Garv1221/parkinsons_research ? reverse-engineered prompt

Build me a Python app for Parkinson’s research that can predict continuous motor severity from two kinds of data, voice features and gait force time series.

I want an interactive dashboard and a simple terminal menu. The app should let me train and evaluate a voice only model, a gait only model, and then combine their predictions in a late fusion step instead of mixing the raw datasets together. Please keep it scientifically careful, with subject separated train and test splits, clear validation results, and a comparison view that shows how the models perform side by side.

It should also handle the gait dataset extraction from the provided clinical files, save trained models and summary results, and generate a chart comparing the approaches. If you need to look up current docs for any library, feel free to do that. Make the experience easy to run from one main entry point, with both a web app and a command line option.