Automated Detection and Recommendation System for Parkinson’s Disease Using Machine Learning

Parkinson’s Disease (PD) is a chronic neurological condition that significantly affects speech and motor control. Early diagnosis plays a vital role in symptom management and slowing disease progression. This project presents an automated machine learning-based system for early detection and severity classification of Parkinson’s Disease using voice signal features. Key voice measurements such as jitter, shimmer, and harmonic-to-noise ratio are extracted from biomedical voice data to train multiple classifiers. An ensemble model combining XGBoost, K-Nearest Neighbors, Decision Tree, and Gaussian Naive Bayes achieves high diagnostic accuracy. The system also incorporates severity prediction (Mild, Moderate, Severe) based on probability scores and provides personalized recommendations related to exercise, diet, and therapy. The best-performing model is deployed in a Flask-based web application, enabling users to input voice features and receive real-time feedback. This non-invasive, cost- effective, and user-friendly system aids in clinical diagnosis, enhances early detection, and empowers patients with actionable health insights.