AI-Driven Multi-Class Diagnosis of Parkinson’s Disease: Enhancing Accuracy and Differentiation from Movement Disorders

Parkinson’s disease is a progressive neurological disorder that is frequently misdiagnosed due to its symptomatic similarities with other movement-related conditions. Ensuring precise and early diagnosis is essential for effective treatment; however, current diagnostic approaches are often subjective and prone to errors. To overcome these limitations, this study introduces a multi-class classification framework designed to differentiate Parkinson’s disease from related disorders, including Essential Tremor, Atypical Parkinsonism, Secondary Parkinsonism, Multiple Sclerosis, and Healthy Controls. The classification process is structured into three tasks: distinguishing Parkinson’s disease from Healthy Controls, differentiating it from other movement disorders, and performing a multi-class classification that includes Parkinson’s disease, Healthy Controls, and additional disorders. Experimental evaluations indicate that XGBoost achieved an F1-score of 93% for Parkinson’s disease versus Healthy Controls, SVM attained 85% for Parkinson’s disease against other movement disorders, and Logistic Regression yielded 68% accuracy in the multi-class classification. These results highlight the potential of machine learning in providing accurate, scalable, and objective diagnostic solutions for Parkinson’s disease.