Table 4.

Model performance

Cohort	Validation		F-measure	Sensitivity	Specificity	PPV	NPV	AUC
ME	Within-site	CV on JHH	0.37±0.01	0.50±0.05	0.84±0.03	0.29±0.01	0.93±0.01	0.75±0.01
		CV on BMC	0.36±0.10	0.50±0.05	0.88±0.02	0.28±0.02	0.95±0.01	0.79±0.02
	Across-site	Test on BMC	0.34	0.47	0.88	0.27	0.95	0.81
		Test on JHH	0.34	0.58	0.76	0.24	0.93	0.78
CP	Within-site	CV on JHH	0.36±0.10	0.50±0.11	0.68±0.20	0.34±0.06	0.84±0.02	0.71±0.05
		CV on BMC	0.28±0.05	0.56±0.19	0.79±0.11	0.20±0.06	0.95±0.02	0.65±0.05
	Across-site	Test on BMC	0.19	0.85	0.41	0.11	0.97	0.65
		Test on JHH	0.38	0.60	0.71	0.27	0.91	0.73

We report sensitivity, specificity, positive predictive values (PPVs), and negative predictive values (NPVs) at the cutoff thresholds that maximize the F-measure. For cross-validation, the table shows the mean±SD of the performance metric over the folds.

AUC, area under the curve; BMC, Bayview Medical Center; CP, patients with chronic pancreatitis; JHH, Johns Hopkins Hospital; ME, all patients admitted to the medical service.