2.052319e-03
## Liver_dysf -8.687839e-04 3.603438e-03
## Intracranial_diso 2.494954e-04 3.124555e-03
## Orthopedics 7.287055e-05 -1.007711e-03
## Anemia 4.958779e-04 1.129606e-03
## Hemoglobinopathy 6.048181e-04 -6.363681e-04
## Sepsis 4.470673e-04 2.860719e-03
## ECMO -2.609078e-03 5.934652e-04
## CPB -3.727258e-04 -1.320809e-03
## Prematurity 1.372234e-03 -1.163683e-03
## Dialysis 5.644165e-04 -5.679688e-04
## RBC_Unit_type 1.925665e-03 -5.734879e-03
## Age_of_RBC_unit 4.305812e-03 -3.084218e-02
## Leukoreduction -2.128141e-04 1.938091e-04
## Irradiation -3.397669e-04 -4.479050e-03
## Washed 2.646263e-04 -3.933427e-04
## Plasma_reduction 2.682400e-04 -2.944859e-04
## tV_12_weight_ml_Kg -7.995780e-01 5.996622e-01
## tV_12_eTBV -1.178709e-02 9.185166e-03
## Infusion_rate_hour_Pt_weight -6.002411e-01 -7.992880e-01
## Infusion_rate_hour_eTBV -8.530322e-03 -9.791716e-03
## Time_from_irradition_to_transfusion_min -1.072706e-03 2.393419e-04
## PC5 PC6
## Group -0.0134148862 0.0063909761
## TAH_occurrence 0.0121567814 -0.0148851748
## Mortality1Day 0.0022579561 0.0074436901
## Sex 0.0001377604 0.0046541401
## Race -0.0103148613 -0.0197798364
## Access -0.0018547446 -0.0210376188
## Age_days 0.0029852358 0.0080068021
## Weight -0.2104778150 -0.9754942889
## Pre_transfusion_K 0.0093767956 -0.0203448516
## Cr_at_transfusion 0.0012817387 -0.0330326565
## Cardiac_dysf 0.0050709619 -0.0004190476
## Respiratory_dysf 0.0037975669 0.0115724934
## Kidney_dysf 0.0052335901 -0.0246848517
## Liver_dysf -0.0023513075 -0.0015183787
## Intracranial_diso 0.0012171943 0.0061996660
## Orthopedics 0.0042395083 0.0069309119
## Anemia -0.0072636545 -0.0078626890
## Hemoglobinopathy -0.0050915045 0.0191308289
## Sepsis 0.0032357682 -0.0039948890
## ECMO 0.0040199506 -0.0059614887
## CPB 0.0037645733 -0.0006361846
## Prematurity 0.0069566854 0.0057695693
## Dialysis -0.0036174500 -0.0020440719
## RBC_Unit_type 0.0124969866 -0.0019131657
## Age_of_RBC_unit -0.9764654274 0.2102471575
## Leukoreduction 0.0056752107 -0.0025077354
## Irradiation 0.0067677066 0.0017646801
## Washed -0.0031094510 0.0016505892
## Plasma_reduction 0.0016759871 -0.0001569927
## tV_12_weight_ml_Kg -0.0245457730 -0.0093568863
## tV_12_eTBV 0.0008480285 -0.0007711355
## Infusion_rate_hour_Pt_weight 0.0233790771 0.0032968523
## Infusion_rate_hour_eTBV 0.0017855441 -0.0005586598
## Time_from_irradition_to_transfusion_min 0.0002362727 -0.0000520760
## PC7 PC8
## Group 1.866208e-03 -1.356795e-01
## TAH_occurrence 4.610002e-02 1.583997e-01
## Mortality1Day -5.286926e-03 5.764553e-02
## Sex -1.480451e-02 6.877814e-02
## Race -9.786705e-01 1.718002e-01
## Access -2.950752e-02 9.727948e-02
## Age_days -1.141410e-04 4.660049e-04
## Weight 1.370760e-02 -4.779933e-02
## Pre_transfusion_K 1.725560e-02 3.772608e-01
## Cr_at_transfusion 1.736937e-01 7.977324e-01
## Cardiac_dysf -3.300141e-02 -5.120213e-02
## Respiratory_dysf -3.777088e-02 -7.351877e-02
## Kidney_dysf 4.708253e-02 2.617228e-01
## Liver_dysf 1.478778e-02 1.268537e-01
## Intracranial_diso -1.463809e-02 2.666229e-02
## Orthopedics 8.444576e-03 -1.600929e-02
## Anemia -8.582766e-03 7.531518e-02
## Hemoglobinopathy 1.686616e-02 -1.664745e-02
## Sepsis 7.028408e-03 5.177050e-03
## ECMO -1.957606e-02 4.578194e-03
## CPB 1.649552e-02 -1.311198e-02
## Prematurity -2.510848e-02 5.294188e-02
## Dialysis 2.672747e-02 1.456317e-01
## RBC_Unit_type 1.721146e-02 4.759557e-02
## Age_of_RBC_unit 8.092024e-03 1.760391e-02
## Leukoreduction 9.345659e-04 9.094863e-03
## Irradiation -6.104120e-03 6.419218e-03
## Washed 1.104185e-02 3.956374e-02
## Plasma_reduction -9.000377e-03 8.120739e-03
## tV_12_weight_ml_Kg 5.068080e-03 -2.950959e-03
## tV_12_eTBV 3.033042e-03 7.945133e-03
## Infusion_rate_hour_Pt_weight 3.826434e-03 3.501348e-03
## Infusion_rate_hour_eTBV 2.377230e-03 1.735590e-02
## Time_from_irradition_to_transfusion_min 2.064839e-05 -1.673871e-07
## PC9 PC10
## Group 2.988287e-01 -0.3486149709
## TAH_occurrence -4.968640e-02 -0.0397305971
## Mortality1Day 2.293215e-02 -0.0115747814
## Sex 1.234783e-01 0.3953575730
## Race 5.217495e-02 -0.0292182783
## Access 2.584062e-02 -0.0439566937
## Age_days -2.508970e-04 0.0002226953
## Weight 2.833601e-03 0.0106484922
## Pre_transfusion_K -7.973041e-01 -0.0761514208
## Cr_at_transfusion 3.788188e-01 0.1011863360
## Cardiac_dysf 1.127680e-01 0.0289267366
## Respiratory_dysf 6.966383e-02 0.3601280281
## Kidney_dysf 2.515517e-02 -0.1272820327
## Liver_dysf 3.773003e-02 -0.1530685880
## Intracranial_diso -3.027407e-02 0.2284137793
## Orthopedics 7.945417e-02 -0.0282608818
## Anemia 3.077025e-02 -0.2193213839
## Hemoglobinopathy 3.834029e-02 -0.0553292470
## Sepsis 4.114433e-02 0.0986513155
## ECMO 6.245546e-02 0.0062762967
## CPB 6.239334e-02 0.0333941149
## Prematurity -5.845956e-02 0.2060736870
## Dialysis 1.404432e-01 -0.0227282445
## RBC_Unit_type 3.182812e-02 -0.5590682177
## Age_of_RBC_unit -1.180102e-02 -0.0014725409
## Leukoreduction -4.399758e-03 0.0362318509
## Irradiation 4.470433e-02 -0.2378600928
## Washed 8.843792e-02 0.0249278771
## Plasma_reduction 6.572556e-02 -0.0133511153
## tV_12_weight_ml_Kg -1.343343e-03 -0.0011745607
## tV_12_eTBV 1.334402e-01 -0.0379727408
## Infusion_rate_hour_Pt_weight -6.171643e-03 0.0040350737
## Infusion_rate_hour_eTBV 1.216809e-01 -0.0252512077
## Time_from_irradition_to_transfusion_min -7.136745e-06 0.0000174985
## PC11 PC12
## Group 3.834541e-01 -1.708349e-01
## TAH_occurrence 4.071590e-01 -1.722310e-01
## Mortality1Day 1.576901e-01 -2.097304e-01
## Sex -3.390846e-01 -2.766613e-01
## Race 2.278818e-02 3.207724e-02
## Access -2.779409e-02 6.143130e-02
## Age_days -1.270143e-04 1.530540e-04
## Weight -4.718912e-03 -9.104923e-03
## Pre_transfusion_K 9.008646e-02 -9.209415e-02
## Cr_at_transfusion -5.128065e-02 2.651143e-01
## Cardiac_dysf 1.542152e-01 -4.998878e-02
## Respiratory_dysf 9.115118e-03 -5.125172e-02
## Kidney_dysf 2.101655e-01 -2.195958e-01
## Liver_dysf 3.617788e-02 -2.737949e-01
## Intracranial_diso 1.885140e-01 -2.977424e-01
## Orthopedics 3.999647e-02 -6.198811e-02
## Anemia -5.344410e-01 -3.890683e-01
## Hemoglobinopathy -4.836307e-02 -1.973784e-01
## Sepsis 1.739089e-01 -2.316150e-01
## ECMO 8.263990e-02 9.990517e-02
## CPB 2.455889e-02 -9.524986e-02
## Prematurity -8.785861e-02 -3.325512e-01
## Dialysis 4.321171e-02 -4.387790e-02
## RBC_Unit_type -2.956266e-01 -9.230558e-02
## Age_of_RBC_unit 5.108446e-03 -5.503323e-04
## Leukoreduction 1.580600e-02 2.447699e-02
## Irradiation 1.311380e-02 1.777752e-01
## Washed 3.506532e-02 -4.801908e-02
## Plasma_reduction -1.457324e-03 -1.245296e-01
## tV_12_weight_ml_Kg -7.523439e-03 4.897431e-03
## tV_12_eTBV 5.783228e-02 -1.999638e-01
## Infusion_rate_hour_Pt_weight 1.751349e-03 -1.542832e-03
## Infusion_rate_hour_eTBV 3.303495e-02 -2.051101e-01
## Time_from_irradition_to_transfusion_min -1.624492e-05 -1.237418e-05
## PC13 PC14
## Group 5.885081e-01 3.958187e-02
## TAH_occurrence -9.924832e-02 1.443046e-01
## Mortality1Day -1.291629e-01 1.121386e-01
## Sex 3.446449e-01 -1.510368e-01
## Race 3.726932e-02 8.481475e-03
## Access -3.575298e-01 -1.191282e-02
## Age_days -4.110281e-04 -3.830111e-05
## Weight 2.067560e-03 4.549811e-03
## Pre_transfusion_K 1.640760e-01 -1.253072e-01
## Cr_at_transfusion 1.018058e-01 1.758736e-02
## Cardiac_dysf -1.498415e-01 -5.929982e-01
## Respiratory_dysf -1.671909e-01 5.021894e-02
## Kidney_dysf -4.365490e-02 -7.220431e-02
## Liver_dysf -1.127020e-01 -2.746009e-02
## Intracranial_diso -8.189081e-02 1.197524e-01
## Orthopedics -1.086715e-01 -1.546030e-05
## Anemia 4.662927e-02 1.809600e-01
## Hemoglobinopathy 4.002106e-02 -6.470718e-02
## Sepsis -4.145292e-02 3.737120e-02
## ECMO -2.944304e-01 -2.577192e-01
## CPB -4.560057e-02 -1.140027e-01
## Prematurity -1.636901e-01 2.167988e-01
## Dialysis -3.392812e-02 -3.797747e-02
## RBC_Unit_type -1.849737e-01 -2.671566e-01
## Age_of_RBC_unit -1.729586e-02 -4.001950e-03
## Leukoreduction -2.699395e-05 2.952279e-02
## Irradiation -2.024508e-01 5.583175e-01
## Washed -5.649996e-02 1.336594e-02
## Plasma_reduction -8.790749e-02 -1.717355e-02
## tV_12_weight_ml_Kg 3.524009e-03 9.394128e-04
## tV_12_eTBV -1.514113e-01 -2.264952e-02
## Infusion_rate_hour_Pt_weight 3.090503e-03 7.702962e-04
## Infusion_rate_hour_eTBV -1.635042e-01 -1.851413e-02
## Time_from_irradition_to_transfusion_min -5.625311e-06 -2.164171e-05
## PC15 PC16
## Group -9.803082e-02 -2.814824e-01
## TAH_occurrence -1.004371e-01 3.369533e-01
## Mortality1Day 1.803523e-03 -2.944676e-02
## Sex -4.845909e-01 -1.522022e-03
## Race 3.590127e-02 4.266920e-02
## Access -6.971938e-01 -2.189252e-01
## Age_days -5.605834e-05 1.018377e-04
## Weight 1.331155e-02 9.680872e-04
## Pre_transfusion_K -4.764796e-02 -2.769981e-01
## Cr_at_transfusion 1.411857e-01 -2.320130e-02
## Cardiac_dysf -8.204299e-03 -2.434503e-01
## Respiratory_dysf 3.526483e-01 -5.997082e-01
## Kidney_dysf 3.560970e-02 -7.327494e-02
## Liver_dysf -3.769230e-02 1.723313e-01
## Intracranial_diso 1.144260e-01 -1.016034e-01
## Orthopedics -4.751640e-03 -1.485347e-02
## Anemia 1.187570e-01 -1.973908e-01
## Hemoglobinopathy 6.784039e-02 6.662553e-02
## Sepsis -2.553553e-03 -3.215160e-02
## ECMO -7.885801e-03 -5.256630e-03
## CPB -5.109054e-02 -5.884583e-02
## Prematurity 8.407602e-02 2.143200e-01
## Dialysis 1.603626e-02 -3.580330e-02
## RBC_Unit_type 1.495965e-01 2.140437e-02
## Age_of_RBC_unit -1.203797e-03 1.310393e-03
## Leukoreduction -1.556070e-02 8.018843e-02
## Irradiation -1.991057e-01 -3.338728e-01
## Washed 3.908477e-02 1.314008e-02
## Plasma_reduction -2.284248e-02 1.552348e-02
## tV_12_weight_ml_Kg 7.157516e-04 -1.731770e-03
## tV_12_eTBV -2.065718e-02 -2.190345e-02
## Infusion_rate_hour_Pt_weight 6.770571e-05 9.331153e-04
## Infusion_rate_hour_eTBV -4.512375e-02 -2.039367e-02
## Time_from_irradition_to_transfusion_min 2.043836e-05 1.737691e-05
## PC17 PC18
## Group 3.238365e-02 1.582647e-01
## TAH_occurrence -1.604382e-01 1.781727e-01
## Mortality1Day 3.025421e-02 1.149140e-01
## Sex -4.080057e-01 -8.106493e-02
## Race -4.499932e-03 -9.560314e-03
## Access 3.524189e-01 3.522156e-01
## Age_days 6.345218e-05 -3.498175e-05
## Weight -1.565079e-02 -3.534590e-03
## Pre_transfusion_K -5.382123e-02 -9.262503e-02
## Cr_at_transfusion -3.982467e-03 4.444659e-02
## Cardiac_dysf -1.519711e-01 -1.000974e-01
## Respiratory_dysf -9.332737e-02 3.167400e-01
## Kidney_dysf 6.160914e-02 -1.567686e-01
## Liver_dysf 3.009857e-02 8.268856e-02
## Intracranial_diso 2.269084e-02 6.503941e-03
## Orthopedics -1.440160e-01 -5.608824e-02
## Anemia 3.952276e-01 -9.723036e-02
## Hemoglobinopathy 6.199028e-02 -3.110660e-01
## Sepsis -1.498333e-02 -8.843732e-04
## ECMO 5.060982e-02 -3.152335e-01
## CPB -2.229905e-02 -4.852337e-02
## Prematurity -2.008221e-01 2.132846e-01
## Dialysis 6.847270e-02 -1.925058e-02
## RBC_Unit_type -4.351580e-01 3.521243e-01
## Age_of_RBC_unit -1.466680e-02 3.240096e-03
## Leukoreduction 8.601871e-02 -1.968688e-02
## Irradiation -4.596765e-01 -3.588049e-01
## Washed 4.093748e-02 -1.752881e-01
## Plasma_reduction 2.161892e-03 -1.090438e-01
## tV_12_weight_ml_Kg -4.688257e-03 3.606054e-03
## tV_12_eTBV 3.907049e-02 -2.138644e-01
## Infusion_rate_hour_Pt_weight 3.230530e-03 4.018876e-03
## Infusion_rate_hour_eTBV 2.863948e-02 -2.201167e-01
## Time_from_irradition_to_transfusion_min 6.963581e-06 1.191890e-05
## PC19 PC20
## Group 8.922562e-02 -7.605264e-02
## TAH_occurrence -5.324519e-01 2.671661e-01
## Mortality1Day -1.290296e-01 -9.651783e-02
## Sex -1.325500e-01 1.268094e-01
## Race -2.687100e-02 -1.137198e-02
## Access 5.909106e-04 -3.717755e-02
## Age_days 3.138800e-05 1.643915e-04
## Weight -2.892534e-03 -1.482126e-02
## Pre_transfusion_K -1.297728e-02 -1.851041e-01
## Cr_at_transfusion 9.282249e-03 -1.634876e-01
## Cardiac_dysf 2.602250e-01 9.144936e-02
## Respiratory_dysf -2.696207e-01 3.992582e-02
## Kidney_dysf 1.857408e-01 6.593720e-01
## Liver_dysf 2.558534e-01 -2.952928e-01
## Intracranial_diso 8.490985e-02 -2.418004e-02
## Orthopedics -4.988288e-03 -1.199581e-01
## Anemia -9.842872e-02 2.142291e-01
## Hemoglobinopathy -3.268988e-01 -2.045846e-01
## Sepsis 5.834043e-02 -2.068358e-01
## ECMO -1.036334e-01 1.239013e-01
## CPB -1.423054e-01 -1.954441e-01
## Prematurity 4.778997e-01 1.430016e-02
## Dialysis 4.219187e-02 5.019983e-02
## RBC_Unit_type -1.252750e-01 -1.840547e-02
## Age_of_RBC_unit -2.652790e-03 8.461898e-03
## Leukoreduction 5.225102e-02 8.896793e-02
## Irradiation 9.977323e-02 2.995169e-02
## Washed 1.780924e-02 -9.463092e-02
## Plasma_reduction 3.376009e-02 -1.176460e-01
## tV_12_weight_ml_Kg 3.700448e-03 1.176415e-03
## tV_12_eTBV -7.750694e-02 -1.854629e-01
## Infusion_rate_hour_Pt_weight -2.632008e-04 2.504094e-03
## Infusion_rate_hour_eTBV -9.351681e-02 -1.859867e-01
## Time_from_irradition_to_transfusion_min -6.443096e-07 -1.566422e-06
## PC21 PC22
## Group -1.064735e-01 -1.034479e-01
## TAH_occurrence 4.797346e-02 -3.611862e-01
## Mortality1Day -3.079395e-02 2.408728e-01
## Sex 8.622409e-02 1.191933e-01
## Race -3.989169e-04 7.881584e-03
## Access -9.515641e-02 2.433031e-02
## Age_days 9.332369e-05 5.117800e-05
## Weight -4.397613e-04 2.917107e-04
## Pre_transfusion_K -5.830602e-02 -5.889866e-02
## Cr_at_transfusion -6.598738e-02 -3.983736e-02
## Cardiac_dysf -8.031734e-03 -3.835660e-01
## Respiratory_dysf 2.662536e-01 -4.741320e-02
## Kidney_dysf 1.377103e-01 3.022318e-01
## Liver_dysf 7.527573e-01 -5.945924e-02
## Intracranial_diso -3.229881e-01 1.566262e-01
## Orthopedics -1.007223e-01 5.910414e-02
## Anemia -1.012620e-02 -2.318705e-01
## Hemoglobinopathy 2.562431e-02 1.538636e-02
## Sepsis 6.898796e-03 4.773123e-01
## ECMO -1.145175e-02 8.076439e-02
## CPB -8.956636e-02 -2.558054e-01
## Prematurity -2.512241e-01 -2.762361e-01
## Dialysis -1.168044e-01 -4.459074e-02
## RBC_Unit_type -1.940629e-01 2.198157e-01
## Age_of_RBC_unit -3.669003e-03 -7.642351e-04
## Leukoreduction -1.473930e-01 -6.676164e-02
## Irradiation 8.460744e-02 -8.760052e-02
## Washed -1.050524e-01 -5.099813e-02
## Plasma_reduction 6.246597e-03 -1.154520e-01
## tV_12_weight_ml_Kg -1.870246e-03 2.004672e-03
## tV_12_eTBV -8.510955e-02 -1.062387e-02
## Infusion_rate_hour_Pt_weight 3.894522e-03 1.229138e-03
## Infusion_rate_hour_eTBV -1.397478e-01 -7.077423e-03
## Time_from_irradition_to_transfusion_min 8.749207e-07 2.713531e-06
## PC23 PC24
## Group -8.638814e-02 -1.751974e-01
## TAH_occurrence -1.100022e-01 1.798562e-01
## Mortality1Day 2.975846e-01 -2.949926e-01
## Sex 8.664283e-02 -5.185182e-02
## Race 3.989404e-03 1.843316e-02
## Access -1.426481e-01 -7.228618e-02
## Age_days 2.915352e-05 1.704684e-04
## Weight 9.635198e-04 -1.339797e-02
## Pre_transfusion_K 3.684338e-02 -7.371816e-02
## Cr_at_transfusion -1.087588e-01 4.229377e-02
## Cardiac_dysf -1.522185e-01 1.921749e-01
## Respiratory_dysf 5.967237e-03 -1.710500e-01
## Kidney_dysf 9.802692e-02 -6.909653e-02
## Liver_dysf 1.030194e-01 -8.753715e-03
## Intracranial_diso -4.341634e-03 2.401749e-01
## Orthopedics 3.160340e-01 6.331895e-02
## Anemia -6.074647e-02 2.414921e-01
## Hemoglobinopathy -4.747296e-01 -3.975491e-01
## Sepsis -3.863390e-01 4.443679e-01
## ECMO -1.879546e-01 -2.591858e-01
## CPB 2.644982e-01 2.485715e-01
## Prematurity -2.398892e-01 -3.281215e-01
## Dialysis 2.278111e-01 -1.785822e-01
## RBC_Unit_type -1.419498e-02 -2.031633e-03
## Age_of_RBC_unit 1.467693e-03 6.378527e-03
## Leukoreduction 2.105690e-01 -1.148012e-01
## Irradiation -7.959883e-02 4.591236e-02
## Washed 1.403403e-01 -2.741803e-02
## Plasma_reduction 4.074500e-02 1.478733e-02
## tV_12_weight_ml_Kg -1.792523e-03 -1.570332e-03
## tV_12_eTBV 1.456043e-01 -4.193131e-02
## Infusion_rate_hour_Pt_weight -2.044046e-03 1.754990e-03
## Infusion_rate_hour_eTBV 1.449391e-01 -3.367438e-02
## Time_from_irradition_to_transfusion_min 6.217692e-07 -1.359675e-05
## PC25 PC26
## Group 1.546565e-01 4.475748e-02
## TAH_occurrence 7.497623e-02 4.211024e-02
## Mortality1Day -6.095621e-02 -6.016039e-01
## Sex 9.218089e-02 -4.753928e-02
## Race -2.411331e-02 1.536663e-02
## Access -3.253125e-02 1.119606e-01
## Age_days -1.198242e-04 6.939376e-06
## Weight -6.461768e-05 -3.102288e-03
## Pre_transfusion_K 3.529048e-02 5.884296e-02
## Cr_at_transfusion 6.442779e-02 -3.252261e-02
## Cardiac_dysf -5.511798e-02 -1.808234e-01
## Respiratory_dysf -2.587226e-02 1.643230e-01
## Kidney_dysf -2.522782e-01 1.531958e-01
## Liver_dysf 1.797820e-01 -1.119443e-02
## Intracranial_diso 4.326919e-01 -1.644618e-01
## Orthopedics 1.824767e-01 6.348937e-01
## Anemia 1.261397e-01 -6.544289e-02
## Hemoglobinopathy -2.580544e-01 1.006258e-01
## Sepsis -2.488390e-01 5.427208e-02
## ECMO 4.763247e-01 -9.674277e-02
## CPB -4.207356e-01 -1.402108e-01
## Prematurity -9.669128e-02 8.210350e-02
## Dialysis -2.565294e-01 3.152168e-02
## RBC_Unit_type 3.243850e-02 -2.398335e-02
## Age_of_RBC_unit -9.253013e-04 1.306942e-03
## Leukoreduction -1.815797e-02 3.211503e-02
## Irradiation -4.132182e-02 -1.212818e-01
## Washed -2.137913e-02 2.418515e-02
## Plasma_reduction 5.917031e-02 -2.648600e-02
## tV_12_weight_ml_Kg -2.210829e-03 -1.916154e-04
## tV_12_eTBV -8.109784e-03 1.313747e-01
## Infusion_rate_hour_Pt_weight 7.968842e-04 -1.260836e-03
## Infusion_rate_hour_eTBV 2.033438e-02 1.110516e-01
## Time_from_irradition_to_transfusion_min 6.942507e-07 3.035767e-06
## PC27 PC28
## Group 5.487652e-02 1.527946e-01
## TAH_occurrence 9.848811e-02 -5.508869e-02
## Mortality1Day 3.260246e-01 -1.721718e-01
## Sex 1.567694e-02 8.718946e-03
## Race -1.178607e-02 3.882755e-03
## Access -1.080852e-01 -7.397995e-03
## Age_days 1.015959e-05 2.159460e-05
## Weight -4.020814e-03 -1.046771e-02
## Pre_transfusion_K 8.557794e-02 3.558125e-02
## Cr_at_transfusion 6.072555e-03 2.907371e-02
## Cardiac_dysf 1.063182e-01 -3.019428e-01
## Respiratory_dysf -3.573955e-03 9.614852e-02
## Kidney_dysf -1.738267e-01 1.577485e-01
## Liver_dysf -1.602970e-01 -1.958443e-03
## Intracranial_diso -5.659568e-01 -1.752624e-01
## Orthopedics 2.273541e-01 -2.516131e-01
## Anemia 2.140824e-01 -4.720207e-02
## Hemoglobinopathy -3.033666e-01 -2.444255e-01
## Sepsis 3.702782e-01 4.960224e-02
## ECMO 2.215644e-01 3.445363e-01
## CPB -2.236603e-01 4.565784e-01
## Prematurity 1.320150e-01 1.902347e-01
## Dialysis 4.245999e-02 -4.595803e-01
## RBC_Unit_type -1.263480e-01 6.570963e-02
## Age_of_RBC_unit 3.303741e-03 5.908917e-03
## Leukoreduction -3.819333e-02 2.215877e-02
## Irradiation -4.912000e-02 -6.057036e-02
## Washed 7.521698e-02 1.993512e-01
## Plasma_reduction 5.877751e-02 -6.347539e-02
## tV_12_weight_ml_Kg -1.759999e-03 -1.374744e-03
## tV_12_eTBV 3.898469e-04 1.361701e-01
## Infusion_rate_hour_Pt_weight 5.126384e-04 -2.221269e-03
## Infusion_rate_hour_eTBV -7.251901e-03 1.115549e-01
## Time_from_irradition_to_transfusion_min -1.205950e-05 -3.498915e-07
## PC29 PC30
## Group -3.782822e-02 -1.088297e-02
## TAH_occurrence -4.702969e-02 -3.630469e-02
## Mortality1Day 3.184632e-01 8.707828e-02
## Sex -4.942886e-02 -3.894799e-02
## Race 6.688113e-03 2.514238e-03
## Access 7.092162e-02 3.980494e-03
## Age_days -4.430746e-05 1.363426e-05
## Weight 4.445976e-03 1.727423e-04
## Pre_transfusion_K -5.768837e-02 -2.416048e-02
## Cr_at_transfusion 1.117500e-01 4.049269e-03
## Cardiac_dysf 1.933088e-01 -5.287446e-02
## Respiratory_dysf -5.286935e-02 -9.750494e-02
## Kidney_dysf 1.508267e-01 8.829237e-02
## Liver_dysf -7.437678e-02 -9.963406e-02
## Intracranial_diso -2.716334e-02 1.091395e-02
## Orthopedics 3.411441e-01 1.221592e-01
## Anemia 2.510874e-02 -1.100835e-02
## Hemoglobinopathy 2.242391e-01 -5.509513e-02
## Sepsis -1.474412e-01 -2.153964e-01
## ECMO -2.358183e-01 -1.441430e-03
## CPB 9.894009e-03 1.052323e-01
## Prematurity -2.552630e-03 9.581243e-02
## Dialysis -7.039441e-01 2.324085e-02
## RBC_Unit_type -5.905405e-02 -1.039470e-01
## Age_of_RBC_unit 9.628049e-04 -5.841892e-03
## Leukoreduction 7.763779e-02 -9.125804e-01
## Irradiation -1.617552e-02 -6.686907e-02
## Washed 1.449198e-01 -1.218055e-01
## Plasma_reduction 2.244678e-02 -6.886828e-02
## tV_12_weight_ml_Kg 1.673797e-03 -3.991557e-04
## tV_12_eTBV -1.394587e-01 3.481674e-02
## Infusion_rate_hour_Pt_weight 9.508919e-04 -1.146126e-03
## Infusion_rate_hour_eTBV -1.066936e-01 6.722608e-02
## Time_from_irradition_to_transfusion_min 7.505348e-06 -3.194662e-06
## PC31 PC32
## Group -5.356718e-02 -5.014129e-03
## TAH_occurrence 5.605142e-02 -4.304228e-02
## Mortality1Day -8.120990e-02 3.983671e-02
## Sex 1.572286e-02 -3.754125e-02
## Race -6.260417e-03 -5.429731e-04
## Access 2.163050e-02 -5.771428e-02
## Age_days 8.624457e-05 2.029822e-05
## Weight -4.213117e-03 -3.713717e-03
## Pre_transfusion_K 1.460721e-02 -3.770221e-03
## Cr_at_transfusion -1.240564e-02 8.185996e-02
## Cardiac_dysf 7.583238e-02 5.793855e-03
## Respiratory_dysf 6.511588e-02 -1.172888e-03
## Kidney_dysf 3.309621e-02 5.630493e-02
## Liver_dysf -9.144719e-02 -9.130943e-02
## Intracranial_diso -8.711660e-02 -8.561472e-02
## Orthopedics -3.651318e-01 -4.999071e-02
## Anemia -9.795592e-02 -3.044558e-02
## Hemoglobinopathy -1.362245e-01 -6.462375e-02
## Sepsis -7.525203e-02 -2.795858e-02
## ECMO -3.495821e-01 -5.861074e-02
## CPB -4.723506e-01 -1.590765e-02
## Prematurity -7.896534e-02 2.103774e-02
## Dialysis -1.509933e-01 -1.951273e-01
## RBC_Unit_type 5.576341e-02 -5.216816e-02
## Age_of_RBC_unit -7.666503e-04 5.763162e-03
## Leukoreduction -1.278485e-01 1.451587e-01
## Irradiation -5.440435e-02 -1.592236e-02
## Washed 3.755186e-01 -8.179367e-01
## Plasma_reduction 1.915768e-01 2.837635e-01
## tV_12_weight_ml_Kg -4.885361e-03 -4.547882e-03
## tV_12_eTBV 3.415771e-01 2.838256e-01
## Infusion_rate_hour_Pt_weight -3.174647e-03 -3.316594e-03
## Infusion_rate_hour_eTBV 3.176204e-01 2.477773e-01
## Time_from_irradition_to_transfusion_min -9.014439e-06 1.387547e-05
## PC33 PC34
## Group -2.296811e-02 -1.648235e-02
## TAH_occurrence 1.825510e-02 2.609544e-03
## Mortality1Day 5.602898e-02 2.020017e-02
## Sex 3.781985e-04 1.741697e-02
## Race 3.642831e-03 -2.570696e-04
## Access -2.901213e-02 1.487976e-02
## Age_days -3.756448e-06 2.473938e-06
## Weight 5.157194e-04 2.103469e-04
## Pre_transfusion_K -1.034070e-02 5.046357e-03
## Cr_at_transfusion 2.678472e-02 4.983163e-03
## Cardiac_dysf 1.493046e-01 7.384210e-03
## Respiratory_dysf -3.938874e-02 -2.232008e-02
## Kidney_dysf -7.731129e-02 -1.806091e-02
## Liver_dysf 7.598533e-02 -2.486299e-02
## Intracranial_diso -1.600259e-02 3.241580e-02
## Orthopedics -1.536367e-02 2.066699e-02
## Anemia 5.703261e-02 1.431381e-02
## Hemoglobinopathy -1.254831e-02 7.457714e-03
## Sepsis -1.487708e-02 -1.344448e-02
## ECMO -5.552682e-02 5.118036e-03
## CPB -8.654801e-02 2.068311e-03
## Prematurity 2.911480e-02 5.269973e-03
## Dialysis -6.555579e-02 -6.819291e-03
## RBC_Unit_type -4.112560e-02 5.979506e-03
## Age_of_RBC_unit -2.853993e-04 -3.765919e-05
## Leukoreduction 9.076232e-02 -1.499021e-02
## Irradiation 3.200197e-02 6.427024e-03
## Washed -1.143387e-01 1.030278e-02
## Plasma_reduction -8.908368e-01 -3.445801e-02
## tV_12_weight_ml_Kg -4.786689e-03 -9.760323e-03
## tV_12_eTBV 2.248565e-01 7.003395e-01
## Infusion_rate_hour_Pt_weight -3.216352e-03 9.384976e-03
## Infusion_rate_hour_eTBV 2.658436e-01 -7.091991e-01
## Time_from_irradition_to_transfusion_min 5.580138e-07 4.546956e-08a <- as.matrix(df_TAH_data_impute_mean[, c(2,5,10)])
p <- plot_ly() %>%
add_trace(x=a[,1], y=a[,2], z=a[,3], type="scatter3d", mode="markers", name="",
marker = list(color=a[,2], colorscale = c("#FFE1A1", "#683531"), opacity = 0.7))
for (k in 1:ncol(loadings)) {
x <- c(0, -loadings[1, k])*scaleLoad # Project PCAs only on the last 3 original data dimensions (6,5,4)
y <- c(0, -loadings[2, k])*scaleLoad
z <- c(0, -loadings[3, k])*scaleLoad
p <- p %>% add_trace(x=x, y=y, z=z, type="scatter3d", mode="lines",
name=paste0("Loading PC ", k, " ", colnames(df_TAH_data_impute_mean)[k]), line=list(width=8), opacity=1)
}
p <- p %>%
layout(legend = list(orientation = 'h'), title="Alternative 3D Projection of 3D Data along with First 3 PCs",
scene = list ( xaxis = list(title = rownames(loadings)[1]),
yaxis = list(title = rownames(loadings)[2]),
zaxis = list(title = rownames(loadings)[3])))
p6.2 Linear 2D modeling
# Define the 3D features
x <- df_TAH_data_impute_mean$Age_days
y <- df_TAH_data_impute_mean$tV_12_eTBV
z <- df_TAH_data_impute_mean$Infusion_rate_hour_eTBV
z1 <- df_TAH_data_impute_mean$TAH_occurrence
myDF <- data.frame(x, y, z)
### Fit a (bivariate-predictor) linear regression model
lm.fit <- lm(z ~ x+y)
str(df_TAH_data_impute_mean)## 'data.frame': 95 obs. of 34 variables:
## $ Group : num 1 1 1 1 2 1 1 1 1 1 ...
## $ TAH_occurrence : num 1 1 1 1 2 1 1 1 1 1 ...
## $ Mortality1Day : num 1 1 1 1 2 1 1 1 1 1 ...
## $ Sex : num 1 2 2 1 2 1 1 2 2 1 ...
## $ Race : num 5 5 5 5 2 1 1 1 2 1 ...
## $ Access : num 2 2 2 2 2 1 2 2 2 2 ...
## $ Age_days : num 26 9 19 13 2 12 23 2 8 24 ...
## $ Weight : num 4.35 0.63 0.99 4.7 2 ...
## $ Pre_transfusion_K : num 5.4 5.4 5.1 4.8 2 3.8 4.6 4.8 4.4 4.4 ...
## $ Cr_at_transfusion : num 0.28 0.89 0.63 0.55 2 0.45 0.3 0.7 0.5 0.4 ...
## $ Cardiac_dysf : num 2 1 1 2 2 1 1 1 1 2 ...
## $ Respiratory_dysf : num 1 2 2 2 2 2 1 1 1 2 ...
## $ Kidney_dysf : num 1 1 1 1 2 1 1 1 1 2 ...
## $ Liver_dysf : num 1 1 1 1 2 1 1 1 1 1 ...
## $ Intracranial_diso : num 1 1 1 1 2 1 1 1 1 1 ...
## $ Orthopedics : num 1 1 1 1 2 1 1 1 1 1 ...
## $ Anemia : num 1 2 2 1 2 1 1 1 1 1 ...
## $ Hemoglobinopathy : num 1 1 1 1 2 1 1 1 1 1 ...
## $ Sepsis : num 1 1 1 1 2 1 1 1 1 1 ...
## $ ECMO : num 2 1 1 2 2 1 1 1 1 2 ...
## $ CPB : num 1 1 1 1 2 1 1 1 1 1 ...
## $ Prematurity : num 1 2 2 1 2 1 2 1 2 1 ...
## $ Dialysis : num 1 1 1 1 2 1 1 1 1 1 ...
## $ RBC_Unit_type : num 2 2 2 2 2 2 1 1 1 2 ...
## $ Age_of_RBC_unit : num 8 7 9 5 2 4 14 14 14 8 ...
## $ Leukoreduction : num 2 2 2 2 2 2 2 2 2 2 ...
## $ Irradiation : num 1 1 1 2 2 1 2 2 2 2 ...
## $ Washed : num 1 1 1 1 2 1 1 1 1 1 ...
## $ Plasma_reduction : num 1 1 1 1 2 1 1 1 1 1 ...
## $ tV_12_weight_ml_Kg : num 13.79 33.33 25.25 4.79 2 ...
## $ tV_12_eTBV : num 0.16 0.33 0.25 0.06 2 0.21 0.2 0.18 0.21 0.16 ...
## $ Infusion_rate_hour_Pt_weight : num 9.2 13.33 14.43 2.99 2 ...
## $ Infusion_rate_hour_eTBV : num 0.11 0.13 0.14 0.04 2 0.14 0.05 0.09 0.07 0.06 ...
## $ Time_from_irradition_to_transfusion_min: num 118 118 118 52 2 118 116 30 67 43 ...# account for the categorical variables ....
