The Movement Disorder Society-Unified Parkinson’s Disease Rating Scale (MDS-UPDRS) is a widely accepted tool for assessing Parkinson’s disease (PD) severity and staging. However, manual assessment is subjective and may lack consistency. Therefore, this study proposes developing a machine learning (ML)-based model for PD staging using MDS-UPDRS features to categorize patients into three stages: normal, mild, and severe. To achieve this, five ML classification algorithms– Decision Tree, Subspace Discriminant Ensemble Classifiers, Support Vector Machine (SVM), Kernel naive Bayes, and K-Nearest Neighbors (KNN) — were applied to evaluate the performance of MDS-UPDRS Parts I, II, III, and IV in distinguishing between PD and control subjects. The SVM model demonstrated a high classification accuracy of 98.47% with the training dataset while achieving an accuracy of 100% with the test dataset. The results suggest that ML techniques can serve as reliable and efficient tools for PD classification, reducing diagnostic uncertainty and aiding clinical decision-making. These models have the potential to be integrated into healthcare systems as intelligent diagnostic tools for early and automated PD prediction, thereby facilitating timely interventions and improving patient management.
Satyam et al. (Tue,) studied this question.