Gait analysis in Parkinson's disease

Abstract

Recent advancement of technology has made it possible to measure the gait recordings of patients with Parkinson’s Disease (PD). Gait is the walking pattern of a person and gait disorders are commonly observed and known to exist in PD patients. In addition, with machine learning techniques improving at a rapid rate, researchers are therefore looking into using machine learning techniques to perform gait analysis as an alternative way to diagnose patients with PD apart from the current diagnosis method which is through the clinician’s recognition of motor symptoms. The impact of this new diagnosis method is potentially significant as the diagnosis will now not be based solely on the clinician’s judgement so it will be less susceptible to human error. In addition, the symptoms will not have to be very severe in order for PD to be detected, and this could result in early and accurate detection of PD which can be very helpful for potential patients. This project will therefore look at the possibility of using some of these gait features that can be extracted from gait recordings of healthy patients and PD patients, as well as explore different feature selection techniques, classification models and performance metrics to see the if using machine learning techniques on gait features could result in accurate classification and hence diagnosis of patients with PD.