Table 5.

The performance of models with different criteria.

Model	Accuracy	Area under the curve	Specificity	Sensitivity
Semihandcrafted Bayesian network	.754	.78	.78	.73
Handcrafted Bayesian network	.709	.75	.73	.69
Decision tree	.736	.77	.69	.78
Discriminant analysis	.709	.75	.73	.69
Logistic regression	.736	.81	.78	.69
Support vector machine	.709	.77	.73	.69
K-nearest neighbor	.727	.78	.67	.78
Ensemble method	.773	.81	.78	.76
Mean scorea	.732	.78	.73	.73
Semihandcrafted Bayesian networkb	.682	.67	.60	.76
Logistic regressionb	.709	.74	.73	.69

^aThe mean score includes the results of the DT, DA, LR, SVM, KNN, and Ensemble method models.

^bThese results are based only on the subjects’ basic characteristics, without the scores of physical fitness tests.