Early detection of subtle motor function changes can play a crucial role in slowing the progression of Parkinson’s disease.
However, these changes often go unnoticed during clinical assessments.
A recent study by University of Florida researcher Diego L. Guarín, Ph.D., offers a promising solution by using artificial intelligence (AI) to detect these subtle signs through video recordings.
Guarín, an assistant professor at the UF College of Health & Human Performance, published his findings in npj Parkinson’s Disease. His research focused on analyzing videos of finger-tapping movements from 66 individuals, including healthy controls, people with idiopathic REM sleep behavior disorder (iRBD), and individuals with early Parkinson’s disease.
iRBD is characterized by people physically acting out their dreams during sleep and is a known precursor to Parkinson’s and related brain disorders, with over 80% of iRBD patients developing such conditions. Because of this, iRBD is a key group for studying early motor changes.
Participants in the study did not show visible signs of Parkinsonism in their finger-tapping videos. According to Guarín, a trained clinician reviewing the videos concluded that all participants appeared healthy. However, AI-driven video analysis told a different story.
Using VisionMD, an open-source machine learning software developed by his team, Guarín found that individuals who appeared healthy to human observers actually displayed smaller and slower finger-tapping movements. This reveals that the software is capable of identifying motor impairments that clinicians may miss.
“Video analysis is allowing us to see movement alterations that the eyes of the clinician cannot see,” said Guarín. “Early identification of these movement alterations is critical for disease management.”
The software also detected the ‘sequence effect’ in both iRBD and Parkinson’s patients—a progressive decline in movement speed and amplitude during repetitive actions. This effect could serve as an early marker of neurological disorders, though its underlying mechanisms are not fully understood.
Guarín’s findings suggest that AI and standard video recordings, even those taken with smartphones or webcams, could become powerful, accessible tools for early diagnosis and screening of Parkinson’s and related brain disorders.
By identifying movement abnormalities invisible to the human eye, this AI-driven method could potentially enable earlier intervention and better disease management.
The study is published in npj Parkinson’s Disease.
Copyright © 2025 Knowridge Science Report. All rights reserved.
