Comparative Analysis of Machine Learning Algorithms for Early Prediction of Parkinson’s Disorder based on Voice Features

Abstract

Parkinson Disorder (PD) is a neurological disorder which is progressive in nature and has no cure. Early diagnosis of PD plays a key role in delaying the progression of the disorder. Dysphonia is the most prominent early symptom which is exhibited by approximately 90% of PD patients. Voice features based early diagnosis with the integration of Artificial Intelligence plays a prominent role in providing accurate, non-invasive, and robust predictions to PD patients. This paper focuses on providing comparative and experimental analysis of Machine Learning (ML) algorithms for the prediction of PD based on the voice features dataset retrieved from the UCI repository. This paper presents the results from the four sampling experiments conducted with different traditional ML algorithms for the retrieved voice dataset. The results of this study make it evident that Naïve Bayes provides a highest accuracy of 89% when compared to other ML algorithms. This study helps in identifying the best ML algorithm among the traditional ML algorithms for PD prediction based on voice features dataset.