Performance of machine learning models in estimation of ground reaction forces during balance exergaming.
Elise Klæbo Vonstad, Kerstin Bach, Beatrix Vereijken and 2 others
PMID 35152877WHAT IT FOUND
An LSTM model accurately estimated vertical ground reaction force from 2D video during balance exergaming, with errors under 11% body weight.
This suggests standard cameras could provide real-time weight-shifting biofeedback for home use, though horizontal force estimation was less accurate.
Key findings
01The LSTM model outperformed XGBoost and linear regression in estimating ground reaction forces using both 3D motion capture and 2D digital video data.
02Using 2D digital video data, the LSTM model achieved a mean root mean square error of 10.7% body weight for the vertical force component.
03The study suggests that standard digital video cameras can provide accurate information about weight-shifting performance for in-home exergaming.
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
Small sample size of only 12 healthy older adults. The movement pattern was restricted to sideways leaning; results may not generalize to other balance exercises. Data was collected in a laboratory setting, not in home environments. Estimation of horizontal forces (anterior-posterior and medio-lateral) was less accurate, especially with 2D data. The models require training data to be usable, which may limit immediate real-world deployment.
Declared interests
The authors declared no competing interests. The study was funded by the Research Council of Norway through the Centre for Advanced Analysis in Sports and Health (CAASH).
The easy way to misread this
Do not assume this technology is ready for clinical use or home practice. The study tested only one specific sideways weight-shifting movement in a lab with healthy older adults, and horizontal force estimation was poor. It is a feasibility study for a potential future tool, not a validated intervention.