An insight on recent advancements and future perspectives in detection techniques of Parkinson's disease

Parkinson's disease has no specific test for diagnosis, which is usually detected through a review of signs and symptoms like tremors, stiffness, and slowness of movement. Over the past 10-15 years various methods were proposed that use motor signs and voice data for the accurate detection of Parkinson's disease (PD). Cutting-edge technologies like machine learning, deep learning, image processing, and several signal Processing mechanisms were employed in the efficient diagnosis of PD. In view of this, we present a structured review of such techniques. In the ongoing research over the past few years, various datasets were generated to assess PD with the salient parameters of the voice signals acquired from healthy people and PD patients. Also, tremor and gait data were collected by recording the movements of PD patients from sensing units such as accelerometers, gyroscopes, and inertial measurement unit sensors. An overall analysis has been done on the various techniques with respect to the voice dataset and motion sensor data. A detailed analysis suggests that machine and deep learning approaches are optimistic in developing clinical applications.