Projects on Repeat
Favorite builds, experiments, and ideas I keep coming back to.
NeuroTrace
Parkinson's Risk Classifier
Data ScienceComputer VisionFull-Stack ML
Liner Notes
An explainable machine learning web app that screens for Parkinson's-related risk patterns using both clinical questionnaire data and spiral drawing analysis. The app combines two model outputs into a clear risk interpretation for Parkinson's disease.
Tools Used
- Python
- Streamlit
- Scikit-learn
- PyTorch
- MobileNetV2
- OpenCV
- SHAP
- Pandas
- Plotly
- Gemini API
Technical Highlights
Multi-modal ML pipeline combining a Random Forest clinical model, MobileNetV2 spiral-drawing CNN, SHAP explainability, and late-fusion risk logic.
Live Preview
Screen early. Understand the risk.