df_TAH_data_impute_mean[ , factorVarNames[-1]] <- lapply(df_TAH_data_impute_mean[ , factorVarNames[-1]] , factor)
lm.fit <- glm(as.factor(TAH_occurrence) ~ Age_days + Age_of_RBC_unit + Irradiation + Time_from_irradition_to_transfusion_min +
tV_12_eTBV + Infusion_rate_hour_eTBV, data=df_TAH_data_impute_mean, family = "binomial")
summary(lm.fit)##
## Call:
## glm(formula = as.factor(TAH_occurrence) ~ Age_days + Age_of_RBC_unit +
## Irradiation + Time_from_irradition_to_transfusion_min + tV_12_eTBV +
## Infusion_rate_hour_eTBV, family = "binomial", data = df_TAH_data_impute_mean)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.4002 -0.8382 -0.4733 0.5344 2.1610
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.485e+00 8.529e-01 -1.742 0.08158
## Age_days 7.411e-05 1.365e-04 0.543 0.58726
## Age_of_RBC_unit -8.585e-02 4.053e-02 -2.118 0.03415
## Irradiation2 3.064e-01 6.240e-01 0.491 0.62335
## Time_from_irradition_to_transfusion_min 6.277e-05 7.465e-05 0.841 0.40039
## tV_12_eTBV 6.063e+00 2.250e+00 2.695 0.00703
## Infusion_rate_hour_eTBV -1.265e+00 1.643e+00 -0.770 0.44134
##
## (Intercept) .
## Age_days
## Age_of_RBC_unit *
## Irradiation2
## Time_from_irradition_to_transfusion_min
## tV_12_eTBV **
## Infusion_rate_hour_eTBV
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 126.074 on 94 degrees of freedom
## Residual deviance: 91.268 on 88 degrees of freedom
## AIC: 105.27
##
## Number of Fisher Scoring iterations: 7lm.fitReduced <- glm(as.factor(TAH_occurrence) ~ Age_of_RBC_unit + tV_12_eTBV, data=df_TAH_data_impute_mean, family = "binomial")
summary(lm.fitReduced)##
## Call:
## glm(formula = as.factor(TAH_occurrence) ~ Age_of_RBC_unit + tV_12_eTBV,
## family = "binomial", data = df_TAH_data_impute_mean)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.3061 -0.8725 -0.5078 0.4966 1.9397
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.05578 0.59395 -1.778 0.07547 .
## Age_of_RBC_unit -0.07176 0.03550 -2.022 0.04323 *
## tV_12_eTBV 4.82842 1.65546 2.917 0.00354 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 126.07 on 94 degrees of freedom
## Residual deviance: 93.45 on 92 degrees of freedom
## AIC: 99.45
##
## Number of Fisher Scoring iterations: 7hControls <- hist(df_TAH_data_TAH_Controls$tV_12_eTBV, breaks = 30, plot = FALSE, xlim=range(0,5))
hPatients <- hist(df_TAH_data_TAH_Patients$tV_12_eTBV, breaks = "FD", plot = FALSE)
h <- hist(df_TAH_data_impute_mean$tV_12_eTBV, breaks = "FD", plot = FALSE)
plControls <- plot_ly(x = hControls$mids, y = hControls$counts, name="Controls") %>% add_bars() # name = df_TAH_data_impute_mean$TAH_occurrence)
plPatients <- plot_ly(x = hPatients$mids, y = hPatients$counts, name="Patients") %>% add_bars()
subplot(plControls, plPatients, nrows = 2)7 Individual Factor-effects on Clinical Outcomes (TAH-occurance and Mortality1Day)
7.1 TAH-occurance
# Age_days, Pre_transfusion_K, Age_of_RBC_unit, tV_12_eTBV, Infusion_rate_hour_eTBV, Time_from_irradition_to_transfusion_min
plot_ly(data=df_TAH_data_impute_mean, x=~tV_12_eTBV, name=~TAH_occurrence, type="histogram", opacity=0.9, nbinsx=100, histnorm="probability") %>%
add_segments(x=0.3, xend=0.3, y=0, yend=0.3, line=list(width=4, color="green", dash='dot'), name="Cutoff") %>%
layout(title = "tV_12_eTBV Distributions by Cohort (Controls (1) & Patients (2))",
xaxis = list(title = "tV_12_eTBV"), yaxis = list(title = "Density"))plot_ly(data=df_TAH_data_impute_mean, x=~tV_12_weight_ml_Kg, name=~TAH_occurrence, type="histogram",
opacity=0.9, nbinsx=20, histnorm="probability") %>%
add_segments(x=30, xend=30, y=0, yend=0.7, line=list(width=4, color="green", dash='dot'), name="Cutoff") %>%
layout(title = "tV_12_weight_ml_Kg Distributions by Cohort (Controls (1) & Patients (2))",
xaxis = list(title = "tV_12_weight_ml_Kg"), yaxis = list(title = "Density"))plot_ly(data=df_TAH_data_impute_mean, x=~tV_12_eTBV, name=~TAH_occurrence, type="histogram", opacity=0.9, nbinsx=100, histnorm="probability") %>%
add_segments(x=0.3, xend=0.3, y=0, yend=0.3, line=list(width=4, color="green", dash='dot'), name="Cutoff") %>%
layout(title = "tV_12_eTBV Distributions by Cohort (Controls (1) & Patients (2))",
xaxis = list(title = "tV_12_eTBV"), yaxis = list(title = "Density"))plot_ly(data=df_TAH_data_impute_mean, x=~Age_days, name=~TAH_occurrence, type="histogram", opacity=0.9, nbinsx=100, histnorm="probability") %>%
# add_segments(x=0.3, xend=0.3, y=0, yend=0.3, line=list(width=4, color="green", dash='dot'), name="Cutoff") %>%
layout(title = "Age_days Distributions by Cohort (Controls (1) & Patients (2))",
xaxis = list(title = "Age_days"), yaxis = list(title = "Density"))plot_ly(data=df_TAH_data_impute_mean, x=~Pre_transfusion_K, name=~TAH_occurrence, type="histogram",
opacity=0.9, nbinsx=30, histnorm="probability") %>%
# add_segments(x=4.5, xend=4.5, y=0, yend=0.2, line=list(width=4, color="green", dash='dot'), name="Cutoff") %>%
layout(title = "Pre_transfusion_K Distributions by Cohort (Controls (1) & Patients (2))",
xaxis = list(title = "Pre_transfusion_K"), yaxis = list(title = "Density"))plot_ly(data=df_TAH_data_impute_mean, x=~Age_of_RBC_unit, name=~TAH_occurrence, type="histogram",
opacity=0.9, nbinsx=30, histnorm="probability") %>%
add_segments(x=7.95, xend=7.95, y=0, yend=0.25, line=list(width=4, color="green", dash='dot'), name="Cutoff") %>%
layout(title = "Age_of_RBC_unit Distributions by Cohort (Controls (1) & Patients (2))",
xaxis = list(title = "Age_of_RBC_unit"), yaxis = list(title = "Density"))plot_ly(data=df_TAH_data_impute_mean, x=~Infusion_rate_hour_eTBV, name=~TAH_occurrence, type="histogram",
opacity=0.9, nbinsx=20, histnorm="probability") %>%
# add_segments(x=0.3, xend=0.3, y=0, yend=0.3, line=list(width=4, color="green", dash='dot'), name="Cutoff") %>%
layout(title = "Infusion_rate_hour_eTBV Distributions by Cohort (Controls (1) & Patients (2))",
xaxis = list(title = "Infusion_rate_hour_eTBV"), yaxis = list(title = "Density"))# plot_ly(data=df_TAH_data_impute_mean, x=~Infusion_rate_hour_Pt_weight, name=~TAH_occurrence, type="histogram",
# opacity=0.9, nbinsx=20, histnorm="probability") %>%
# add_segments(x=10, xend=10, y=0, yend=0.5, line=list(width=4, color="green", dash='dot'), name="Cutoff") %>%
# layout(title = "Infusion_rate_hour_Pt_weight Distributions by Cohort (Controls (1) & Patients (2))",
# xaxis = list(title = "Infusion_rate_hour_Pt_weight"), yaxis = list(title = "Density"))
plot_ly(data=df_TAH_data_impute_mean, x=~Time_from_irradition_to_transfusion_min, name=~TAH_occurrence, type="histogram",
opacity=0.9, nbinsx=40, histnorm="probability") %>%
# add_segments(x=0.3, xend=0.3, y=0, yend=0.3, line=list(width=4, color="green", dash='dot'), name="Cutoff") %>%
layout(title = "Time_from_irradition_to_transfusion_min Distributions by Cohort (Controls (1) & Patients (2))",
xaxis = list(title = "Time from irradition to transfusion (min)"), yaxis = list(title = "Density"))plot_ly(data=df_TAH_data_impute_mean, x=~Age_days, name=~TAH_occurrence, type="histogram",
opacity=0.9, nbinsx=30, histnorm="probability") %>%
# add_segments(x=7.95, xend=7.95, y=0, yend=0.25, line=list(width=4, color="green", dash='dot'), name="Cutoff") %>%
layout(title = "Age (in days) Distributions by Cohort (Controls (1) & Patients (2))",
xaxis = list(title = "Age (days)"), yaxis = list(title = "Density"))plot_ly(data=df_TAH_data_impute_mean, x=~Weight, name=~TAH_occurrence, type="histogram",
opacity=0.9, nbinsx=30, histnorm="probability") %>%
# add_segments(x=7.95, xend=7.95, y=0, yend=0.25, line=list(width=4, color="green", dash='dot'), name="Cutoff") %>%
layout(title = "Weight Distributions by Cohort (Controls (1) & Patients (2))",
xaxis = list(title = "Weight"), yaxis = list(title = "Density"))# plot_ly(data=df_TAH_data_impute_mean, x=~Pre_transfusion_K, name=~TAH_occurrence, type="histogram",
# opacity=0.9, nbinsx=30, histnorm="probability") %>%
# add_segments(x=5.5, xend=5.5, y=0, yend=0.25, line=list(width=4, color="green", dash='dot'), name="Cutoff") %>%
# layout(title = "Pre-transfusion K Distributions by Cohort (Controls (1) & Patients (2))",
# xaxis = list(title = "Pre-transfusion K"), yaxis = list(title = "Density"))
# plot_ly(data=df_TAH_data_impute_mean, x=~Infusion_rate_hour_eTBV, name=~TAH_occurrence, type="histogram",
# opacity=0.9, nbinsx=30, histnorm="probability") %>%
# add_segments(x=0.15, xend=0.15, y=0, yend=0.7, line=list(width=4, color="green", dash='dot'), name="Cutoff") %>%
# layout(title = "Infusion rate/hour eTBV Distributions by Cohort (Controls (1) & Patients (2))",
# xaxis = list(title = "Infusion rate/hour eTBV"), yaxis = list(title = "Density"))
plot_ly(data=df_TAH_data_impute_mean, x=~ECMO, name=~TAH_occurrence, type="histogram",
opacity=0.9, nbinsx=30, histnorm="probability") %>%
# add_segments(x=1.5, xend=1.5, y=0, yend=0.7, line=list(width=4, color="green", dash='dot'), name="Cutoff") %>%
layout(title = "ECMO Distributions by Cohort (Controls (1) & Patients (2))",
xaxis = list(title = "ECMO"), yaxis = list(title = "Density"))plot_ly(data=df_TAH_data_impute_mean, x=~CPB, name=~TAH_occurrence, type="histogram",
opacity=0.9, nbinsx=30, histnorm="probability") %>%
# add_segments(x=1.5, xend=1.5, y=0, yend=0.7, line=list(width=4, color="green", dash='dot'), name="Cutoff") %>%
layout(title = "CPB Distributions by Cohort (Controls (1) & Patients (2))",
xaxis = list(title = "CPB"), yaxis = list(title = "Density"))7.2 Mortality1Day
# Age_days, Pre_transfusion_K, Age_of_RBC_unit, tV_12_eTBV, Infusion_rate_hour_eTBV, Time_from_irradition_to_transfusion_min
# pl_list <- list()
# for (i in 1:dim(df_TAH_data_impute_mean_2Way_Mortality)[2]) {
# if (i !=3 ) {
# pl_list[[i]] <- plot_ly(x=~df_TAH_data_impute_mean_2Way_Mortality[, i], name=~df_TAH_data_impute_mean_2Way_Mortality[, 3],
# type="histogram", opacity=0.9, nbinsx=10, histnorm="probability") %>%
# layout(title = paste0(colnames(df_TAH_data_impute_mean_2Way_Mortality)[i], " Distributions by 1-Day Mortality Cohort"),
# xaxis = list(title = colnames(df_TAH_data_impute_mean_2Way_Mortality)[i]), yaxis = list(title = "Density"))
# }
# }
#
# pl_list[-3] %>% subplot(nrows = 10) %>% layout(title="Mortality1Day")
plot_ly(data=df_TAH_data_impute_mean_2Way_Mortality, x=~Pre_transfusion_K, name=~Mortality1Day, type="histogram",
opacity=0.9, nbinsx=30, histnorm="probability") %>%
layout(title = "Pre-transfusion K Distributions by by 1-Day Mortality Cohort",
xaxis = list(title = "Pre-transfusion K"), yaxis = list(title = "Density"))plot_ly(data=df_TAH_data_impute_mean_2Way_Mortality, x=~Infusion_rate_hour_eTBV, name=~Mortality1Day, type="histogram",
opacity=0.9, nbinsx=30, histnorm="probability") %>%
layout(title = "Infusion_rate_hour_eTBVDistributions by by 1-Day Mortality Cohort",
xaxis = list(title = "Infusion_rate_hour_eTBV"), yaxis = list(title = "Density"))plot_ly(data=df_TAH_data_impute_mean_2Way_Mortality, x=~ECMO, name=~Mortality1Day, type="histogram",
opacity=0.9, nbinsx=30, histnorm="probability") %>%
layout(title = "ECMO Distributions by by 1-Day Mortality Cohort",
xaxis = list(title = "ECMO"), yaxis = list(title = "Density"))plot_ly(data=df_TAH_data_impute_mean_2Way_Mortality, x=~CPB, name=~Mortality1Day, type="histogram",
opacity=0.9, nbinsx=30, histnorm="probability") %>%
layout(title = "CPB Distributions by by 1-Day Mortality Cohort",
xaxis = list(title = "CPB"), yaxis = list(title = "Density"))plot_ly(data=df_TAH_data_impute_mean_2Way_Mortality, x=~Group, name=~Mortality1Day, type="histogram",
opacity=0.9, nbinsx=30, histnorm="probability") %>%
layout(title = "Group Distributions by by 1-Day Mortality Cohort",
xaxis = list(title = "Group"), yaxis = list(title = "Density"))8 Linear Modeling (TAH_occurrence)
# Backwards step linear model selection
lm_full.fit <- glm(as.factor(TAH_occurrence) ~ ., data=df_TAH_data_impute_mean, family = "binomial")
summary(lm.fit)##
## Call:
## glm(formula = as.factor(TAH_occurrence) ~ Age_days + Age_of_RBC_unit +
## Irradiation + Time_from_irradition_to_transfusion_min + tV_12_eTBV +
## Infusion_rate_hour_eTBV, family = "binomial", data = df_TAH_data_impute_mean)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.4002 -0.8382 -0.4733 0.5344 2.1610
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.485e+00 8.529e-01 -1.742 0.08158
