Early Diagnosis of Parkinson’s Disease via Pro-Saccadic Eye Movement Analysis: Multimodal Intermediate Fusion Framework

Abstract

Early detection and timely treatment are essential for improving patient outcomes, but the lack of reliable biomarkers impedes early diagnosis of Parkinson’s Disease (PD). Consequently, eye movement abnormalities, knownas early symptoms of PD, are gaining attention as crucial clues for early diagnosis. This study proposes a novel multimodal intermediate fusion framework for the early diagnosis of PD using eye-tracking data. The proposed framework improves the performance of classifying abnormal eye movement patterns in PD by integrating local features from time-series data and global features from encoded time-series images. Focusing on pro-saccade eye movements, this framework captures significant abnormalities like reduced peak saccadic velocity and multi-step saccades frequently observed in PD. The experimental results show a precision of 82% and a recall of 96% for PD, which demonstrates the effectiveness of the framework in minimizing missed diagnoses during early detection. In addition, this study highlights the potential of eye-tracking data as a biomarker for the early diagnosis of PD and predicts the advanced application of integrating wearable smart glasses for daily monitoring of neurodegenerative diseases.