Overview

• A peer-reviewed npj Parkinson’s Disease paper reports analysis of 66 finger-tapping videos from healthy participants, people with idiopathic REM sleep behavior disorder, and individuals with early Parkinson’s disease.
• Expert reviewers judged the recordings as normal, yet the AI pipeline quantified smaller and slower movements consistent with bradykinesia.
• The system detected the sequence effect, a progressive decline in movement amplitude or speed, in both the iRBD and Parkinson’s groups.
• VisionMD, released as open-source and built on MediaPipe hand tracking, extracted amplitude, speed, and sequence-effect metrics from standard video.
• Researchers propose low-cost screening using smartphone or webcam recordings for high-risk groups such as iRBD, with broader validation needed before routine clinical use.