Validity of multiple human pose estimation tools for measuring knee impact angles in video-captured falls of older adults
Reese Michaels, Justin Ehrlich, Yajun Mei, Jongsang Son, Stephen N Robinovitch, Jacob J Sosnoff, Yaejin Moon · 2026
Read Validity of multiple human pose estimation tools for measuring knee impact angles in video-captured falls of older adults by Reese Michaels, Justin Ehrlich, Yajun Mei, Jongsang Son, Stephen N Robinovitch, Jacob J Sosnoff, Yaejin Moon (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/plos-pone-0335108.txt.
Description
Michaels and colleagues checked AI pose tools against true motion data for older adults' falls in a 2026 study. Falls injure many older people, and bending the knees may protect the hip, but knee motion in real falls is hard to measure. More than one hundred lab videos of sideways falls were analyzed by three pose tools.
The tools captured broad differences between fall styles, from stiff falls to tucked rolls. They missed finer differences between the two legs, especially in tuck-and-roll falls. This first validation gives fall researchers a usable, cheaper tool with known limits, and it points to fall-specific training as the way to sharpen accuracy.
Catalog record
| CREATOR | Reese Michaels, Justin Ehrlich, Yajun Mei, Jongsang Son, Stephen N Robinovitch, Jacob J Sosnoff, Yaejin Moon |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 46601 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |