Multi-faceted approach for Parkinson’s disease detection and monitoring

Abstract

Parkinson's disease (PD) is a paralyzing neurodegenerative disorder which affects millions of people around the world. To improve diagnosis, monitoring, and treatment, a novel and comprehensive approach is needed. In this project, we develop a machine learning (ML) and deep learning (DL) based mobile application—consisting with four functional components—to positively impact diagnosis, monitoring, and treatment processes for PD. The first component includes a wearable device that can continuously track the symptoms and identify the stage of PD. The second component includes a real-time facial detection system to pick up hypomimia symptoms through facial expressions. The third component is implemented to perform voice and speech analysis to find subtle changes in speech patterns linked to PD, and the fourth component can detect common motor symptoms of PD using the gyroscope of a smartphone. The trained ML and DL models that are associated with the four functional components show 96%, 92%, 97%, 92% testing accuracies for the first, second, third, and fourth components respectively suggesting that the developed system is promising in facilitating symptom detection and personalized care for PD patients.