Project 02 / 06

Parkinson's

Severity prediction from speech

Predicts Parkinson's severity scores from speech features. Combining clustering-based feature selection with an Extra Trees regressor gave state-of-the-art R² on both targets. Published in Scientific Reports (2026).

View code

Results

What it achieved

R² motor UPDRS: 0.999
R² total UPDRS: 0.997

Built with

Scikit-Learn
Clustering
Extra Trees
Feature Selection