Table 4.

The prediction results of 10 algorithms in the test cohort.

Model	R2	MAE	MSE	RMSE	Accuracy of the predicted TDM within ±30% of the actual TDM
XGBoost	0.512	10.97	198.55	14.09	54.82%
LightGBM	0.269	12.85	297.52	17.25	47.72%
CatBoost	0.403	11.55	242.88	15.58	51.78%
AdaBoost	0.344	13.13	267.07	16.34	45.18%
Random forest	0.487	11.17	208.69	14.45	52.82%
SVM	0.486	10.60	209.39	14.47	53.81%
KNN	0.181	13.76	333.38	18.26	48.22%
Linear regression	0.482	11.18	210.98	14.53	54.31%
Lasso regression	0.482	11.19	210.97	14.52	54.31%
Ridge regression	0.482	11.19	210.99	14.53	54.31%

XGBoost, Extreme Gradient Boosting; SVM, support vector machine; KNN, k-Nearest Neighbor; MSE, Mean Square Error; RMSE, Root Mean Square Error; MAE, Mean Absolute Error. Bold values mean the best prediction performance among ten models.