Assessment of Parkinson's Disease Progression by Feature Relevance Analysis and Regression Techniques Using Machine Learning Algorithms

Abstract

Remote patient tracking has been gaining increased attention due to its low-cost non-invasive methods. Unified Parkinson's Disease Rating Scale (UPDRS) is used often to track Parkinson's Disease (PD) symptoms which requires the patient's visit to the clinic and time consuming medical tests that may not be feasible for most of the elderly PD patients. One of the major concerns to predict the PD in early stages is that PD symptoms overlap with the symptoms of other diseases such as Multiple Sclerosis, Alzheimer's disease. Moreover, most of the current methods used for tracking PD rely on expert clinical raters, from which PD symptoms assessment may be difficult due to inter-individual variability. Predicting relevant features using machine learning algorithms is helpful in providing the scientific decision-making classification rules necessary to assess the disease progression in early stages.