The feasibility of using deep learning technologies to preliminarily identify patients with advanced knee osteoarthritis via smartphone videos.
Zhengkuan Zhao, Mingkuan Zhao, Min Lu and 7 others
PMID 41792743WHAT IT FOUND
Self-recorded smartphone videos of sit-to-stand and walking were analyzed by deep learning.
The best model helped separate advanced knee osteoarthritis from age-related physical decline using sit-to-stand plus age and body mass index. This is preliminary screening feasibility, not a validated diagnostic tool.
Key findings
01The best reported model used sit-to-stand video plus age and body mass index, with an AUC of 0.981 (95% CI 0.977 to 0.985).
02The differences between groups were more pronounced in the sit-to-stand test than in gait analysis.
03The reported AUC for the combined sit-to-stand and gait model was 0.890 (95% CI 0.878 to 0.901), while the sit-to-stand-only model had an AUC of 0.937 (95% CI 0.929 to 0.944).
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 study was done at one hospital, and participants had to be able to walk normally and not regularly use adaptive walking aids. Videos were self-recorded without controlled angle or distance, and the study did not quantify how the remote smartphone data differed from laboratory motion capture. The reported performance came from 5-fold cross-validation and an 8:2 training-to-testing ratio on the collected dataset; the supplied methods do not describe testing on a separate outside sample. The two groups differed in age, body mass index and sex distribution, and the best model included demographic characteristics, so it is unclear how much of the performance came from movement data rather than demographic differences. The study only compared advanced knee osteoarthritis with age-related physical decline and did not examine early knee osteoarthritis.
Declared interests
The study was funded by the Natural Science Foundation of Tianjin Municipality. The supplied text does not include other conflict-of-interest declarations.
The easy way to misread this
Do not use this method to diagnose advanced knee osteoarthritis in individual patients. The paper reports preliminary feasibility, 5-fold cross-validation, no quantified comparison with laboratory motion capture, and no testing for early knee osteoarthritis.
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 →