Designing a Machine Learning Model to Predict Parkinson’s Disease from Voice Recordings

Abstract

Parkinson’s disease is one of the most prevalent neurodegenerative disease among the older population more than 60 years, it directly affects the brain cells, and it keeps deteriorating the health of the patient without being cured. It has been found in our research work that there is no cure to Parkinson’s disease, so it must be detected early to prevent the fatal rate. Machine learning based approach has been used by many researchers to get a breakthrough for the early prediction of Parkinson’s disease. PD cannot be diagnosed through traditional diagnostic methods, so there must be some other method for diagnosing it through cheaper and faster diagnostic tool. This paper proposed a predictive model through which PD can be diagnosed attaining highest accuracy and precision employing a dataset which is composed of voice recordings of the healthy subjects and PD patients. The results obtained from the experiment showed that the Two-Class SVM is the best model for the available dataset with an accuracy of 99%.