Predicting Parkinson’s Disease using Machine Learning

Parkinson’s disease (PD) is a progressive neurodegenerative disease which affects motor functions of the body, including tremors, rigidity and bradykinesia. Early detection of PD allows effective interventions to be initiated in time to improve the quality of life of patients, but is not easy due to disease symptoms being similar to other disorders. In this work, we propose a machine learning based diagnostic system which attempts to leverage voice measurements along with clinical characteristics in order to improve early diagnosis of PD. We used several machine learning models such as Support Vector Machine (SVM), Random Forest, K-Nearest Neighbors (KNN) and Logistic Regression. The combination of genetic risk score along with clinical assessment resulted in better performance, penalized Logistic Regression and XGBoost. As compared to traditional approaches, a noninvasive, telemedicine supported and scaled out diagnostic solution is proposed.