## Age_days 7.411e-05 1.365e-04 0.543 0.58726
## Age_of_RBC_unit -8.585e-02 4.053e-02 -2.118 0.03415
## Irradiation2 3.064e-01 6.240e-01 0.491 0.62335
## Time_from_irradition_to_transfusion_min 6.277e-05 7.465e-05 0.841 0.40039
## tV_12_eTBV 6.063e+00 2.250e+00 2.695 0.00703
## Infusion_rate_hour_eTBV -1.265e+00 1.643e+00 -0.770 0.44134
##
## (Intercept) .
## Age_days
## Age_of_RBC_unit *
## Irradiation2
## Time_from_irradition_to_transfusion_min
## tV_12_eTBV **
## Infusion_rate_hour_eTBV
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 126.074 on 94 degrees of freedom
## Residual deviance: 91.268 on 88 degrees of freedom
## AIC: 105.27
##
## Number of Fisher Scoring iterations: 7backwards = stats::step(lm_full.fit)## Start: AIC=82
## as.factor(TAH_occurrence) ~ Group + Mortality1Day + Sex + Race +
## Access + Age_days + Weight + Pre_transfusion_K + Cr_at_transfusion +
## Cardiac_dysf + Respiratory_dysf + Kidney_dysf + Liver_dysf +
## Intracranial_diso + Orthopedics + Anemia + Hemoglobinopathy +
## Sepsis + ECMO + CPB + Prematurity + Dialysis + RBC_Unit_type +
## Age_of_RBC_unit + Leukoreduction + Irradiation + Washed +
## Plasma_reduction + tV_12_weight_ml_Kg + tV_12_eTBV + Infusion_rate_hour_Pt_weight +
## Infusion_rate_hour_eTBV + Time_from_irradition_to_transfusion_min
##
## Df Deviance AIC
## - Group 4 4.7732e-09 74
## - Race 4 9.2336e-09 74
## - RBC_Unit_type 2 1.8283e-08 78
## - Liver_dysf 1 3.3783e-09 80
## - ECMO 1 3.3863e-09 80
## - tV_12_eTBV 1 3.3901e-09 80
## - Irradiation 1 3.3994e-09 80
## - Sex 1 3.4087e-09 80
## - Hemoglobinopathy 1 3.4143e-09 80
## - Infusion_rate_hour_Pt_weight 1 3.4143e-09 80
## - Sepsis 1 3.4233e-09 80
## - Kidney_dysf 1 3.4246e-09 80
## - Infusion_rate_hour_eTBV 1 3.4322e-09 80
## - Washed 1 3.4483e-09 80
## - Cardiac_dysf 1 3.4524e-09 80
## - CPB 1 3.4541e-09 80
## - Age_days 1 3.4920e-09 80
## - Leukoreduction 1 3.5064e-09 80
## - Plasma_reduction 1 3.5106e-09 80
## - Mortality1Day 1 3.5574e-09 80
## - Age_of_RBC_unit 1 3.5938e-09 80
## - Respiratory_dysf 1 3.6503e-09 80
## - Access 1 3.6522e-09 80
## - tV_12_weight_ml_Kg 1 3.6643e-09 80
## - Dialysis 1 3.6947e-09 80
## - Pre_transfusion_K 1 3.7557e-09 80
## - Weight 1 3.8228e-09 80
## - Prematurity 1 3.8547e-09 80
## - Cr_at_transfusion 1 4.0420e-09 80
## - Intracranial_diso 1 4.5167e-09 80
## - Orthopedics 1 5.7284e-09 80
## - Time_from_irradition_to_transfusion_min 1 6.6280e-09 80
## - Anemia 1 1.7408e-08 80
## <none> 3.3893e-09 82
##
## Step: AIC=74
## as.factor(TAH_occurrence) ~ Mortality1Day + Sex + Race + Access +
## Age_days + Weight + Pre_transfusion_K + Cr_at_transfusion +
## Cardiac_dysf + Respiratory_dysf + Kidney_dysf + Liver_dysf +
## Intracranial_diso + Orthopedics + Anemia + Hemoglobinopathy +
## Sepsis + ECMO + CPB + Prematurity + Dialysis + RBC_Unit_type +
## Age_of_RBC_unit + Leukoreduction + Irradiation + Washed +
## Plasma_reduction + tV_12_weight_ml_Kg + tV_12_eTBV + Infusion_rate_hour_Pt_weight +
## Infusion_rate_hour_eTBV + Time_from_irradition_to_transfusion_min
##
## Df Deviance AIC
## - Race 4 0.00 66.00
## - RBC_Unit_type 2 0.00 70.00
## - Mortality1Day 1 0.00 72.00
## - Prematurity 1 0.00 72.00
## - ECMO 1 0.00 72.00
## - Dialysis 1 0.00 72.00
## - Liver_dysf 1 0.00 72.00
## - Infusion_rate_hour_eTBV 1 0.00 72.00
## - Leukoreduction 1 0.00 72.00
## - Age_days 1 0.00 72.00
## - Cr_at_transfusion 1 0.00 72.00
## - Sepsis 1 0.00 72.00
## - Hemoglobinopathy 1 0.00 72.00
## - Infusion_rate_hour_Pt_weight 1 0.00 72.00
## - Irradiation 1 0.00 72.00
## - Washed 1 0.00 72.00
## - Kidney_dysf 1 0.00 72.00
## - Sex 1 0.00 72.00
## - Plasma_reduction 1 0.00 72.00
## - Weight 1 0.00 72.00
## - CPB 1 0.00 72.00
## - tV_12_eTBV 1 0.00 72.00
## - Pre_transfusion_K 1 0.00 72.00
## - Age_of_RBC_unit 1 0.00 72.00
## - Cardiac_dysf 1 0.00 72.00
## - Access 1 0.00 72.00
## - Intracranial_diso 1 0.00 72.00
## - Time_from_irradition_to_transfusion_min 1 0.00 72.00
## - tV_12_weight_ml_Kg 1 0.00 72.00
## - Orthopedics 1 0.00 72.00
## - Respiratory_dysf 1 0.00 72.00
## <none> 0.00 74.00
## - Anemia 1 504.61 576.61
##
## Step: AIC=66
## as.factor(TAH_occurrence) ~ Mortality1Day + Sex + Access + Age_days +
## Weight + Pre_transfusion_K + Cr_at_transfusion + Cardiac_dysf +
## Respiratory_dysf + Kidney_dysf + Liver_dysf + Intracranial_diso +
## Orthopedics + Anemia + Hemoglobinopathy + Sepsis + ECMO +
## CPB + Prematurity + Dialysis + RBC_Unit_type + Age_of_RBC_unit +
## Leukoreduction + Irradiation + Washed + Plasma_reduction +
## tV_12_weight_ml_Kg + tV_12_eTBV + Infusion_rate_hour_Pt_weight +
## Infusion_rate_hour_eTBV + Time_from_irradition_to_transfusion_min
##
## Df Deviance AIC
## - tV_12_weight_ml_Kg 1 0.000 64.000
## - Mortality1Day 1 0.000 64.000
## - Plasma_reduction 1 0.000 64.000
## - ECMO 1 0.000 64.000
## - Liver_dysf 1 0.000 64.000
## - Infusion_rate_hour_Pt_weight 1 0.000 64.000
## - tV_12_eTBV 1 0.000 64.000
## - Infusion_rate_hour_eTBV 1 0.000 64.000
## - Leukoreduction 1 0.000 64.000
## - Sepsis 1 0.000 64.000
## - Age_of_RBC_unit 1 0.000 64.000
## - Washed 1 0.000 64.000
## - Sex 1 0.000 64.000
## - Hemoglobinopathy 1 0.000 64.000
## - Irradiation 1 0.000 64.000
## - Age_days 1 0.000 64.000
## - CPB 1 0.000 64.000
## - Intracranial_diso 1 0.000 64.000
## - Pre_transfusion_K 1 0.000 64.000
## - Access 1 0.000 64.000
## - Weight 1 0.000 64.000
## - Cardiac_dysf 1 0.000 64.000
## - Orthopedics 1 0.000 64.000
## - Cr_at_transfusion 1 0.000 64.000
## - Prematurity 1 0.000 64.000
## - Dialysis 1 0.000 64.000
## - Time_from_irradition_to_transfusion_min 1 0.000 64.000
## - Respiratory_dysf 1 0.000 64.000
## - Kidney_dysf 1 0.000 64.000
## <none> 0.000 66.000
## - Anemia 1 25.977 89.977
## - RBC_Unit_type 2 37.923 99.923
##
## Step: AIC=64
## as.factor(TAH_occurrence) ~ Mortality1Day + Sex + Access + Age_days +
## Weight + Pre_transfusion_K + Cr_at_transfusion + Cardiac_dysf +
## Respiratory_dysf + Kidney_dysf + Liver_dysf + Intracranial_diso +
## Orthopedics + Anemia + Hemoglobinopathy + Sepsis + ECMO +
## CPB + Prematurity + Dialysis + RBC_Unit_type + Age_of_RBC_unit +
## Leukoreduction + Irradiation + Washed + Plasma_reduction +
## tV_12_eTBV + Infusion_rate_hour_Pt_weight + Infusion_rate_hour_eTBV +
## Time_from_irradition_to_transfusion_min
##
## Df Deviance AIC
## - Mortality1Day 1 0.00 62.00
## - ECMO 1 0.00 62.00
## - Leukoreduction 1 0.00 62.00
## - Plasma_reduction 1 0.00 62.00
## - Liver_dysf 1 0.00 62.00
## - Sepsis 1 0.00 62.00
## - Infusion_rate_hour_Pt_weight 1 0.00 62.00
## - Washed 1 0.00 62.00
## - Sex 1 0.00 62.00
## - Hemoglobinopathy 1 0.00 62.00
## - Age_of_RBC_unit 1 0.00 62.00
## - Infusion_rate_hour_eTBV 1 0.00 62.00
## - Irradiation 1 0.00 62.00
## - Cardiac_dysf 1 0.00 62.00
## - Age_days 1 0.00 62.00
## - Pre_transfusion_K 1 0.00 62.00
## - CPB 1 0.00 62.00
## - Access 1 0.00 62.00
## - Weight 1 0.00 62.00
## - Intracranial_diso 1 0.00 62.00
## - Orthopedics 1 0.00 62.00
## - Cr_at_transfusion 1 0.00 62.00
## - Respiratory_dysf 1 0.00 62.00
## - Dialysis 1 0.00 62.00
## - Prematurity 1 0.00 62.00
## - Kidney_dysf 1 0.00 62.00
## <none> 0.00 64.00
## - Anemia 1 26.30 88.30
## - RBC_Unit_type 2 38.30 98.30
## - tV_12_eTBV 1 39.61 101.61
## - Time_from_irradition_to_transfusion_min 1 720.87 782.87
##
## Step: AIC=62
## as.factor(TAH_occurrence) ~ Sex + Access + Age_days + Weight +
## Pre_transfusion_K + Cr_at_transfusion + Cardiac_dysf + Respiratory_dysf +
## Kidney_dysf + Liver_dysf + Intracranial_diso + Orthopedics +
## Anemia + Hemoglobinopathy + Sepsis + ECMO + CPB + Prematurity +
## Dialysis + RBC_Unit_type + Age_of_RBC_unit + Leukoreduction +
## Irradiation + Washed + Plasma_reduction + tV_12_eTBV + Infusion_rate_hour_Pt_weight +
## Infusion_rate_hour_eTBV + Time_from_irradition_to_transfusion_min
##
## Df Deviance AIC
## - Sepsis 1 0.00 60.00
## - Plasma_reduction 1 0.00 60.00
## - ECMO 1 0.00 60.00
## - Leukoreduction 1 0.00 60.00
## - Liver_dysf 1 0.00 60.00
## - Infusion_rate_hour_Pt_weight 1 0.00 60.00
## - Washed 1 0.00 60.00
## - Hemoglobinopathy 1 0.00 60.00
## - Sex 1 0.00 60.00
## - Age_of_RBC_unit 1 0.00 60.00
## - Infusion_rate_hour_eTBV 1 0.00 60.00
## - Irradiation 1 0.00 60.00
## - CPB 1 0.00 60.00
## - Age_days 1 0.00 60.00
## - Intracranial_diso 1 0.00 60.00
## - Cardiac_dysf 1 0.00 60.00
## - Pre_transfusion_K 1 0.00 60.00
## - Weight 1 0.00 60.00
## - Cr_at_transfusion 1 0.00 60.00
## - Orthopedics 1 0.00 60.00
## - Access 1 0.00 60.00
## - Respiratory_dysf 1 0.00 60.00
## - Prematurity 1 0.00 60.00
## - Kidney_dysf 1 0.00 60.00
## - Dialysis 1 0.00 60.00
## <none> 0.00 62.00
## - Anemia 1 37.30 97.30
## - RBC_Unit_type 2 40.60 98.60
## - tV_12_eTBV 1 48.96 108.96
## - Time_from_irradition_to_transfusion_min 1 648.79 708.79
##
## Step: AIC=60
## as.factor(TAH_occurrence) ~ Sex + Access + Age_days + Weight +
## Pre_transfusion_K + Cr_at_transfusion + Cardiac_dysf + Respiratory_dysf +
## Kidney_dysf + Liver_dysf + Intracranial_diso + Orthopedics +
## Anemia + Hemoglobinopathy + ECMO + CPB + Prematurity + Dialysis +
## RBC_Unit_type + Age_of_RBC_unit + Leukoreduction + Irradiation +
## Washed + Plasma_reduction + tV_12_eTBV + Infusion_rate_hour_Pt_weight +
## Infusion_rate_hour_eTBV + Time_from_irradition_to_transfusion_min
##
## Df Deviance AIC
## - Plasma_reduction 1 0.000 58.000
## - ECMO 1 0.000 58.000
## - Liver_dysf 1 0.000 58.000
## - Infusion_rate_hour_Pt_weight 1 0.000 58.000
## - Leukoreduction 1 0.000 58.000
## - Washed 1 0.000 58.000
## - Hemoglobinopathy 1 0.000 58.000
## - Sex 1 0.000 58.000
## - Age_of_RBC_unit 1 0.000 58.000
## - Infusion_rate_hour_eTBV 1 0.000 58.000
## - Irradiation 1 0.000 58.000
## - CPB 1 0.000 58.000
## - Age_days 1 0.000 58.000
## - Intracranial_diso 1 0.000 58.000
## - Pre_transfusion_K 1 0.000 58.000
## - Cardiac_dysf 1 0.000 58.000
## - Weight 1 0.000 58.000
## - Cr_at_transfusion 1 0.000 58.000
## - Orthopedics 1 0.000 58.000
## - Access 1 0.000 58.000
## - Dialysis 1 0.000 58.000
## - Respiratory_dysf 1 0.000 58.000
## - Time_from_irradition_to_transfusion_min 1 0.000 58.000
## - Prematurity 1 0.000 58.000
## - Kidney_dysf 1 0.000 58.000
## <none> 0.000 60.000
## - Anemia 1 38.047 96.047
## - RBC_Unit_type 2 40.617 96.617
## - tV_12_eTBV 1 51.002 109.002
##
## Step: AIC=58
## as.factor(TAH_occurrence) ~ Sex + Access + Age_days + Weight +
## Pre_transfusion_K + Cr_at_transfusion + Cardiac_dysf + Respiratory_dysf +
## Kidney_dysf + Liver_dysf + Intracranial_diso + Orthopedics +
## Anemia + Hemoglobinopathy + ECMO + CPB + Prematurity + Dialysis +
## RBC_Unit_type + Age_of_RBC_unit + Leukoreduction + Irradiation +
## Washed + tV_12_eTBV + Infusion_rate_hour_Pt_weight + Infusion_rate_hour_eTBV +
## Time_from_irradition_to_transfusion_min
##
## Df Deviance AIC
## - Liver_dysf 1 0.00 56.00
## - ECMO 1 0.00 56.00
## - Leukoreduction 1 0.00 56.00
## - Infusion_rate_hour_Pt_weight 1 0.00 56.00
## - Washed 1 0.00 56.00
## - Hemoglobinopathy 1 0.00 56.00
## - Sex 1 0.00 56.00
## - Age_of_RBC_unit 1 0.00 56.00
## - Infusion_rate_hour_eTBV 1 0.00 56.00
## - Irradiation 1 0.00 56.00
## - CPB 1 0.00 56.00
## - Age_days 1 0.00 56.00
## - Intracranial_diso 1 0.00 56.00
## - Pre_transfusion_K 1 0.00 56.00
## - Cardiac_dysf 1 0.00 56.00
## - Weight 1 0.00 56.00
## - Cr_at_transfusion 1 0.00 56.00
## - Orthopedics 1 0.00 56.00
## - Access 1 0.00 56.00
## - Dialysis 1 0.00 56.00
## - Respiratory_dysf 1 0.00 56.00
## - Prematurity 1 0.00 56.00
## - Kidney_dysf 1 0.00 56.00
## <none> 0.00 58.00
## - RBC_Unit_type 2 40.82 94.82
## - Anemia 1 39.93 95.93
## - tV_12_eTBV 1 51.86 107.86
## - Time_from_irradition_to_transfusion_min 1 937.13 993.13
##
## Step: AIC=56
## as.factor(TAH_occurrence) ~ Sex + Access + Age_days + Weight +
## Pre_transfusion_K + Cr_at_transfusion + Cardiac_dysf + Respiratory_dysf +
## Kidney_dysf + Intracranial_diso + Orthopedics + Anemia +
## Hemoglobinopathy + ECMO + CPB + Prematurity + Dialysis +
## RBC_Unit_type + Age_of_RBC_unit + Leukoreduction + Irradiation +
## Washed + tV_12_eTBV + Infusion_rate_hour_Pt_weight + Infusion_rate_hour_eTBV +
## Time_from_irradition_to_transfusion_min
##
## Df Deviance AIC
## - ECMO 1 0.00 54.00
## - Leukoreduction 1 0.00 54.00
## - Hemoglobinopathy 1 0.00 54.00
## - Infusion_rate_hour_Pt_weight 1 0.00 54.00
## - Washed 1 0.00 54.00
## - Sex 1 0.00 54.00
## - Age_of_RBC_unit 1 0.00 54.00
## - Infusion_rate_hour_eTBV 1 0.00 54.00
## - Irradiation 1 0.00 54.00
## - CPB 1 0.00 54.00
## - Age_days 1 0.00 54.00
## - Pre_transfusion_K 1 0.00 54.00
## - Intracranial_diso 1 0.00 54.00
## - Weight 1 0.00 54.00
## - Cardiac_dysf 1 0.00 54.00
## - Dialysis 1 0.00 54.00
## - Cr_at_transfusion 1 0.00 54.00
## - Access 1 0.00 54.00
## - Prematurity 1 0.00 54.00
## - Respiratory_dysf 1 0.00 54.00
## - Kidney_dysf 1 0.00 54.00
## <none> 0.00 56.00
## - RBC_Unit_type 2 41.69 93.69
## - Anemia 1 41.95 95.95
## - tV_12_eTBV 1 59.09 113.09
## - Orthopedics 1 576.70 630.70
## - Time_from_irradition_to_transfusion_min 1 648.79 702.79
##
## Step: AIC=54
## as.factor(TAH_occurrence) ~ Sex + Access + Age_days + Weight +
## Pre_transfusion_K + Cr_at_transfusion + Cardiac_dysf + Respiratory_dysf +
## Kidney_dysf + Intracranial_diso + Orthopedics + Anemia +
## Hemoglobinopathy + CPB + Prematurity + Dialysis + RBC_Unit_type +
## Age_of_RBC_unit + Leukoreduction + Irradiation + Washed +
## tV_12_eTBV + Infusion_rate_hour_Pt_weight + Infusion_rate_hour_eTBV +
## Time_from_irradition_to_transfusion_min
