Vulnerability

Machine learning analysis for predicting days with missing exercise behavior using clinical and wearable-derived data.

Repository layout

• 1_Raw/: Source data files (demographics, exercise logs, glucose, insulin).
• 2_Aggregated/: Processed day-level and minute-level datasets.
• 3_Code and Results/: Analysis scripts and manuscript/result artifacts.

Main analysis scripts

Located in 3_Code and Results/:

• datapreprocessing.py
• modelexploration.py
• ManuscriptBaselineModelsCovariateOnly.py
• ManuscriptCovariateWithPreviousDay.py
• ManuscriptWeekLookBackFeatures.py
• ManuscriptTSFreshFull.py

Quick start

1. Use Python 3.9+.
2. Install required packages used by the scripts (for example: pandas, numpy, scikit-learn, and plotting libraries used in your environment).
3. Run preprocessing first, then model scripts from 3_Code and Results/.

Example:

cd "3_Code and Results"
python datapreprocessing.py
python modelexploration.py

Notes

• Some data files are large (including spreadsheets and PDFs).
• The repository contains study data and analysis outputs; handle and share according to your data governance requirements.