Automated video analysis for early detection of bradykinesia in Parkinson's disease.
Diego L Guarín, Jackson G Wolfe, Sofia Kane and 2 others
PMID 41664165WHAT IT FOUND
An AI system analyzing finger-tapping videos accurately distinguished people with Parkinson's disease from healthy controls, even when clinicians rated the movement as normal or only slightly impaired.
This suggests subtle motor signs invisible to the eye can be detected objectively.
Key findings
01The feature-based model achieved a ROC-AUC of 0.94 and balanced accuracy of 0.87, successfully distinguishing patients from controls.
02Patients showed significantly reduced movement amplitude and speed, and increased variability compared to controls, even with mild clinical ratings.
03The system detected Parkinson's disease in individuals clinically rated as having normal or only slight motor dysfunction on the finger-tapping task.
STILL TO COME
How it was doneWhat they foundWhat it means for PTsWhat it means for OTs
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The study was retrospective and moderate in size, with no separate test set, which may lead to optimistic performance estimates. Clinician ratings used as the ground truth were subjective and subject to inter-rater variability. Important confounders like disease duration and hand laterality were not available for all videos and were not controlled for. The study focused on early-stage patients with minimal motor signs, so results may not generalize to more advanced disease.
Declared interests
Funded by the Norman Fixel Institute for Neurological Diseases, University of Florida Intelligent Clinical Care Center, Deutsche Forschungsgemeinschaft, and Interdisziplinäres Zentrum für Klinische Forschung, Universitätsklinikum Würzburg.
The easy way to misread this
Do not interpret the high accuracy as evidence that this tool is ready for clinical diagnosis or screening. The study was retrospective, lacked a separate external test set, and relied on subjective clinical ratings as the ground truth, which limits the certainty of its real-world diagnostic utility.
Summarised by AI from the full paper, without a clinician reviewing it. Check it against the source before it changes what you do. Read it on PubMed →