Continuous assessment of daily-living gait using self-supervised learning of wrist-worn accelerometer data.

Compared with a fully supervised model, ElderNet showed significantly lower errors (MAE: p < 0.001; RMSE: p = 0.045), higher ICC ( p = 0.02), and a trend towards a higher R 2, though this did not reach statistical significance ( p = 0.07).

also in 111,027 other papers

For cadence and stride length, ElderNet consistently outperformed the supervised model across most evaluation metrics, with statistically significant improvements observed for all metrics except cadence R2, which showed a positive trend.

also in 11,005 other papers