Innovative Approaches to Enhance Accuracy of Parkinson’s Disease

Abstract

Parkinson’s disease (PD) is a slowly worsening neurological condition that affects a person’s ability to control movement. Detecting PD at an early stage can make treatment more effective and help patients manage their symptoms better. In this work, we develop a real-time system that analyzes a person’s voice to predict the likelihood of Parkinson’s disease using machine learning. The system is trained on the UCI Parkinson’s dataset, which contains several important vocal features such as jitter, shimmer, and harmonic-to-noise ratio. Multiple machine learning models including Support Vector Machine (SVM), Random Forest (RF), Decision Tree (DT), and Logistic Regression (LR) are trained and evaluated using accuracy, F1-score, ROC, and AUC metrics. The model with the best performance is deployed in a user-friendly Stream lit web application that allows real-time predictions from simple voice recordings. The proposed system is non-invasive, affordable, and easy to use, making it a promising tool for preliminary screening and early identification of Parkinson’s disease.