##
## Df Deviance AIC
## - Leukoreduction 1 0.00 52.00
## - Infusion_rate_hour_Pt_weight 1 0.00 52.00
## - Hemoglobinopathy 1 0.00 52.00
## - Washed 1 0.00 52.00
## - Sex 1 0.00 52.00
## - Age_of_RBC_unit 1 0.00 52.00
## - Infusion_rate_hour_eTBV 1 0.00 52.00
## - Irradiation 1 0.00 52.00
## - CPB 1 0.00 52.00
## - Age_days 1 0.00 52.00
## - Pre_transfusion_K 1 0.00 52.00
## - Weight 1 0.00 52.00
## - Intracranial_diso 1 0.00 52.00
## - Access 1 0.00 52.00
## - Cr_at_transfusion 1 0.00 52.00
## - Prematurity 1 0.00 52.00
## - Dialysis 1 0.00 52.00
## <none> 0.00 54.00
## - RBC_Unit_type 2 41.94 91.94
## - Anemia 1 42.36 94.36
## - tV_12_eTBV 1 59.78 111.78
## - Orthopedics 1 360.44 412.44
## - Cardiac_dysf 1 504.61 556.61
## - Time_from_irradition_to_transfusion_min 1 576.70 628.70
## - Kidney_dysf 1 648.79 700.79
## - Respiratory_dysf 1 1081.31 1133.31
##
## Step: AIC=52
## as.factor(TAH_occurrence) ~ Sex + Access + Age_days + Weight +
## Pre_transfusion_K + Cr_at_transfusion + Cardiac_dysf + Respiratory_dysf +
## Kidney_dysf + Intracranial_diso + Orthopedics + Anemia +
## Hemoglobinopathy + CPB + Prematurity + Dialysis + RBC_Unit_type +
## Age_of_RBC_unit + Irradiation + Washed + tV_12_eTBV + Infusion_rate_hour_Pt_weight +
## Infusion_rate_hour_eTBV + Time_from_irradition_to_transfusion_min
##
## Df Deviance AIC
## - Infusion_rate_hour_Pt_weight 1 0.00 50.00
## - Hemoglobinopathy 1 0.00 50.00
## - Washed 1 0.00 50.00
## - Sex 1 0.00 50.00
## - Infusion_rate_hour_eTBV 1 0.00 50.00
## - Irradiation 1 0.00 50.00
## - CPB 1 0.00 50.00
## - Age_days 1 0.00 50.00
## - Pre_transfusion_K 1 0.00 50.00
## - Weight 1 0.00 50.00
## - Intracranial_diso 1 0.00 50.00
## - Cardiac_dysf 1 0.00 50.00
## - Access 1 0.00 50.00
## - Cr_at_transfusion 1 0.00 50.00
## - Prematurity 1 0.00 50.00
## - Age_of_RBC_unit 1 0.00 50.00
## - Dialysis 1 0.00 50.00
## - Respiratory_dysf 1 0.00 50.00
## <none> 0.00 52.00
## - RBC_Unit_type 2 41.96 89.96
## - Anemia 1 42.41 92.41
## - tV_12_eTBV 1 60.01 110.01
## - Time_from_irradition_to_transfusion_min 1 360.44 410.44
## - Kidney_dysf 1 576.70 626.70
## - Orthopedics 1 648.79 698.79
##
## Step: AIC=50
## as.factor(TAH_occurrence) ~ Sex + Access + Age_days + Weight +
## Pre_transfusion_K + Cr_at_transfusion + Cardiac_dysf + Respiratory_dysf +
## Kidney_dysf + Intracranial_diso + Orthopedics + Anemia +
## Hemoglobinopathy + CPB + Prematurity + Dialysis + RBC_Unit_type +
## Age_of_RBC_unit + Irradiation + Washed + tV_12_eTBV + Infusion_rate_hour_eTBV +
## Time_from_irradition_to_transfusion_min
##
## Df Deviance AIC
## - Hemoglobinopathy 1 0.000 48.000
## - Washed 1 0.000 48.000
## - Age_of_RBC_unit 1 0.000 48.000
## - Sex 1 0.000 48.000
## - Irradiation 1 0.000 48.000
## - CPB 1 0.000 48.000
## - Age_days 1 0.000 48.000
## - Infusion_rate_hour_eTBV 1 0.000 48.000
## - Pre_transfusion_K 1 0.000 48.000
## - Weight 1 0.000 48.000
## - Intracranial_diso 1 0.000 48.000
## - Cardiac_dysf 1 0.000 48.000
## - Orthopedics 1 0.000 48.000
## - Cr_at_transfusion 1 0.000 48.000
## - Access 1 0.000 48.000
## - Prematurity 1 0.000 48.000
## - Dialysis 1 0.000 48.000
## - Time_from_irradition_to_transfusion_min 1 0.000 48.000
## - Kidney_dysf 1 0.000 48.000
## - Respiratory_dysf 1 0.000 48.000
## <none> 0.000 50.000
## - RBC_Unit_type 2 41.965 87.965
## - Anemia 1 43.131 91.131
## - tV_12_eTBV 1 60.704 108.704
##
## Step: AIC=48
## as.factor(TAH_occurrence) ~ Sex + Access + Age_days + Weight +
## Pre_transfusion_K + Cr_at_transfusion + Cardiac_dysf + Respiratory_dysf +
## Kidney_dysf + Intracranial_diso + Orthopedics + Anemia +
## CPB + Prematurity + Dialysis + RBC_Unit_type + Age_of_RBC_unit +
## Irradiation + Washed + tV_12_eTBV + Infusion_rate_hour_eTBV +
## Time_from_irradition_to_transfusion_min
##
## Df Deviance AIC
## - Washed 1 0.000 46.000
## - Age_of_RBC_unit 1 0.000 46.000
## - Sex 1 0.000 46.000
## - Irradiation 1 0.000 46.000
## - CPB 1 0.000 46.000
## - Age_days 1 0.000 46.000
## - Infusion_rate_hour_eTBV 1 0.000 46.000
## - Pre_transfusion_K 1 0.000 46.000
## - Weight 1 0.000 46.000
## - Intracranial_diso 1 0.000 46.000
## - Cardiac_dysf 1 0.000 46.000
## - Orthopedics 1 0.000 46.000
## - Cr_at_transfusion 1 0.000 46.000
## - Access 1 0.000 46.000
## - Prematurity 1 0.000 46.000
## - Dialysis 1 0.000 46.000
## - Time_from_irradition_to_transfusion_min 1 0.000 46.000
## - Kidney_dysf 1 0.000 46.000
## - Respiratory_dysf 1 0.000 46.000
## <none> 0.000 48.000
## - RBC_Unit_type 2 42.808 86.808
## - Anemia 1 44.390 90.390
## - tV_12_eTBV 1 60.711 106.711
##
## Step: AIC=46
## as.factor(TAH_occurrence) ~ Sex + Access + Age_days + Weight +
## Pre_transfusion_K + Cr_at_transfusion + Cardiac_dysf + Respiratory_dysf +
## Kidney_dysf + Intracranial_diso + Orthopedics + Anemia +
## CPB + Prematurity + Dialysis + RBC_Unit_type + Age_of_RBC_unit +
## Irradiation + tV_12_eTBV + Infusion_rate_hour_eTBV + Time_from_irradition_to_transfusion_min
##
## Df Deviance AIC
## - Age_of_RBC_unit 1 0.000 44.000
## - Sex 1 0.000 44.000
## - Irradiation 1 0.000 44.000
## - CPB 1 0.000 44.000
## - Age_days 1 0.000 44.000
## - Infusion_rate_hour_eTBV 1 0.000 44.000
## - Pre_transfusion_K 1 0.000 44.000
## - Weight 1 0.000 44.000
## - Intracranial_diso 1 0.000 44.000
## - Cardiac_dysf 1 0.000 44.000
## - Orthopedics 1 0.000 44.000
## - Cr_at_transfusion 1 0.000 44.000
## - Dialysis 1 0.000 44.000
## - Access 1 0.000 44.000
## - Time_from_irradition_to_transfusion_min 1 0.000 44.000
## - Prematurity 1 0.000 44.000
## - Kidney_dysf 1 0.000 44.000
## - Respiratory_dysf 1 0.000 44.000
## <none> 0.000 46.000
## - RBC_Unit_type 2 43.297 85.297
## - Anemia 1 44.399 88.399
## - tV_12_eTBV 1 62.367 106.367
##
## Step: AIC=44
## as.factor(TAH_occurrence) ~ Sex + Access + Age_days + Weight +
## Pre_transfusion_K + Cr_at_transfusion + Cardiac_dysf + Respiratory_dysf +
## Kidney_dysf + Intracranial_diso + Orthopedics + Anemia +
## CPB + Prematurity + Dialysis + RBC_Unit_type + Irradiation +
## tV_12_eTBV + Infusion_rate_hour_eTBV + Time_from_irradition_to_transfusion_min
##
## Df Deviance AIC
## - Sex 1 0.000 42.000
## - CPB 1 0.000 42.000
## - Irradiation 1 0.000 42.000
## - Age_days 1 0.000 42.000
## - Infusion_rate_hour_eTBV 1 0.000 42.000
## - Pre_transfusion_K 1 0.000 42.000
## - Intracranial_diso 1 0.000 42.000
## - Weight 1 0.000 42.000
## - Orthopedics 1 0.000 42.000
## - Cardiac_dysf 1 0.000 42.000
## - Access 1 0.000 42.000
## - Cr_at_transfusion 1 0.000 42.000
## - Time_from_irradition_to_transfusion_min 1 0.000 42.000
## - Dialysis 1 0.000 42.000
## - Respiratory_dysf 1 0.000 42.000
## - Prematurity 1 0.000 42.000
## - Kidney_dysf 1 0.000 42.000
## <none> 0.000 44.000
## - RBC_Unit_type 2 47.443 87.443
## - Anemia 1 46.692 88.692
## - tV_12_eTBV 1 70.859 112.859
##
## Step: AIC=42
## as.factor(TAH_occurrence) ~ Access + Age_days + Weight + Pre_transfusion_K +
## Cr_at_transfusion + Cardiac_dysf + Respiratory_dysf + Kidney_dysf +
## Intracranial_diso + Orthopedics + Anemia + CPB + Prematurity +
## Dialysis + RBC_Unit_type + Irradiation + tV_12_eTBV + Infusion_rate_hour_eTBV +
## Time_from_irradition_to_transfusion_min
##
## Df Deviance AIC
## - Irradiation 1 0.000 40.000
## - CPB 1 0.000 40.000
## - Age_days 1 0.000 40.000
## - Pre_transfusion_K 1 0.000 40.000
## - Weight 1 0.000 40.000
## - Infusion_rate_hour_eTBV 1 0.000 40.000
## - Orthopedics 1 0.000 40.000
## - Cardiac_dysf 1 0.000 40.000
## - Cr_at_transfusion 1 0.000 40.000
## - Intracranial_diso 1 0.000 40.000
## - Access 1 0.000 40.000
## - Dialysis 1 0.000 40.000
## - Time_from_irradition_to_transfusion_min 1 0.000 40.000
## <none> 0.000 42.000
## - Prematurity 1 16.250 56.250
## - Respiratory_dysf 1 19.486 59.486
## - Kidney_dysf 1 19.731 59.731
## - Anemia 1 46.706 86.706
## - RBC_Unit_type 2 49.062 87.062
## - tV_12_eTBV 1 73.650 113.650
##
## Step: AIC=40
## as.factor(TAH_occurrence) ~ Access + Age_days + Weight + Pre_transfusion_K +
## Cr_at_transfusion + Cardiac_dysf + Respiratory_dysf + Kidney_dysf +
## Intracranial_diso + Orthopedics + Anemia + CPB + Prematurity +
## Dialysis + RBC_Unit_type + tV_12_eTBV + Infusion_rate_hour_eTBV +
## Time_from_irradition_to_transfusion_min
##
## Df Deviance AIC
## - CPB 1 0.00 38.00
## - Age_days 1 0.00 38.00
## - Pre_transfusion_K 1 0.00 38.00
## - Infusion_rate_hour_eTBV 1 0.00 38.00
## - Intracranial_diso 1 0.00 38.00
## - Orthopedics 1 0.00 38.00
## - Weight 1 0.00 38.00
## - Cr_at_transfusion 1 0.00 38.00
## - Dialysis 1 0.00 38.00
## - Access 1 0.00 38.00
## <none> 0.00 40.00
## - Cardiac_dysf 1 6.69 44.69
## - Prematurity 1 18.55 56.55
## - Kidney_dysf 1 20.45 58.45
## - Respiratory_dysf 1 21.41 59.41
## - Anemia 1 46.93 84.93
## - RBC_Unit_type 2 49.08 85.08
## - tV_12_eTBV 1 73.85 111.85
## - Time_from_irradition_to_transfusion_min 1 865.05 903.05
##
## Step: AIC=38
## as.factor(TAH_occurrence) ~ Access + Age_days + Weight + Pre_transfusion_K +
## Cr_at_transfusion + Cardiac_dysf + Respiratory_dysf + Kidney_dysf +
## Intracranial_diso + Orthopedics + Anemia + Prematurity +
## Dialysis + RBC_Unit_type + tV_12_eTBV + Infusion_rate_hour_eTBV +
## Time_from_irradition_to_transfusion_min
##
## Df Deviance AIC
## - Age_days 1 0.00 36.00
## - Pre_transfusion_K 1 0.00 36.00
## - Infusion_rate_hour_eTBV 1 0.00 36.00
## - Intracranial_diso 1 0.00 36.00
## - Orthopedics 1 0.00 36.00
## - Cr_at_transfusion 1 0.00 36.00
## - Dialysis 1 0.00 36.00
## - Cardiac_dysf 1 0.00 36.00
## <none> 0.00 38.00
## - Prematurity 1 19.39 55.39
## - Kidney_dysf 1 20.49 56.49
## - Respiratory_dysf 1 24.98 60.98
## - RBC_Unit_type 2 49.11 83.11
## - Anemia 1 48.39 84.39
## - tV_12_eTBV 1 77.79 113.79
## - Time_from_irradition_to_transfusion_min 1 360.44 396.44
## - Weight 1 865.05 901.05
## - Access 1 937.13 973.13
##
## Step: AIC=36
## as.factor(TAH_occurrence) ~ Access + Weight + Pre_transfusion_K +
## Cr_at_transfusion + Cardiac_dysf + Respiratory_dysf + Kidney_dysf +
## Intracranial_diso + Orthopedics + Anemia + Prematurity +
## Dialysis + RBC_Unit_type + tV_12_eTBV + Infusion_rate_hour_eTBV +
## Time_from_irradition_to_transfusion_min
##
## Df Deviance AIC
## - Pre_transfusion_K 1 0.00 34.00
## - Infusion_rate_hour_eTBV 1 0.00 34.00
## - Intracranial_diso 1 0.00 34.00
## - Orthopedics 1 0.00 34.00
## - Weight 1 0.00 34.00
## - Cr_at_transfusion 1 0.00 34.00
## - Time_from_irradition_to_transfusion_min 1 0.00 34.00
## <none> 0.00 36.00
## - Cardiac_dysf 1 12.28 46.28
## - Prematurity 1 20.15 54.15
## - Kidney_dysf 1 22.07 56.07
## - Respiratory_dysf 1 25.39 59.39
## - RBC_Unit_type 2 49.20 81.20
## - Anemia 1 49.11 83.11
## - tV_12_eTBV 1 78.63 112.63
## - Dialysis 1 576.70 610.70
## - Access 1 720.87 754.87
##
## Step: AIC=34
## as.factor(TAH_occurrence) ~ Access + Weight + Cr_at_transfusion +
## Cardiac_dysf + Respiratory_dysf + Kidney_dysf + Intracranial_diso +
## Orthopedics + Anemia + Prematurity + Dialysis + RBC_Unit_type +
## tV_12_eTBV + Infusion_rate_hour_eTBV + Time_from_irradition_to_transfusion_min
##
## Df Deviance AIC
## - Infusion_rate_hour_eTBV 1 0.00 32.00
## - Orthopedics 1 0.00 32.00
## - Intracranial_diso 1 0.00 32.00
## - Weight 1 0.04 32.04
## <none> 0.00 34.00
## - Cr_at_transfusion 1 11.78 43.78
## - Access 1 12.08 44.08
## - Time_from_irradition_to_transfusion_min 1 12.23 44.23
## - Cardiac_dysf 1 12.44 44.44
## - Prematurity 1 20.78 52.78
## - Kidney_dysf 1 24.46 56.46
## - Respiratory_dysf 1 25.40 57.40
## - RBC_Unit_type 2 50.11 80.11
## - Anemia 1 50.10 82.10
## - tV_12_eTBV 1 79.83 111.83
## - Dialysis 1 1153.40 1185.40
##
## Step: AIC=32
## as.factor(TAH_occurrence) ~ Access + Weight + Cr_at_transfusion +
## Cardiac_dysf + Respiratory_dysf + Kidney_dysf + Intracranial_diso +
## Orthopedics + Anemia + Prematurity + Dialysis + RBC_Unit_type +
## tV_12_eTBV + Time_from_irradition_to_transfusion_min
##
## Df Deviance AIC
## - Orthopedics 1 0.00 30.00
## - Weight 1 0.00 30.00
## <none> 0.00 32.00
## - Time_from_irradition_to_transfusion_min 1 13.14 43.14
## - Cr_at_transfusion 1 14.57 44.57
## - Access 1 14.97 44.97
## - Cardiac_dysf 1 15.31 45.31
## - Dialysis 1 15.76 45.76
## - Prematurity 1 21.04 51.04
## - Kidney_dysf 1 27.38 57.38
## - Respiratory_dysf 1 29.12 59.12
## - RBC_Unit_type 2 50.38 78.38
## - Anemia 1 50.35 80.35
## - tV_12_eTBV 1 87.28 117.28
## - Intracranial_diso 1 1009.22 1039.22
##
## Step: AIC=30
## as.factor(TAH_occurrence) ~ Access + Weight + Cr_at_transfusion +
## Cardiac_dysf + Respiratory_dysf + Kidney_dysf + Intracranial_diso +
## Anemia + Prematurity + Dialysis + RBC_Unit_type + tV_12_eTBV +
## Time_from_irradition_to_transfusion_min
##
## Df Deviance AIC
## - Intracranial_diso 1 0.000 28.000
## <none> 0.000 30.000
## - Weight 1 9.442 37.442
## - Time_from_irradition_to_transfusion_min 1 14.967 42.967
## - Cr_at_transfusion 1 15.806 43.806
## - Cardiac_dysf 1 16.527 44.527
## - Access 1 17.528 45.528
## - Dialysis 1 17.631 45.631
## - Prematurity 1 22.364 50.364
## - Kidney_dysf 1 27.591 55.591
## - Respiratory_dysf 1 29.502 57.502
## - RBC_Unit_type 2 50.426 76.426
## - Anemia 1 51.521 79.521
## - tV_12_eTBV 1 89.996 117.996
##
## Step: AIC=28
## as.factor(TAH_occurrence) ~ Access + Weight + Cr_at_transfusion +
## Cardiac_dysf + Respiratory_dysf + Kidney_dysf + Anemia +
## Prematurity + Dialysis + RBC_Unit_type + tV_12_eTBV + Time_from_irradition_to_transfusion_min
##
## Df Deviance AIC
## <none> 0.000 28.000
## - Cardiac_dysf 1 20.641 46.641
## - Time_from_irradition_to_transfusion_min 1 20.840 46.840
## - Cr_at_transfusion 1 21.171 47.171
## - Dialysis 1 22.368 48.368
## - Weight 1 23.319 49.319
## - Access 1 24.496 50.496
## - Respiratory_dysf 1 31.869 57.869
## - Prematurity 1 35.940 61.940
## - Kidney_dysf 1 39.583 65.583
## - RBC_Unit_type 2 51.488 75.488
## - Anemia 1 57.356 83.356
## - tV_12_eTBV 1 93.729 119.729formula(backwards)## as.factor(TAH_occurrence) ~ Access + Weight + Cr_at_transfusion +
## Cardiac_dysf + Respiratory_dysf + Kidney_dysf + Anemia +
## Prematurity + Dialysis + RBC_Unit_type + tV_12_eTBV + Time_from_irradition_to_transfusion_minsummary(backwards)##
## Call:
## glm(formula = as.factor(TAH_occurrence) ~ Access + Weight + Cr_at_transfusion +
## Cardiac_dysf + Respiratory_dysf + Kidney_dysf + Anemia +
## Prematurity + Dialysis + RBC_Unit_type + tV_12_eTBV + Time_from_irradition_to_transfusion_min,
## family = "binomial", data = df_TAH_data_impute_mean)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -2.277e-04 -2.100e-08 -2.100e-08 2.100e-08 2.150e-04
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.070e+03 8.926e+04 -0.012 0.990
