We aim to identify the features that are most important for explaining how biologically older a patient is compared to their chronological age, referred to as age acceleration. First, we train a simple model to predict biological age. We then train a more complex model to predict age acceleration itself. Finally, we apply SHAP analysis to the complex model to determine which features contribute most to patients exhibiting greater age acceleration than would be predicted by the simple age-based model.
To run this execute
cd py
python train.py
python shap_analysis.py