Construction of a community-based primary screening and hospital-based confirmatory screening pathway in pediatric nonalcoholic fatty liver disease

Table 3 Diagnostic performances of the machine learning models based on the BIU index for the diagnosis of ≥ F2 in the training and validation sets

	AUC (95%CI)	P value1	Sensitivity	Specificity	Accuracy	PPV	NPV	Kappa
Training set
LR	0.81 (0.71, 0.90)	Reference	0.84	0.74	0.80	0.81	0.79	0.59
DT	0.89 (0.82, 0.95)	0.547	0.80	0.83	0.81	0.86	0.76	0.62
SVM	0.98 (0.95, 1.00)	< 0.001	0.91	0.91	0.91	0.93	0.89	0.82
RF	1.00 (1.00, 1.00)	< 0.001	1.00	1.00	1.00	1.00	1.00	1.00
ANN	0.82 (0.73, 0.91)	0.534	0.84	0.80	0.83	0.84	0.80	0.64
XGBoost	1.00 (0.99, 1.00)	< 0.001	0.98	1.00	0.99	1.00	0.97	0.98
Internal validation set
LR	0.88 (0.70, 1.00)	Reference	0.82	0.70	0.76	0.75	0.78	0.52
DT	0.80 (0.61, 0.96)	0.187	0.73	0.80	0.76	0.80	0.73	0.53
SVM	0.95 (0.84, 1.00)	0.377	0.91	0.60	0.76	0.71	0.86	0.52
RF	0.86 (0.67, 0.99)	0.841	1.00	0.60	0.81	0.73	1.00	0.61
ANN	0.90 (0.76, 1.00)	0.762	0.91	0.80	0.86	0.83	0.89	0.71
XGBoost	0.75 (0.50, 0.95)	0.325	0.82	0.60	0.71	0.69	0.75	0.42

¹Delong test (two-sided 5% significance level).

The BIU index is a logistic regression model developed in this study that includes three parameters: Body mass index, insulin, and uric acid. LR: Logistic regression; DT: Decision tree; SVM: Support vector machine; RF: Random forest; ANN: Artificial neural network; XGBoost: EXtreme gradient boosting; AUC: Area under the curve; PPV: Positive predictive; NPV: Negative predictive value.