## Access2 1.823e+02 1.973e+04 0.009 0.993
## Weight 5.166e+00 6.385e+02 0.008 0.994
## Cr_at_transfusion 1.038e+02 2.970e+04 0.003 0.997
## Cardiac_dysf2 -1.914e+02 1.649e+04 -0.012 0.991
## Respiratory_dysf2 -4.174e+02 3.395e+04 -0.012 0.990
## Kidney_dysf2 3.051e+02 2.934e+04 0.010 0.992
## Anemia2 -6.487e+02 5.351e+04 -0.012 0.990
## Prematurity2 3.307e+02 2.966e+04 0.011 0.991
## Dialysis2 -4.551e+02 7.155e+04 -0.006 0.995
## RBC_Unit_type2 6.421e+02 5.225e+04 0.012 0.990
## RBC_Unit_type3 3.182e+02 1.743e+05 0.002 0.999
## tV_12_eTBV 2.280e+03 1.861e+05 0.012 0.990
## Time_from_irradition_to_transfusion_min 1.702e-02 2.061e+00 0.008 0.993
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1.2607e+02 on 94 degrees of freedom
## Residual deviance: 3.7847e-07 on 81 degrees of freedom
## AIC: 28
##
## Number of Fisher Scoring iterations: 25# Forward step model selection process
baseMod <- glm(as.factor(TAH_occurrence) ~ 1, data=df_TAH_data_impute_mean, family = "binomial")
forwards = stats::step(baseMod, scope=list(lower=formula(baseMod), upper=formula(lm_full.fit)), direction="forward")## Start: AIC=128.07
## as.factor(TAH_occurrence) ~ 1
##
## Df Deviance AIC
## + tV_12_eTBV 1 98.627 102.63
## + tV_12_weight_ml_Kg 1 101.516 105.52
## + Mortality1Day 1 109.315 113.31
## + Kidney_dysf 1 109.454 113.45
## + Age_of_RBC_unit 1 117.696 121.70
## + Infusion_rate_hour_eTBV 1 117.923 121.92
## + Liver_dysf 1 118.425 122.42
## + Cr_at_transfusion 1 119.436 123.44
## + Infusion_rate_hour_Pt_weight 1 120.292 124.29
## + Pre_transfusion_K 1 120.998 125.00
## + Anemia 1 121.207 125.21
## + ECMO 1 122.688 126.69
## + Intracranial_diso 1 122.688 126.69
## + Sepsis 1 122.694 126.69
## + Leukoreduction 1 123.156 127.16
## + Group 4 117.309 127.31
## + Sex 1 123.594 127.59
## <none> 126.074 128.07
## + Hemoglobinopathy 1 124.420 128.42
## + Dialysis 1 124.926 128.93
## + Age_days 1 124.933 128.93
## + CPB 1 125.019 129.02
## + RBC_Unit_type 2 123.067 129.07
## + Respiratory_dysf 1 125.077 129.08
## + Prematurity 1 125.226 129.23
## + Weight 1 125.233 129.23
## + Access 1 125.287 129.29
## + Cardiac_dysf 1 125.795 129.79
## + Plasma_reduction 1 125.950 129.95
## + Time_from_irradition_to_transfusion_min 1 125.967 129.97
## + Irradiation 1 125.993 129.99
## + Washed 1 126.046 130.05
## + Orthopedics 1 126.064 130.06
## + Race 4 122.963 132.96
##
## Step: AIC=102.63
## as.factor(TAH_occurrence) ~ tV_12_eTBV
##
## Df Deviance AIC
## + Kidney_dysf 1 78.715 84.715
## + Pre_transfusion_K 1 86.142 92.142
## + Mortality1Day 1 89.816 95.816
## + Cr_at_transfusion 1 90.296 96.296
## + RBC_Unit_type 2 89.327 97.327
## + Age_of_RBC_unit 1 93.450 99.450
## + Respiratory_dysf 1 93.659 99.659
## + Liver_dysf 1 93.781 99.781
## + Cardiac_dysf 1 95.729 101.729
## + Anemia 1 96.274 102.274
## <none> 98.627 102.627
## + Leukoreduction 1 96.768 102.768
## + Access 1 96.944 102.944
## + Dialysis 1 97.032 103.032
## + Hemoglobinopathy 1 97.424 103.424
## + Sepsis 1 97.595 103.595
## + Sex 1 97.730 103.730
## + Prematurity 1 97.747 103.747
## + Washed 1 97.778 103.778
## + CPB 1 97.918 103.918
## + tV_12_weight_ml_Kg 1 97.927 103.927
## + Irradiation 1 97.998 103.998
## + Intracranial_diso 1 98.005 104.005
## + Plasma_reduction 1 98.040 104.040
## + Infusion_rate_hour_eTBV 1 98.109 104.109
## + Infusion_rate_hour_Pt_weight 1 98.226 104.226
## + Orthopedics 1 98.293 104.293
## + Time_from_irradition_to_transfusion_min 1 98.358 104.358
## + Weight 1 98.398 104.398
## + Age_days 1 98.621 104.621
## + ECMO 1 98.626 104.626
## + Group 4 93.955 105.955
## + Race 4 97.308 109.308
##
## Step: AIC=84.71
## as.factor(TAH_occurrence) ~ tV_12_eTBV + Kidney_dysf
##
## Df Deviance AIC
## + Cardiac_dysf 1 72.605 80.605
## + Respiratory_dysf 1 73.259 81.259
## + RBC_Unit_type 2 72.094 82.094
## + Mortality1Day 1 75.010 83.010
## + Pre_transfusion_K 1 75.328 83.328
## + Anemia 1 75.354 83.354
## + Age_of_RBC_unit 1 75.905 83.905
## + Washed 1 76.651 84.651
## <none> 78.715 84.715
## + Liver_dysf 1 77.451 85.451
## + Dialysis 1 77.656 85.656
## + Sex 1 77.658 85.658
## + tV_12_weight_ml_Kg 1 77.816 85.816
## + Leukoreduction 1 77.854 85.854
## + Prematurity 1 77.966 85.966
## + Time_from_irradition_to_transfusion_min 1 77.997 85.997
## + Irradiation 1 78.096 86.096
## + Weight 1 78.139 86.139
## + Access 1 78.145 86.145
## + Age_days 1 78.255 86.255
## + CPB 1 78.407 86.407
## + Plasma_reduction 1 78.425 86.425
## + ECMO 1 78.430 86.430
## + Cr_at_transfusion 1 78.449 86.449
## + Infusion_rate_hour_eTBV 1 78.459 86.459
## + Infusion_rate_hour_Pt_weight 1 78.503 86.503
## + Sepsis 1 78.664 86.664
## + Intracranial_diso 1 78.676 86.676
## + Hemoglobinopathy 1 78.701 86.701
## + Orthopedics 1 78.714 86.714
## + Group 4 73.931 87.931
## + Race 4 77.340 91.340
##
## Step: AIC=80.61
## as.factor(TAH_occurrence) ~ tV_12_eTBV + Kidney_dysf + Cardiac_dysf
##
## Df Deviance AIC
## + Anemia 1 67.458 77.458
## + Age_of_RBC_unit 1 68.348 78.348
## + RBC_Unit_type 2 66.844 78.844
## + Respiratory_dysf 1 69.559 79.559
## + Washed 1 69.854 79.854
## + Pre_transfusion_K 1 70.111 80.111
## + Mortality1Day 1 70.580 80.580
## <none> 72.605 80.605
## + Dialysis 1 71.100 81.100
## + Sex 1 71.446 81.446
## + Prematurity 1 71.853 81.853
## + Leukoreduction 1 71.868 81.868
## + Time_from_irradition_to_transfusion_min 1 71.993 81.993
## + Liver_dysf 1 72.032 82.032
## + Access 1 72.067 82.067
## + tV_12_weight_ml_Kg 1 72.076 82.076
## + Irradiation 1 72.362 82.362
## + Weight 1 72.379 82.379
## + Age_days 1 72.384 82.384
## + Intracranial_diso 1 72.388 82.388
## + CPB 1 72.512 82.512
## + Cr_at_transfusion 1 72.524 82.524
## + Sepsis 1 72.540 82.540
## + Hemoglobinopathy 1 72.560 82.560
## + Infusion_rate_hour_Pt_weight 1 72.562 82.562
## + Infusion_rate_hour_eTBV 1 72.562 82.562
## + ECMO 1 72.570 82.570
## + Plasma_reduction 1 72.572 82.572
## + Orthopedics 1 72.603 82.603
## + Group 4 68.535 84.535
## + Race 4 71.688 87.688
##
## Step: AIC=77.46
## as.factor(TAH_occurrence) ~ tV_12_eTBV + Kidney_dysf + Cardiac_dysf +
## Anemia
##
## Df Deviance AIC
## + RBC_Unit_type 2 50.436 64.436
## + Respiratory_dysf 1 62.694 74.694
## + Washed 1 63.078 75.078
## + Age_of_RBC_unit 1 64.100 76.100
## + Pre_transfusion_K 1 64.592 76.592
## + Age_days 1 64.840 76.840
## + Weight 1 64.912 76.912
## <none> 67.458 77.458
## + Mortality1Day 1 65.922 77.922
## + Time_from_irradition_to_transfusion_min 1 66.178 78.178
## + Liver_dysf 1 66.233 78.233
## + Dialysis 1 66.330 78.330
## + Access 1 66.454 78.454
## + Sex 1 66.662 78.662
## + Leukoreduction 1 67.060 79.060
## + Prematurity 1 67.124 79.124
## + Infusion_rate_hour_Pt_weight 1 67.292 79.292
## + tV_12_weight_ml_Kg 1 67.331 79.331
## + Hemoglobinopathy 1 67.343 79.343
## + ECMO 1 67.344 79.344
## + Infusion_rate_hour_eTBV 1 67.347 79.347
## + CPB 1 67.406 79.406
## + Intracranial_diso 1 67.408 79.408
## + Orthopedics 1 67.428 79.428
## + Sepsis 1 67.444 79.444
## + Plasma_reduction 1 67.446 79.446
## + Irradiation 1 67.448 79.448
## + Cr_at_transfusion 1 67.454 79.454
## + Group 4 61.809 79.809
## + Race 4 66.416 84.416
##
## Step: AIC=64.44
## as.factor(TAH_occurrence) ~ tV_12_eTBV + Kidney_dysf + Cardiac_dysf +
## Anemia + RBC_Unit_type
##
## Df Deviance AIC
## + Respiratory_dysf 1 43.685 59.685
## + Age_days 1 45.119 61.119
## + Weight 1 45.682 61.682
## + Pre_transfusion_K 1 47.260 63.260
## + Group 4 41.531 63.531
## + Intracranial_diso 1 48.071 64.071
## <none> 50.436 64.436
## + Time_from_irradition_to_transfusion_min 1 48.465 64.465
## + ECMO 1 48.685 64.685
## + Infusion_rate_hour_Pt_weight 1 48.715 64.715
## + Access 1 49.142 65.142
## + Infusion_rate_hour_eTBV 1 49.220 65.220
## + Mortality1Day 1 49.498 65.498
## + Liver_dysf 1 49.511 65.511
## + Washed 1 49.752 65.752
## + Orthopedics 1 49.754 65.754
## + Prematurity 1 49.822 65.822
## + Irradiation 1 50.029 66.029
## + Leukoreduction 1 50.058 66.058
## + Cr_at_transfusion 1 50.092 66.092
## + CPB 1 50.181 66.181
## + Sepsis 1 50.210 66.210
## + Dialysis 1 50.245 66.245
## + Sex 1 50.290 66.290
## + Hemoglobinopathy 1 50.420 66.420
## + Age_of_RBC_unit 1 50.421 66.421
## + tV_12_weight_ml_Kg 1 50.428 66.428
## + Plasma_reduction 1 50.434 66.434
## + Race 4 48.845 70.845
##
## Step: AIC=59.69
## as.factor(TAH_occurrence) ~ tV_12_eTBV + Kidney_dysf + Cardiac_dysf +
## Anemia + RBC_Unit_type + Respiratory_dysf
##
## Df Deviance AIC
## + Intracranial_diso 1 32.537 50.537
## + Prematurity 1 37.073 55.073
## + Pre_transfusion_K 1 38.378 56.378
## + tV_12_weight_ml_Kg 1 39.889 57.889
## + Mortality1Day 1 40.586 58.586
## + Group 4 35.530 59.530
## + ECMO 1 41.682 59.682
## <none> 43.685 59.685
## + Age_days 1 41.918 59.918
## + Washed 1 42.030 60.030
## + Weight 1 42.185 60.185
## + Orthopedics 1 42.549 60.549
## + Infusion_rate_hour_eTBV 1 42.573 60.573
## + Dialysis 1 42.682 60.682
## + Infusion_rate_hour_Pt_weight 1 42.692 60.692
## + Access 1 42.889 60.889
## + Time_from_irradition_to_transfusion_min 1 42.938 60.938
## + Liver_dysf 1 43.066 61.066
## + Sepsis 1 43.152 61.152
## + Age_of_RBC_unit 1 43.271 61.271
## + Irradiation 1 43.420 61.420
## + Hemoglobinopathy 1 43.483 61.483
## + Sex 1 43.485 61.485
## + Leukoreduction 1 43.539 61.539
## + Cr_at_transfusion 1 43.619 61.619
## + Plasma_reduction 1 43.681 61.681
## + CPB 1 43.682 61.682
## + Race 4 43.203 67.203
##
## Step: AIC=50.54
## as.factor(TAH_occurrence) ~ tV_12_eTBV + Kidney_dysf + Cardiac_dysf +
## Anemia + RBC_Unit_type + Respiratory_dysf + Intracranial_diso
##
## Df Deviance AIC
## + Orthopedics 1 28.965 48.965
## + Access 1 30.233 50.233
## + Age_days 1 30.319 50.319
## + Weight 1 30.363 50.363
## + Group 4 24.513 50.513
## <none> 32.537 50.537
## + Prematurity 1 30.627 50.627
## + Pre_transfusion_K 1 31.269 51.269
## + Hemoglobinopathy 1 31.597 51.597
## + Liver_dysf 1 31.641 51.641
## + Irradiation 1 31.833 51.833
## + Time_from_irradition_to_transfusion_min 1 31.838 51.838
## + Cr_at_transfusion 1 31.869 51.869
## + Mortality1Day 1 31.961 51.961
## + Washed 1 32.047 52.047
## + tV_12_weight_ml_Kg 1 32.131 52.131
## + Sex 1 32.197 52.197
## + ECMO 1 32.291 52.291
## + Dialysis 1 32.305 52.305
## + Sepsis 1 32.409 52.409
## + Leukoreduction 1 32.440 52.440
## + CPB 1 32.457 52.457
## + Age_of_RBC_unit 1 32.530 52.530
## + Infusion_rate_hour_Pt_weight 1 32.531 52.531
## + Plasma_reduction 1 32.537 52.537
## + Infusion_rate_hour_eTBV 1 32.537 52.537
## + Race 4 31.194 57.194
##
## Step: AIC=48.96
## as.factor(TAH_occurrence) ~ tV_12_eTBV + Kidney_dysf + Cardiac_dysf +
## Anemia + RBC_Unit_type + Respiratory_dysf + Intracranial_diso +
## Orthopedics
##
## Df Deviance AIC
## + Prematurity 1 26.204 48.204
## <none> 28.965 48.965
## + Liver_dysf 1 27.450 49.450
## + Access 1 27.457 49.457
## + Weight 1 27.672 49.672
## + Pre_transfusion_K 1 27.841 49.841
## + Sex 1 27.881 49.881
## + Age_days 1 27.992 49.993
## + Mortality1Day 1 28.267 50.267
## + Cr_at_transfusion 1 28.291 50.291
## + tV_12_weight_ml_Kg 1 28.293 50.293
## + Hemoglobinopathy 1 28.417 50.417
## + Washed 1 28.462 50.462
## + Irradiation 1 28.571 50.571
## + Group 4 22.736 50.736
## + Sepsis 1 28.739 50.739
## + Dialysis 1 28.747 50.747
## + ECMO 1 28.779 50.779
## + Time_from_irradition_to_transfusion_min 1 28.817 50.817
## + Age_of_RBC_unit 1 28.820 50.820
## + CPB 1 28.886 50.886
## + Leukoreduction 1 28.909 50.909
## + Infusion_rate_hour_Pt_weight 1 28.920 50.920
## + Infusion_rate_hour_eTBV 1 28.921 50.921
## + Plasma_reduction 1 28.965 50.965
## + Race 4 25.385 53.385
##
## Step: AIC=48.2
## as.factor(TAH_occurrence) ~ tV_12_eTBV + Kidney_dysf + Cardiac_dysf +
## Anemia + RBC_Unit_type + Respiratory_dysf + Intracranial_diso +
## Orthopedics + Prematurity
##
## Df Deviance AIC
## + Weight 1 21.152 45.152
## + Age_days 1 21.492 45.492
## + Access 1 22.861 46.861
## <none> 26.204 48.204
## + Group 4 18.480 48.480
## + Liver_dysf 1 24.933 48.933
## + Cr_at_transfusion 1 25.000 49.000
## + Mortality1Day 1 25.213 49.213
## + Infusion_rate_hour_Pt_weight 1 25.302 49.302
## + Pre_transfusion_K 1 25.376 49.376
## + Infusion_rate_hour_eTBV 1 25.562 49.562
## + Time_from_irradition_to_transfusion_min 1 25.572 49.572
## + Sex 1 25.688 49.688
## + Sepsis 1 25.791 49.791
## + Washed 1 25.815 49.815
## + Hemoglobinopathy 1 26.020 50.020
## + Age_of_RBC_unit 