Automated machine learning for early prediction of acute kidney injury in acute pancreatitis

Table 2 Comparison of LR and AutoML models for early prediction of AKI in the validation cohort

AUC(95%CI)		p-value	Sensitivity	Specificity	Accuracy	PPV	NPV	LR+	LR−
AutoML
GBM	0.812(0.705–0.801)	＜0.001	0.759	0.811	0.798	0.550	0.917	4.004	0.298
DRF	0.800(0.696–0.904)	＜0.001	0.655	0.853	0.806	0.576	0.890	4.446	0.404
GLM	0.734(0.630–0.838)	＜0.001	0.655	0.768	0.742	0.463	0.880	2.829	0.449
DL Logistic regression	0.830(0.734–0.926)	＜0.001	0.759	0.832	0.815	0.579	0.919	4.504	0.290
LASSO	0.799(0.694–0.905)	＜0.001	0.586	0.947	0.799	0.773	0.882	11.138	0.437

LR: logistic regression; AutoML: automated machine learning; AKI: acute kidney injury; PPV: positive predictive value; NPV: negative predictive value; LR+: positive likelihood ration; LR−: negative likelihood ratio

ISSN: 1472-6947