Predicting Impulse Control Disorder in Parkinson’s Disease Using Machine Learning

Abstract

Parkinson's disease is a progressive neurodegenerative disorder that primarily affects movement, balance, and motor coordination due to the gradual loss of dopamine-producing neurons. Early identification of the disease is essential for timely medical intervention and improved patient management. This study presents a deep learning-based framework for the automatic prediction of Parkinson's disease using brain Magnetic Resonance Imaging (MRI) images. Three deep learning models, namely a Custom Convolutional Neural Network (CNN), a fine-tuned ResNet50, and a fine-tuned EfficientNetB0, were developed and evaluated using the same MRI dataset. The input images were resized, preprocessed, and augmented to improve model robustness and generalization. Model performance was assessed using accuracy, precision, recall, F1-score, classification report, and confusion matrix. Among the evaluated models, the fine-tuned ResNet50 achieved the highest classification accuracy of 92.66%, outperforming the Custom CNN and EfficientNetB0 models. To demonstrate its practical applicability, the best-performing model was integrated into a Flask-based web application that enables users to upload MRI images and receive automated prediction results with confidence scores. The application also stores prediction records in an SQLite database for future reference. The proposed framework provides a reliable computer-aided screening tool for Parkinson's disease and has the potential to support healthcare professionals by improving diagnostic efficiency and consistency while complementing clinical decision-making.