Implementation of machine learning algorithms for Parkinsonian gait data

In this study, we used the Physionet gait database and extracted gait features such as step/stride regularities and symmetries to build a classifier for Parkinson’s disease (PD) subjects and healthy controls. We also improved the number of features using the mean and standard deviation of step times during their usual, self-selected pace for approximately 2 minutes on level ground. Extracted features were used in three different machine learning algorithms. PD is a neurodegenerative disorder caused by the neurodegeneration of regions of the basal ganglia. Gait abnormality is one of the main symptoms of PD. Motor symptoms in Parkinson’s disease cause a lack of control over movements and difficulty initiating muscle movements such as shuffling steps, quicker strides, or moving slower than expected for the corresponding age. The proposed approach can be used for the diagnosis of PD that can be automated or performed remotely.