Toward Automation of Motor Assessments in Stroke: Proof-of-Concept Model to Interpret Balance and Gait in Patients With Stroke.
Yee Mah, Vassilios Tahtis, Abigail Beddard and 3 others
PMID 42769357WHAT IT FOUND
A video model recognized sit, stand, and turn actions from balance and gait assessment recordings, but performed poorly on feet together, nudge, and walking.
It is not ready to replace clinician observation.
Key findings
01The model labeled POMA video clips with mean action classification accuracy of 78.9%, and it was best for sit, stand, and turn while standing feet together and walking were the most challenging.
02The visual mapping of movement clips showed that healthy participants grouped more tightly during sit-to-stand, while participants with stroke were more spread out.
STILL TO COME
How it was doneWhat they foundWhat it means for PTs
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 stroke group was small compared with the healthy group, and all data came from one hospital. The controls were not age matched, so differences could reflect age rather than stroke. The actions were not equally represented, and some tasks produced many clips while others produced few. The model used fixed one-second clips, which did not always match the true start and end of each task, especially for brief actions. The pose estimation system has joint localization error, so it is not suitable for fine biomechanical analysis. The study did not test whether automated assessment changes clinical decisions or patient outcomes.
Declared interests
One author holds shares in AMD. Another holds stocks in Hologen and Elaitra and has received UKRI scholarship and EU MSCA grant funding. Another is affiliated with Hologen and has received Wellcome Trust and UCLH NIHR Biomedical Research Centre funding. The remaining authors declared nothing.
The easy way to misread this
Do not conclude that automated POMA assessment is ready for clinical use. The study showed proof of concept only, and the model performed poorly on several balance and gait actions.