Center for Complexity and Self-Management of Chronic Disease
(CSCD): Core 2: Methods and Analytics Progress (2019-2020)
In 2019-2020, CSCD students, trainees and investigators continued to
make advances building protocols for data analytics. Several examples of resources are included below.
Longitudinal Data Analysis Protocol
This development includes
implementing, validating and sharing a complex end-to-end analytical protocol for
analyzing longitudinal data, e.g., fMRI. A summary of this protocol is included below and the
complete protocol documentation is available here.
Figure: High-Dimensional Longitudinal Data Analytics.
- Package Loading and Data Manipulation
- Time-series graphs - Interactive time-series visualization
- 3 Kime-series/Kime-surfaces (spacekime analytics protocol) -
Pseudo-code;
Function main steps illustration: Kime-surfaces / kime-series construction
Function result
- Interactive plotly Example - Plotly method: interactive way;
Forecasting with time series
- Motor area detection - fMRI data simulation;
Stimulus detection
- Motor area visualization - Visualization and comparison of p-value;
Comparison of performance of different methods on the same fMRI data
- Three-phase ROI Analysis -
Phase1: Detect Potential Activated ROI;
Phase2: ROI-Based Tensor-on-Tensor Regression;
Phase3: FDR Correction and Spatial Clustering;
3D visualization based on the activated areas by regions
Pressure Injury Modeling and Prediction Webapp
We developed an
AI/ML webapp forecasting the probability of hospitalized patients to
develop pressure injuries. This app allows clinicians to identify and preemptively target
patients at high risk for developing pressure injuries.
Figure: Pressure Injury Modeling and Prediction Webapp.
Curricular Developments
