Dysphonic Voice Pattern Based Parkinson Disease Detection Using Machine Learning Models

Parkinson's Disease stands as a widespread cerebral degenerative disorder impacting millions of people across the world. Timely and precise diagnosis of PD is of paramount importance, enabling early interventions and enhancing overall patient care. The presented work focuses on non-invasive detection of PD through voice analysis, targeting early intervention. Dysphonic voice patterns, indicative of PD progression, are analyzed using advanced signal processing and machine learning. Methodology involves collecting voice recordings from PD-affected individuals and healthy controls, with rigorous analysis of acoustic features. This approach minimizes patient discomfort and reduces healthcare costs, potentially revolutionizing PD care and enhancing patient quality of life. This study explored the efficacy of 4 machine learning models - Decision Trees, k-Nearest Neighbors, Support Vector Machine, and Naive Bayes- in Parkinson's Disease (PD) detection. k-Nearest Neighbors classifier demonstrates much better performance compared to the other three, achieving an accuracy of 96.61%.