Code for MWKNeuroReha (in preparation)

This repository hosts the analysis notebooks for the MWKNeuroReha project.

Most notebooks reproduce figures and model comparisons from the study; due to patient privacy, only the proof-of-concept notebook is runnable without access to the protected dataset.

.
├── 00-gpclust-proof-of-concept.ipynb
├── 01-figure-3-longitudinal-gp-clustering.ipynb
├── 02-train-test-split-and-imputation.ipynb
├── 03-average-models.ipynb
├── 04-ensemble-of-models.ipynb
├── 05-reduced-models.ipynb
├── 06-reduced-model-performance.ipynb
├── 07-figure-4-full-model-weights.ipynb
├── 08-figure-5-reduced-model-weights-top7.ipynb
├── GPclust

Notebooks 01–08 are for display only and expect non-public patient data. Only 00 runs out-of-the-box.

We bundled a slightly modified version of GPclust to ensure compatibility with newer Python versions. Install it in editable mode:

# with uv
uv sync 
uv pip install -e GPclust

# or without
pip install numpy scipy matplotlib pandas seaborn gpy
pip install -e GPclust