Additional File 2: Table S2 shows the validation results for the three algorithms (eMERGE, modified eMERGE and modified MGB) in the two independent chart review datasets from MGB (138 cases, 70 non-cases; case prevalence 66.3%) and UAB (184 cases, 14 non-cases; case prevalence 92.9%). PPVs ranged from 0.79 to 0.94 with the highest PPV observed for the modified MGB algorithm. Sensitivities ranged from 0.61 to 0.92, with the highest sensitivity observed for the modified eMERGE algorithm. NPVs in the UAB validation dataset were zero for both eMERGE algorithms because all 14 non-cases were classified as cases. Based on these results, we selected the modified MGB algorithm for implementation in eMERGE to define T2D cases and controls for the PRS analysis.
