Clinically Meaningful Gait Feature Selection for Classification of Parkinson’s Disease

The diagnosis of Parkinson’s disease (PD) is challenging due to the overlap of its clinical signs with those of other neurodegenerative diseases. Gait analysis is a promising tool that can capture early and evolving motor abnormalities in PD. This study focuses on the use of inertial measurement units (IMUs) to identify gait characteristics for the classification of PD, particularly targeting bradykinesia. By using simple metrics (mean, standard deviation and range of motion of joint angles and acceleration), our exploratory analysis aims to provide a straightforward means of quantifying movement abnormalities. We used random forest (RF) and recursive feature elimination (RFE) to select robust features, highlighting the superiority of RF in distinguishing PD cases with an accuracy of 0.75 with only three features. This approach also distinguished the most informative sensors (hip and thigh) with potential to improve the diagnosis and monitoring of PD.