Results There was a highly significant moderate agreement κ = 0.461 ** ( p < 0.001; Bonferroni-corrected p -value [ p B ] = 0.006) (95% confidence interval [CI], 0.281–0.642) between the RULA scores assessed by both technologies.
A clear trend emerged when examining the variability (SD) across the body part scores: the IMU consistently had a higher SD than pose estimation, indicating that it can capture more detailed movement fluctuations, whereas pose estimation appears to produce less fluctuating postural estimates.