Characterization and detection of Parkinson’s Disease, A data driven approach

Abstract

This study aims to examine diagnostic data of patients suffering from the Parkinson’s disease to identify characteristics that are distinctive in the presence of Parkinson’s. The study discovered numerous new correlations such as male Parkinson’s subjects being heavier than their non-Parkinson’s counterparts but indicated no such trend in females. The study also validated previously existing theories including the morphological alterations of the Caudate and Putamen nuclei in the brain as a result of Parkinson’s. Independent datasets obtained from the Parkinson’s Progression Markers Initiative dataset are explored in this study. Furthermore, datasets are created by combining the available data and standard machine learning models are employed to detect the presence of the Parkinson’s disease. A maximum accuracy of 96% was achieved by the Decision Tree model on a merged dataset consisting of medical history, socio-economic background and mobility data.