Explainable AI Based Parkinson’s Disease Detection From Voice Data Using Machine Learning

Development of a robust preprocessing and acoustic
feature extraction pipeline for Parkinson’s voice analysis.
• Comprehensive comparative evaluation of seven machine
learning models and one CNN-based deep learning model
for Parkinson’s Disease detection.
• Identification of Random Forest as the best-performing
classifier with an accuracy of 84.93% and ROC-AUC of
0.8906.
• Integration of SHAP and LIME to enhance model transparency and support clinical interpretation