Table 6.

Performance metrics, including mean squared Error (MSE), of the AI predictions and human ratings, compared to the ground truth (mean of human scores).

Model	Metric	Respect for Tissue	Time and Motion	Quality of Final Product	Overall Performance
Image Model	MSE	-	-	0.146	-
	RMSE	-	-	0.392	-
	MAE	-	-	0.293	-
	R2	-	-	0.778	-
Kinematic Model	MSE	0.336	0.420	-	0.373
	RMSE	0.579	0.648	-	0.610
	MAE	0.523	0.456	-	0.431
	R2	0.337	0.244	-	0.453
Multi-modal Model	MSE	0.480	0.356	0.186	0.194
	RMSE	0.693	0.597	0.431	0.440
	MAE	0.545	0.459	0.331	0.315
	R2	0.136	0.476	0.838	0.618
Rater 1	MSE	0.464	0.348	0.531	0.505
	RMSE	0.681	0.590	0.729	0.710
	MAE	0.528	0.474	0.449	0.407
Rater 2	MSE	0.546	0.553	0.545	0.466
	RMSE	0.739	0.744	0.738	0.683
	MAE	0.586	0.483	0.425	0.436
Rater 3	MSE	0.288	0.363	0.193	0.290
	RMSE	0.537	0.602	0.439	0.539
	MAE	0.409	0.426	0.291	0.336