Overview

• University of Florida researchers analyzed finger-tapping videos from 66 participants spanning healthy controls, iRBD, and early Parkinson’s disease.
• Videos were processed with VisionMD, an open-source system that derives movement amplitude and speed by tracking fingertip distance using Google’s MediaPipe.
• Even when an expert clinician deemed the recordings normal, the AI detected smaller and slower movements in people with Parkinson’s disease.
• The analysis identified the sequence effect in both iRBD and Parkinson’s groups, pointing to a potential early motor biomarker.
• The findings were published in npj Parkinson’s Disease, and the team notes broader, prospective validation is needed before clinical use, with smartphone or webcam recordings proposed for future screening.