Table 3.

Comparison of model performance between the model with and without Sasang constitution.

		AUC	Sensitivity	Specificity	F1-score	Balanced classification rate	Accuracy
KNN	Without SC	0.73	0.31	0.91	0.39	0.61	0.75
	With SC	0.73	0.31	0.91	0.40	0.61	0.75
Naive Bayes	Without SC	0.79	0.40	0.91	0.48	0.65	0.78
	With SC	0.79	0.49	0.87	0.53	0.68	0.77
Random forest	Without SC	0.77	0.36	0.92	0.45	0.64	0.78
	With SC	0.78	0.37	0.92	0.46	0.64	0.77
Decision tree	Without SC	0.78	0.35	0.93	0.45	0.64	0.78
	With SC	0.77	0.39	0.92	0.47	0.65	0.78
MLP	Without SC	0.8	0.45	0.91	0.52	0.68	0.79
	With SC	0.8	0.47	0.91	0.53	0.69	0.79
SVM	Without SC	0.8	0.37	0.93	0.48	0.65	0.79
	With SC	0.8	0.38	0.93	0.48	0.65	0.79
Logistic regression	Without SC	0.8	0.39	0.93	0.49	0.66	0.79
	With SC	0.8	0.40	0.93	0.49	0.66	0.79

Sex, age, education, marriage status, smoking, body mass index, alcohol, activity, and stress were included. AUC, area under the receiver operating characteristic curve; KNN, K-nearest neighbor; MLP, multilayer perceptron; SC, Sasang constitution; SVM, support vector machine.