1 26.054 50.054
## + ECMO 1 26.108 50.107
## + Dialysis 1 26.136 50.136
## + Irradiation 1 26.145 50.145
## + CPB 1 26.149 50.149
## + Leukoreduction 1 26.177 50.177
## + tV_12_weight_ml_Kg 1 26.197 50.197
## + Plasma_reduction 1 26.203 50.203
## + Race 4 24.649 54.649
##
## Step: AIC=45.15
## as.factor(TAH_occurrence) ~ tV_12_eTBV + Kidney_dysf + Cardiac_dysf +
## Anemia + RBC_Unit_type + Respiratory_dysf + Intracranial_diso +
## Orthopedics + Prematurity + Weight
##
## Df Deviance AIC
## + Access 1 16.148 42.148
## + Infusion_rate_hour_Pt_weight 1 18.046 44.046
## + Infusion_rate_hour_eTBV 1 18.451 44.451
## <none> 21.152 45.152
## + Liver_dysf 1 19.523 45.523
## + Cr_at_transfusion 1 20.458 46.458
## + Pre_transfusion_K 1 20.465 46.465
## + Time_from_irradition_to_transfusion_min 1 20.519 46.519
## + Sex 1 20.559 46.559
## + Mortality1Day 1 20.624 46.624
## + Age_of_RBC_unit 1 20.930 46.930
## + Washed 1 20.991 46.991
## + Sepsis 1 21.002 47.002
## + Hemoglobinopathy 1 21.113 47.112
## + ECMO 1 21.114 47.114
## + CPB 1 21.133 47.133
## + tV_12_weight_ml_Kg 1 21.145 47.146
## + Dialysis 1 21.151 47.151
## + Leukoreduction 1 21.151 47.151
## + Irradiation 1 21.152 47.152
## + Age_days 1 21.152 47.152
## + Plasma_reduction 1 21.152 47.152
## + Race 4 15.464 47.464
## + Group 4 18.006 50.006
##
## Step: AIC=42.15
## as.factor(TAH_occurrence) ~ tV_12_eTBV + Kidney_dysf + Cardiac_dysf +
## Anemia + RBC_Unit_type + Respiratory_dysf + Intracranial_diso +
## Orthopedics + Prematurity + Weight + Access
##
## Df Deviance AIC
## + Infusion_rate_hour_eTBV 1 13.81 41.81
## + Infusion_rate_hour_Pt_weight 1 13.86 41.86
## <none> 16.15 42.15
## + Dialysis 1 15.37 43.37
## + Age_of_RBC_unit 1 15.37 43.37
## + Liver_dysf 1 15.61 43.61
## + Time_from_irradition_to_transfusion_min 1 15.91 43.91
## + Irradiation 1 15.92 43.92
## + Mortality1Day 1 16.05 44.05
## + Cr_at_transfusion 1 16.06 44.06
## + Sepsis 1 16.09 44.09
## + Hemoglobinopathy 1 16.09 44.09
## + tV_12_weight_ml_Kg 1 16.10 44.10
## + Pre_transfusion_K 1 16.11 44.11
## + Washed 1 16.12 44.12
## + ECMO 1 16.13 44.13
## + Age_days 1 16.14 44.14
## + CPB 1 16.15 44.15
## + Plasma_reduction 1 16.15 44.15
## + Leukoreduction 1 16.15 44.15
## + Race 4 13.87 47.87
## + Group 4 15.66 49.66
## + Sex 1 865.05 893.05
##
## Step: AIC=41.81
## as.factor(TAH_occurrence) ~ tV_12_eTBV + Kidney_dysf + Cardiac_dysf +
## Anemia + RBC_Unit_type + Respiratory_dysf + Intracranial_diso +
## Orthopedics + Prematurity + Weight + Access + Infusion_rate_hour_eTBV
##
## Df Deviance AIC
## + Sex 1 0.0000 30.000
## + Pre_transfusion_K 1 10.7276 40.728
## + Time_from_irradition_to_transfusion_min 1 11.7763 41.776
## <none> 13.8146 41.815
## + Age_of_RBC_unit 1 12.5904 42.590
## + Liver_dysf 1 12.7831 42.783
## + Irradiation 1 13.2617 43.262
## + Cr_at_transfusion 1 13.4501 43.450
## + tV_12_weight_ml_Kg 1 13.5202 43.520
## + Age_days 1 13.5458 43.546
## + Hemoglobinopathy 1 13.7692 43.769
## + CPB 1 13.7826 43.783
## + Dialysis 1 13.7854 43.785
## + ECMO 1 13.7929 43.793
## + Infusion_rate_hour_Pt_weight 1 13.8085 43.808
## + Washed 1 13.8108 43.811
## + Mortality1Day 1 13.8117 43.812
## + Sepsis 1 13.8146 43.815
## + Plasma_reduction 1 13.8146 43.815
## + Leukoreduction 1 13.8146 43.815
## + Race 4 8.5957 44.596
## + Group 4 12.8173 48.817
##
## Step: AIC=30
## as.factor(TAH_occurrence) ~ tV_12_eTBV + Kidney_dysf + Cardiac_dysf +
## Anemia + RBC_Unit_type + Respiratory_dysf + Intracranial_diso +
## Orthopedics + Prematurity + Weight + Access + Infusion_rate_hour_eTBV +
## Sex
##
## Df Deviance AIC
## <none> 0.0 30.0
## + Age_of_RBC_unit 1 0.0 32.0
## + Irradiation 1 0.0 32.0
## + Pre_transfusion_K 1 0.0 32.0
## + Cr_at_transfusion 1 0.0 32.0
## + Dialysis 1 0.0 32.0
## + ECMO 1 0.0 32.0
## + CPB 1 0.0 32.0
## + Hemoglobinopathy 1 0.0 32.0
## + Sepsis 1 0.0 32.0
## + Plasma_reduction 1 0.0 32.0
## + Leukoreduction 1 0.0 32.0
## + Washed 1 0.0 32.0
## + Mortality1Day 1 0.0 32.0
## + Race 4 0.0 38.0
## + Group 4 0.0 38.0
## + Time_from_irradition_to_transfusion_min 1 576.7 608.7
## + Liver_dysf 1 720.9 752.9
## + Age_days 1 865.0 897.0
## + Infusion_rate_hour_Pt_weight 1 1225.5 1257.5
## + tV_12_weight_ml_Kg 1 3390.6 3422.6formula(forwards)## as.factor(TAH_occurrence) ~ tV_12_eTBV + Kidney_dysf + Cardiac_dysf +
## Anemia + RBC_Unit_type + Respiratory_dysf + Intracranial_diso +
## Orthopedics + Prematurity + Weight + Access + Infusion_rate_hour_eTBV +
## Sexsummary(forwards)##
## Call:
## glm(formula = as.factor(TAH_occurrence) ~ tV_12_eTBV + Kidney_dysf +
## Cardiac_dysf + Anemia + RBC_Unit_type + Respiratory_dysf +
## Intracranial_diso + Orthopedics + Prematurity + Weight +
## Access + Infusion_rate_hour_eTBV + Sex, family = "binomial",
## data = df_TAH_data_impute_mean)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -4.844e-04 -2.000e-08 -2.000e-08 2.000e-08 3.862e-04
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -2771.98 136568.26 -0.020 0.984
## tV_12_eTBV 5528.03 273482.73 0.020 0.984
## Kidney_dysf2 603.59 32060.88 0.019 0.985
## Cardiac_dysf2 -1234.85 1534899.83 -0.001 0.999
## Anemia2 -1743.08 86202.48 -0.020 0.984
## RBC_Unit_type2 1675.58 82332.37 0.020 0.984
## RBC_Unit_type3 647.35 169521.48 0.004 0.997
## Respiratory_dysf2 -698.78 36519.57 -0.019 0.985
## Intracranial_diso2 646.63 1532717.06 0.000 1.000
## Orthopedics2 309.36 23306.62 0.013 0.989
## Prematurity2 805.95 39588.73 0.020 0.984
## Weight 18.87 933.40 0.020 0.984
## Access2 480.68 24090.86 0.020 0.984
## Infusion_rate_hour_eTBV -760.83 70586.62 -0.011 0.991
## Sex2 331.83 17147.07 0.019 0.985
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1.2607e+02 on 94 degrees of freedom
## Residual deviance: 1.1911e-06 on 80 degrees of freedom
## AIC: 30
##
## Number of Fisher Scoring iterations: 25# Other Models
lm.fit <- glm(as.factor(Mortality1Day) ~ Age_days + Irradiation, # + Time_from_irradition_to_transfusion_min + tV_12_eTBV,
data=df_TAH_data_impute_mean, family = "binomial")
summary(lm.fit)##
## Call:
## glm(formula = as.factor(Mortality1Day) ~ Age_days + Irradiation,
## family = "binomial", data = df_TAH_data_impute_mean)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.5095 -0.4988 -0.4291 -0.2883 2.5731
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -2.1247321 0.7927310 -2.680 0.00736 **
## Age_days -0.0002729 0.0002413 -1.131 0.25803
## Irradiation2 0.1486727 0.8582696 0.173 0.86248
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 54.898 on 94 degrees of freedom
## Residual deviance: 53.202 on 92 degrees of freedom
## AIC: 59.202
##
## Number of Fisher Scoring iterations: 6coef.lm.fit <- coef(lm.fit)
### Reparameterize the 2D (x,y) grid, and define the corresponding model values z on the grid
# x.seq <- seq(min(x),max(x),length.out=100)
# y.seq <- seq(min(y),max(y),length.out=100)
# z.seq <- function(x,y) coef.lm.fit[1]+coef.lm.fit[2]*x+coef.lm.fit[3]*y
# # define the values of z = z(x.seq, y.seq), as a Matrix of dimension c(dim(x.seq), dim(y.seq))
# z <- t(outer(x.seq, y.seq, z.seq))
#
# #Setup Axis
# axis_x <- seq(min(x), max(x), length.out=100)
# axis_y <- seq(min(y), max(y), length.out=100)
#
# #Sample points
# library(reshape2)
# lm_surface <- expand.grid(x = axis_x, y = axis_y, KEEP.OUT.ATTRS = F)
# lm_surface$z <- predict.lm(lm.fit, newdata = lm_surface)
# lm_surface <- acast(lm_surface, x ~ y, value.var = "z") # z ~ 0 + x + y
#
# plot_ly(myDF, x = ~x, y = ~y, z = ~z,
# text = df_TAH_data_impute_mean$TAH_Occurrence, type="scatter3d", mode="markers",
# color=df_TAH_data_impute_mean$Mortality1Day) %>%
# add_trace(x=~axis_x, y=~axis_y, z=~lm_surface, type="surface", color="gray", name="LM model", opacity=0.3) %>%
# layout(title="3D Plane Regression (Infusion_rate_hour ~ Age_days + eTBV); Color=Mortality", showlegend = F,
# scene = list ( xaxis = list(title = "Age_days"),
# yaxis = list(title = "eTBV"),
# zaxis = list(title = "Infusion_rate_hour"))) %>%
# hide_colorbar()9 Non-linear Dimensionality Reduction
9.1 2D t-SNE
library(Rtsne)
execTime_tSNE <- system.time(tsne_digits <- Rtsne(df_TAH_data_impute_mean, dims = 2, perplexity=7,
check_duplicates = FALSE, verbose=TRUE, max_iter = 500))## Performing PCA
## Read the 95 x 42 data matrix successfully!
## OpenMP is working. 1 threads.
## Using no_dims = 2, perplexity = 7.000000, and theta = 0.500000
## Computing input similarities...
## Building tree...
## Done in 0.00 seconds (sparsity = 0.308033)!
## Learning embedding...
## Iteration 50: error is 57.417063 (50 iterations in 0.00 seconds)
## Iteration 100: error is 57.096917 (50 iterations in 0.00 seconds)
## Iteration 150: error is 55.922942 (50 iterations in 0.00 seconds)
## Iteration 200: error is 52.954287 (50 iterations in 0.00 seconds)
## Iteration 250: error is 54.233427 (50 iterations in 0.00 seconds)
## Iteration 300: error is 1.540796 (50 iterations in 0.00 seconds)
## Iteration 350: error is 0.540761 (50 iterations in 0.00 seconds)
## Iteration 400: error is 0.235010 (50 iterations in 0.00 seconds)
## Iteration 450: error is 0.189666 (50 iterations in 0.00 seconds)
## Iteration 500: error is 0.179151 (50 iterations in 0.00 seconds)
## Fitting performed in 0.04 seconds.execTime_tSNE## user system elapsed
## 0.05 0.00 0.05# 2D t-SNE Plot
df <- data.frame(tsne_digits$Y, colors=df_TAH_data_impute_mean$TAH_occurrence)
df$col2 <- ifelse(df$colors==1, "green", "red") # map binary color to green (control) or red (patient)
plot_ly(df, x = ~X1, y = ~X2, type="scatter", mode = 'markers', color=~colors,
name=paste0("Case: ", rownames(df_TAH_data_impute_mean)),
marker=list(color = ~col2,
size=5*as.numeric(df_TAH_data_impute_mean$Group))) %>%
layout(title = "t-SNE 2D Embedding (Color=TAH; Size=Group)", xaxis = list(title = ""), yaxis = list(title = "")) %>%
hide_colorbar() %>% hide_legend()## Warning in RColorBrewer::brewer.pal(N, "Set2"): minimal value for n is 3, returning requested palette with 3 different levels
## Warning in RColorBrewer::brewer.pal(N, "Set2"): minimal value for n is 3, returning requested palette with 3 different levels9.2 3D t-SNE
execTime_tSNE <- system.time(tsne_digits3D <- Rtsne(df_TAH_data_impute_mean, dims = 3, perplexity=10,
verbose=TRUE, max_iter = 500)); execTime_tSNE## Performing PCA
## Read the 95 x 42 data matrix successfully!
## OpenMP is working. 1 threads.
## Using no_dims = 3, perplexity = 10.000000, and theta = 0.500000
## Computing input similarities...
## Building tree...
## Done in 0.00 seconds (sparsity = 0.439889)!
## Learning embedding...
## Iteration 50: error is 51.722068 (50 iterations in 0.01 seconds)
## Iteration 100: error is 53.824599 (50 iterations in 0.01 seconds)
## Iteration 150: error is 53.630193 (50 iterations in 0.01 seconds)
## Iteration 200: error is 50.524562 (50 iterations in 0.01 seconds)
## Iteration 250: error is 52.212905 (50 iterations in 0.01 seconds)
## Iteration 300: error is 1.100251 (50 iterations in 0.00 seconds)
## Iteration 350: error is 0.370685 (50 iterations in 0.00 seconds)
## Iteration 400: error is 0.175874 (50 iterations in 0.00 seconds)
## Iteration 450: error is 0.147498 (50 iterations in 0.00 seconds)
## Iteration 500: error is 0.143203 (50 iterations in 0.00 seconds)
## Fitting performed in 0.06 seconds.## user system elapsed
## 0.08 0.00 0.07df3D <- data.frame(tsne_digits3D$Y, colors=df_TAH_data_impute_mean$TAH_occurrence)
df3D$col2 <- ifelse(df3D$colors==1, "green", "red") # map binary color to green (control) or red (patient)
plot_ly(df3D, x = ~df3D[, 1], y = ~df3D[, 2], z= ~df3D[, 3], type="scatter3d", mode = 'markers',
color=~col2, name=paste0("Case: ", rownames(df_TAH_data_impute_mean)),
marker=list(color = ~col2,
size=5*as.numeric(df_TAH_data_impute_mean$Group))) %>%
layout(title = "t-SNE 3D Embedding",
scene = list(xaxis = list(title=""), yaxis=list(title=""), zaxis=list(title=""))) %>%
hide_colorbar() %>% hide_legend()df <- as.data.frame(df_TAH_data_impute_mean)9.3 UMAP
library(plotly)
# install.packages('umap')
library(umap)
#
# # Specify the input data
# umapData <- iris[ , 1:4]
# umapLabels <- iris[ , 5]
#
# # UMAP projection in 2D
# umapData.umap <- umap(umapData) # exclude the iris 3-class labels
#
# # Using Python through R-reticulate interface
# # library(reticulate)
# # iris.umap_learn <- umap(umapData, method="umap-learn")
#
# # Plot UMAP reduction in 2D
# plot(umapData.umap$layout, col=as.factor(umapLabels))
# https://rdrr.io/cran/umap/man/umap.defaults.html
custom.settings = umap.defaults
custom.settings$n_neighbors = 5
custom.settings$n_components=3
# custom.settings
df_TAH_data_impute_meanNum <- as.data.frame(lapply(df_TAH_data_impute_mean , as.numeric))
execTime_UMAP <- system.time(umap_digits3D <- umap(df_TAH_data_impute_meanNum, umap.config=custom.settings)); execTime_UMAP## user system elapsed
## 0.29 0.03 0.32cols <- palette(rainbow(10))
# 2D UMAP Plot
dfUMAP <- data.frame(umap_digits3D$layout, colors=df_TAH_data_impute_mean$TAH_occurrence)
dfUMAP$col2 <- ifelse(dfUMAP$colors==1, "green", "red") # map binary color to green (control) or red (patient)
plot_ly(dfUMAP, x = ~X1, y = ~X2, type="scatter", mode = 'markers',
color=~col2, name=paste0("Case: ", rownames(df_TAH_data_impute_mean)),
marker=list(color = ~col2, #as.factor(colors),
size=5*as.numeric(df_TAH_data_impute_mean$Group))) %>%
layout(title="UMAP (40D->2D) Embedding", xaxis = list(title = ""), yaxis = list(title = "")) %>%
hide_colorbar() %>% hide_legend()Use UMAP to predict TAH binary label.
