Abstract
Parkinson's disease is a prevalent neurodegenerative disorder, where early diagnosis is essential for slowing disease progression and optimizing treatment strategies. The latest developments in artificial intelligence (AI) have introduced new opportunities for early detection. Studies have demonstrated that before obvious motor symptoms appear, PD patients exhibit a range of subtle but quantifiable motor abnormalities. This article provides an overview of AI-driven early detection approaches based on various motor symptoms of PD, including eye movement, facial expression, speech, handwriting, finger tapping, and gait. Specifically, we summarized the characteristic manifestations of these motor symptoms, analyzed the features of the data currently collected for AI-assisted diagnosis, collected the publicly available datasets, evaluated the performance of existing diagnostic models, and discussed their limitations. By scrutinizing the existing research methodologies, this review summarizes the application progress of motor symptom-based AI technology in the early detection of PD, explores the key challenges from experimental techniques to clinical translation applications, and proposes future research directions to promote the clinical practice of AI technology in PD diagnosis.