Table 3.

Performance of each machine learning approach predicting case duration of spine surgery. Calculation is based on the median quantified by k-fold cross-validation for the bagging regressor, linear regression, random forest regressor, and XGBoost regressor. Current method is based on average of the last 5 instances of the surgery with surgeons input to modify time.

Model or method	Explained variance	Max error	MAEa (minutes)	RMSEb (minutes)	R 2
Current method	0.012	847	180.32	243.30	–0.57
Linear regression	0.345	526.29	127.21	162.84	0.34
RFc regressor	0.768	454.59	62.82	96.51	0.76
Bagging regressor	0.769	454.90	62.83	96.51	0.76
XGBoost regressor	0.778	475.72	44.31	92.95	0.77

^aMAE: mean absolute error.

^bRMSE: root mean square error.

^cRF: random forest.