Table 2.

Performance of predictive models under consideration

Algorithm	AUC (95% CI)	Sensitivity (95% CI)	Specificity (95% CI)	Accuracy (95% CI)	F1 (95% CI)
Bayesian Network	0.825 (0.797–0.853)	0.794 (0.751–0.837)	0.720 (0.696–0.744)	0.735 (0.714–0.756)	0.545 (0.521–0.569)
Random Forest	0.842 (0.815–0.869)	0.794 (0.751–0.837)	0.724 (0.700-0.748)	0.738 (0.717–0.759)	0.548 (0.524–0.572)
Gradient Boosting	0.846 (0.819–0.873)	0.776 (0.732–0.820)	0.759 (0.736–0.782)	0.762 (0.742–0.782)	0.567 (0.543–0.591)
Logistic Regression	0.838 (0.811–0.865)	0.779 (0.735–0.823)	0.738 (0.715–0.761)	0.746 (0.725–0.767)	0.552 (0.528–0.576)
Neural Network	0.836 (0.808–0.864)	0.788 (0.745–0.831)	0.723 (0.699–0.747)	0.736 (0.715–0.757)	0.544 (0.520–0.568)

Adjusted for gender, age, BMI, smoking, drinking, SBP, DBP, TC, HDL_C, LDL_C, TG, BUN, Creatinine, and ALT. The champion model was Gradient Boosting. The 95% confidence interval for the Sensitivity (0.756, 0.796), and the 95% confidence interval for the Specificity (0.749, 0.769