umapData_Pred <- predict(umap_digits3D, df_TAH_data_impute_meanNum) # mind transpose of data (case * Feature)
# train.labels.colors <- colMap(train.labels)
# 2D UMAP Plot
dfUMAP <- data.frame(x=umap_digits3D$layout[, 1], y=umap_digits3D$layout[, 2],
col=as.factor(df_TAH_data_impute_mean$TAH_occurrence),
text = df_TAH_data_impute_mean$TAH_occurrence,
predX = umapData_Pred[ , 1], predY=umapData_Pred[ , 2])
dfUMAP$col2 <- ifelse(dfUMAP$col==1, "green", "red") # map binary color to green (control) or red (patient)
plot_ly(dfUMAP) %>%
add_text(x = ~x, y = ~y, mode = 'text', name="Observed", text=text,
textfont=list(color=~col2, size=15), showlegend=F) %>%
# Add training data UMAP projection scatter Digits
# add_text(text=text, textfont=list(color=~col, size=15), showlegend=F) %>%
add_markers(x=~predX, y=~predY, name ="Predicted", marker = list(color=~col2)) %>%
layout(title="UMAP Prediction: Projection of TAH using the training Data",
xaxis = list(title = ""), yaxis = list(title = ""))dfUMAP <- data.frame(x=umap_digits3D$layout[, 1], y=umap_digits3D$layout[, 2],
col=as.factor(df_TAH_data_impute_mean$TAH_occurrence), textLbl = df_TAH_data_impute_mean$TAH_occurrence,
predX = umapData_Pred[ , 1], predY=umapData_Pred[ , 2])
dfUMAP$col2 <- ifelse(dfUMAP$col==1, "green", "red") # map binary color to green (control) or red (patient)
plot_ly(dfUMAP, x = ~x, y = ~y, mode = 'text', name="Observed") %>%
# Add training data UMAP projection scatter Digits
add_text(text=~textLbl, textfont=list(color=~col2, size=15)) %>%
add_markers(x=~predX, y=~predY, name ="Predicted", marker=list(color=~col2, size=15, opacity=0.5)) %>%
layout(title="UMAP TAH Prediction: Projection of Controls (green) and Patients (red) using the training Data",
xaxis = list(title = ""), yaxis = list(title = ""))10 Regularized Linear Model Prediction
Specify the predictors (\(X\)) and the outcomes (\(Y\)).
y_TAH <- df_TAH_data_impute_meanNum$TAH_occurrence
y_Mortality <- df_TAH_data_impute_meanNum$Mortality1Day
y_Mortality <- ifelse(as.numeric(y_Mortality) > 1, "2", "1")
X <- select(df_TAH_data_impute_meanNum, -c(TAH_occurrence, Mortality1Day))
dim(X) # [1] 95 34## [1] 95 3210.1 Ridge and LASSO Regularized LM
10.1.1 Model TAH
library("glmnet")
library(doParallel)
cl <- makePSOCKcluster(6)
registerDoParallel(cl); getDoParWorkers()## [1] 6cv.ridge <- cv.glmnet(x=as.matrix(X), y=as.factor(y_TAH), type.measure="mse", alpha=0, parallel=T, family="binomial")
## alpha =1 for lasso only, alpha = 0 for ridge only, and 0<alpha<1 to blend ridge & lasso penalty !!!!
# plot(cv.ridge)
plotCV.glmnet <- function(cv.glmnet.object, name="") {
df <- as.data.frame(cbind(x=log(cv.glmnet.object$lambda), y=cv.glmnet.object$cvm,
errorBar=cv.glmnet.object$cvsd), nzero=cv.glmnet.object$nzero)
featureNum <- cv.glmnet.object$nzero
xFeature <- log(cv.glmnet.object$lambda)
yFeature <- max(cv.glmnet.object$cvm)+max(cv.glmnet.object$cvsd)
dataFeature <- data.frame(featureNum, xFeature, yFeature)
plot_ly(data = df) %>%
# add error bars for each CV-mean at log(lambda)
add_trace(x = ~x, y = ~y, type = 'scatter', mode = 'markers',
name = 'CV MSE', error_y = ~list(array = errorBar)) %>%
# add the lambda-min and lambda 1SD vertical dash lines
add_lines(data=df, x=c(log(cv.glmnet.object$lambda.min), log(cv.glmnet.object$lambda.min)),
y=c(min(cv.glmnet.object$cvm)-max(df$errorBar), max(cv.glmnet.object$cvm)+max(df$errorBar)),
showlegend=F, line=list(dash="dash"), name="lambda.min", mode = 'lines+markers') %>%
add_lines(data=df, x=c(log(cv.glmnet.object$lambda.1se), log(cv.glmnet.object$lambda.1se)),
y=c(min(cv.glmnet.object$cvm)-max(df$errorBar), max(cv.glmnet.object$cvm)+max(df$errorBar)),
showlegend=F, line=list(dash="dash"), name="lambda.1se") %>%
# Add Number of Features Annotations on Top
add_trace(dataFeature, x = ~xFeature, y = ~yFeature, type = 'scatter', name="Number of Features",
mode = 'text', text = ~featureNum, textposition = 'middle right',
textfont = list(color = '#000000', size = 9)) %>%
# Add top x-axis (non-zero features)
# add_trace(data=df, x=~c(min(cv.glmnet.object$nzero),max(cv.glmnet.object$nzero)),
# y=~c(max(y)+max(errorBar),max(y)+max(errorBar)), showlegend=F,
# name = "Non-Zero Features", yaxis = "ax", mode = "lines+markers", type = "scatter") %>%
layout(title = paste0("Cross-Validation MSE (", name, ")"),
xaxis = list(title=paste0("log(",TeX("\\lambda"),")"), side="bottom", showgrid=TRUE), # type="log"
hovermode = "x unified", legend = list(orientation='h'), # xaxis2 = ax,
yaxis = list(title = cv.glmnet.object$name, side="left", showgrid = TRUE))
}
plotCV.glmnet(cv.ridge, "Ridge")coef(cv.ridge)## 33 x 1 sparse Matrix of class "dgCMatrix"
## s1
## (Intercept) -2.710086e+00
## Group -4.345753e-03
## Sex -1.428609e-01
## Race -1.971094e-02
## Access 6.911144e-02
## Age_days -1.935071e-06
## Weight 1.157368e-04
## Pre_transfusion_K 1.336358e-01
## Cr_at_transfusion 1.099052e-01
## Cardiac_dysf -6.273937e-02
## Respiratory_dysf -1.487991e-01
## Kidney_dysf 3.969724e-01
## Liver_dysf 3.127293e-01
## Intracranial_diso 2.548654e-01
## Orthopedics 2.364015e-02
## Anemia -2.162983e-01
## Hemoglobinopathy -1.241734e-01
## Sepsis 2.467923e-01
## ECMO 1.512490e-01
## CPB 1.902543e-01
## Prematurity 6.670518e-02
## Dialysis 1.043460e-01
## RBC_Unit_type -6.145441e-02
## Age_of_RBC_unit -1.285021e-02
## Leukoreduction 3.240957e-01
## Irradiation 2.699466e-02
## Washed -1.855221e-01
## Plasma_reduction -2.575890e-02
## tV_12_weight_ml_Kg 3.492291e-03
## tV_12_eTBV 2.244901e-01
## Infusion_rate_hour_Pt_weight 1.850471e-03
## Infusion_rate_hour_eTBV 1.463833e-01
## Time_from_irradition_to_transfusion_min 3.868063e-06cv.lasso <- cv.glmnet(x=as.matrix(X), y=as.factor(y_TAH), type.measure="mse", alpha=1, parallel=T, family="binomial")
plotCV.glmnet(cv.lasso, "LASSO")coef(cv.lasso)## 33 x 1 sparse Matrix of class "dgCMatrix"
## s1
## (Intercept) -2.664830935
## Group .
## Sex .
## Race .
## Access .
## Age_days .
## Weight .
## Pre_transfusion_K 0.158588353
## Cr_at_transfusion .
## Cardiac_dysf .
## Respiratory_dysf -0.093961072
## Kidney_dysf 1.306782298
## Liver_dysf 0.142076628
## Intracranial_diso .
## Orthopedics .
## Anemia -0.390875362
## Hemoglobinopathy .
## Sepsis .
## ECMO .
## CPB .
## Prematurity .
## Dialysis .
## RBC_Unit_type .
## Age_of_RBC_unit -0.015845717
## Leukoreduction .
## Irradiation .
## Washed .
## Plasma_reduction .
## tV_12_weight_ml_Kg 0.009916597
## tV_12_eTBV 0.824338604
## Infusion_rate_hour_Pt_weight .
## Infusion_rate_hour_eTBV .
## Time_from_irradition_to_transfusion_min .df_ridge_lasso <- as.data.frame(cbind(feature=str(coef(cv.lasso)@Dimnames[[1]]),
ridge=as.matrix(coef(cv.ridge))[,1],
lasso=as.matrix(coef(cv.lasso))[,1]))## chr [1:33] "(Intercept)" "Group" "Sex" "Race" "Access" "Age_days" "Weight" ...dim(df_ridge_lasso) # [1] 35 2## [1] 33 2head(df_ridge_lasso)## ridge lasso
## (Intercept) -2.710086e+00 -2.664831
## Group -4.345753e-03 0.000000
## Sex -1.428609e-01 0.000000
## Race -1.971094e-02 0.000000
## Access 6.911144e-02 0.000000
## Age_days -1.935071e-06 0.000000variables <- rownames(df_ridge_lasso[-1,])
plot_ly(df_ridge_lasso[-1,], x = ~variables, y = ~ridge, type = 'bar', name = 'Ridge') %>%
add_trace(y = ~lasso, name = 'LASSO') %>%
layout(yaxis = list(title = 'Effect'), barmode = 'group', title="Variable Importance Plot (TAH Prediction Model)")#### Report also the SE/CI of the LASSO effect-size estimates
# install.packages("glmnetSE")
library(glmnetSE)
df_data <- cbind(y_TAH=y_TAH, X)
glmnetSE.model <- glmnetSE(data=df_data, cf.no.shrnkg = c("y_TAH"), alpha=1, method="10CVoneSE", r=250, seed = 123, family="gaussian", perf.metric="mse") ### family="binomial")
summary(glmnetSE.model)## LASSO model:
## Outcome variable: y_TAH
## Variables without shrinkage: y_TAH
##
##
## Coefficients Estimates Std.Error CI.low CI.up
## Group 0 <NA> <NA> <NA>
## Sex 0 <NA> <NA> <NA>
## Race 0 <NA> <NA> <NA>
## Access 0 <NA> <NA> <NA>
## Age_days 0 <NA> <NA> <NA>
## Weight 0 <NA> <NA> <NA>
## Pre_transfusion_K 0 <NA> <NA> <NA>
## Cr_at_transfusion 0 <NA> <NA> <NA>
## Cardiac_dysf 0 <NA> <NA> <NA>
## Respiratory_dysf 0 <NA> <NA> <NA>
## Kidney_dysf 0.161 <NA> <NA> <NA>
## Liver_dysf 0 <NA> <NA> <NA>
## Intracranial_diso 0 <NA> <NA> <NA>
## Orthopedics 0 <NA> <NA> <NA>
## Anemia 0 <NA> <NA> <NA>
## Hemoglobinopathy 0 <NA> <NA> <NA>
## Sepsis 0 <NA> <NA> <NA>
## ECMO 0 <NA> <NA> <NA>
## CPB 0 <NA> <NA> <NA>
## Prematurity 0 <NA> <NA> <NA>
## Dialysis 0 <NA> <NA> <NA>
## RBC_Unit_type 0 <NA> <NA> <NA>
## Age_of_RBC_unit 0 <NA> <NA> <NA>
## Leukoreduction 0 <NA> <NA> <NA>
## Irradiation 0 <NA> <NA> <NA>
## Washed 0 <NA> <NA> <NA>
## Plasma_reduction 0 <NA> <NA> <NA>
## tV_12_weight_ml_Kg 4e-04 <NA> <NA> <NA>
## tV_12_eTBV 0.0667 <NA> <NA> <NA>
## Infusion_rate_hour_Pt_weight 0 <NA> <NA> <NA>
## Infusion_rate_hour_eTBV 0 <NA> <NA> <NA>
## Time_from_irradition_to_transfusion_min 0 <NA> <NA> <NA>
## p.value Sig.
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
## <NA>
##
##
## Performance Metric:
## Metric Value CI.low CI.up
## mse 0.2 0.18 0.2910.1.2 Model 1-Day Mortality
# has_NA = apply(is.na(X), 1, any) #= 1 if any column in that row is NA
# X = X[!has_NA, ]
cv.ridge <- cv.glmnet(x=as.matrix(X), y=as.factor(y_Mortality),
type.measure="mse", alpha=0, parallel=T, family="binomial")
plotCV.glmnet(cv.ridge, "Ridge")coef(cv.ridge)## 33 x 1 sparse Matrix of class "dgCMatrix"
## s1
## (Intercept) -2.386467e+00
## Group -1.722271e-38
## Sex -5.917855e-38
## Race 1.344001e-38
## Access 6.362398e-38
## Age_days -1.664252e-41
## Weight -2.568711e-39
## Pre_transfusion_K 1.395261e-38
## Cr_at_transfusion 2.347954e-38
## Cardiac_dysf 1.430001e-38
## Respiratory_dysf 9.379509e-38
## Kidney_dysf 1.587302e-37
## Liver_dysf 2.387188e-37
## Intracranial_diso 3.192094e-37
## Orthopedics 1.234568e-37
## Anemia -4.432624e-38
## Hemoglobinopathy 1.782531e-38
## Sepsis 2.779735e-37
## ECMO 1.115047e-37
## CPB 1.234568e-37
## Prematurity 1.114431e-37
## Dialysis 2.197298e-37
## RBC_Unit_type -3.290092e-38
## Age_of_RBC_unit -3.197691e-39
## Leukoreduction 8.783487e-38
## Irradiation 1.830673e-38
## Washed 2.598448e-37
## Plasma_reduction 4.290214e-37
## tV_12_weight_ml_Kg 1.806057e-39
## tV_12_eTBV 1.535367e-37
## Infusion_rate_hour_Pt_weight 1.469213e-39
## Infusion_rate_hour_eTBV 1.681580e-37
## Time_from_irradition_to_transfusion_min -5.857657e-42cv.lasso <- cv.glmnet(x=as.matrix(X), y=as.factor(y_Mortality),
type.measure="mse", alpha=1, parallel=T, family="binomial")
plotCV.glmnet(cv.lasso, "LASSO")coef(cv.lasso)## 33 x 1 sparse Matrix of class "dgCMatrix"
## s1
## (Intercept) -2.386467
## Group .
## Sex .
## Race .
## Access .
## Age_days .
## Weight .
## Pre_transfusion_K .
## Cr_at_transfusion .
## Cardiac_dysf .
## Respiratory_dysf .
## Kidney_dysf .
## Liver_dysf .
## Intracranial_diso .
## Orthopedics .
## Anemia .
## Hemoglobinopathy .
## Sepsis .
## ECMO .
## CPB .
## Prematurity .
## Dialysis .
## RBC_Unit_type .
## Age_of_RBC_unit .
## Leukoreduction .
## Irradiation .
## Washed .
## Plasma_reduction .
## tV_12_weight_ml_Kg .
## tV_12_eTBV .
## Infusion_rate_hour_Pt_weight .
## Infusion_rate_hour_eTBV .
## Time_from_irradition_to_transfusion_min .df_ridge_lasso <- as.data.frame(cbind(feature=str(coef(cv.lasso)@Dimnames[[1]]),
ridge=as.matrix(coef(cv.ridge))[,1],
lasso=as.matrix(coef(cv.lasso))[,1]))## chr [1:33] "(Intercept)" "Group" "Sex" "Race" "Access" "Age_days" "Weight" ...dim(df_ridge_lasso) # [1] 35 2## [1] 33 2head(df_ridge_lasso)## ridge lasso
## (Intercept) -2.386467e+00 -2.386467
## Group -1.722271e-38 0.000000
## Sex -5.917855e-38 0.000000
## Race 1.344001e-38 0.000000
## Access 6.362398e-38 0.000000
## Age_days -1.664252e-41 0.000000variables <- rownames(df_ridge_lasso[-1,])
plot_ly(df_ridge_lasso[-1,], x = ~variables, y = ~ridge, type = 'bar', name = 'Ridge') %>%
add_trace(y = ~lasso, name = 'LASSO') %>%
layout(yaxis = list(title = 'Effect'), barmode = 'group', title="Variable Importance Plot (1-Day Mortality Prediction Model)")11 Probability Distribution Models for each TAH and Control Cohorts
# df_TAH_data_TAH_Controls<- filter(df_TAH_data_impute_mean, TAH_Occurrence==0)
# df_TAH_data_TAH_Patients
library(tidyr)
# df_TAH_data_TAH_Controls_long <- gather(df_TAH_data_TAH_Controls, feature, value,
# Group:Irradiation, factor_key=TRUE)
# pl1 <- plot_ly(df_TAH_data_TAH_Controls_long, y = ~log(value), color = ~feature, type = "box", showlegend=F) %>%
# layout(title="Feature Distributions: Controls (TAH=1)")
#
# df_TAH_data_TAH_Patients_long <- gather(df_TAH_data_TAH_Patients, feature, value,
# Group:Irradiation, factor_key=TRUE)
# pl2 <- plot_ly(df_TAH_data_TAH_Patients_long, y = ~log(value), color = ~feature, type = "box", showlegend=F) %>%
# layout(title="Feature Distributions: Patients (TAH=2)")
#
# subplot(pl1, pl2, nrows=2, shareX = TRUE) %>% layout(title="Feature Distributions: Controls (Top) vs. Patients (Bottom)")
#
# subplot(pl1, pl2, nrows=2, shareX = TRUE) %>% layout(title="Feature Distributions: Controls vs. Patients",
# updatemenus = list(list(
# active = 0,
# buttons= list(
# list(label = 'linear',
# method = 'update',
# args = list(list(visible = c(T,F)), list(yaxis = list(type = 'linear')))),
# list(label = 'log',
# method = 'update',
# args = list(list(visible = c(F,T)), list(yaxis = list(type = 'log'))))))))
CNTR_data <- as.data.frame(sapply(df_TAH_data_TAH_Controls, as.numeric)) %>%
gather(variable, value, -c("Age_days", "Time_from_irradition_to_transfusion_min"))
# %>% separate(variable, "TAH_occurrence", sep = "\\_")
pl_Controls <- plot_ly(CNTR_data, x=~variable, y=~value, color=~variable, type="box", showlegend=F) %>%
layout(title="Feature Distributions: Controls (non-TAH)")
PTNT_data <- as.data.frame(sapply(df_TAH_data_TAH_Patients, as.numeric)) %>%
gather(variable, value, -c("Age_days", "Time_from_irradition_to_transfusion_min"))
# %>% separate(variable, "TAH_occurrence", sep = "\\_")
pl_Patients <- plot_ly(PTNT_data, x=~variable, y=~value, color=~variable, type="box", showlegend=F) %>%
layout(title="Feature Distributions: Patients (TAH)")
subplot(pl_Controls, pl_Patients, nrows=2, shareX = TRUE) %>%
layout(title="Feature Distributions: Controls (Top) vs. Patients (Bottom)")CNTR_data$Cohort <- "Controls"
PTNT_data$Cohort <- "Patients"
df_CNTR_Pat <- rbind(CNTR_data, PTNT_data)
plot_ly(df_CNTR_Pat, x=~variable, y=~value, color=~Cohort, type="box", showlegend=F) %>%
layout(title="Feature Distributions: Controls (non-TAH) vs. Patients (TAH)", boxmode = "group",
xaxis = list(title='Features'), yaxis = list(title='values'))Other SOCR publicaitons, webapps, research projects, Ebooks, and instructional resources are available online.