Table 2.

Selected diagnostic test characteristics for all linear models to detect an M value in the lowest quartile

Model covariates	Sensitivity	Specificity	PPV	NPV	Balanced accuracy
RISC → RISC
Proteins	0.458	0.963	0.805	0.842	0.711
Centre-Age-Sex-BMI	0.208	0.986	0.833	0.789	0.597
Centre-Age-Sex-BMI-Proteins	0.542	0.912	0.672	0.857	0.727
Centre-Age-Sex-BMI-Lipids-SBP	0.250	0.981	0.818	0.796	0.616
Centre-Age-Sex-BMI-Lipids-SBP-Proteins	0.444	0.916	0.640	0.831	0.680
BMI-Lipids-SBP	0.236	0.981	0.810	0.793	0.609
BMI-Lipids-SBP-Proteins	0.486	0.930	0.700	0.844	0.708
ULSAM → ULSAM
Proteins	0.250	0.964	0.700	0.792	0.607
Centre-Age-Sex-BMI	0.107	1.000	1.000	0.769	0.554
Centre-Age-Sex-BMI-Proteins	0.357	0.976	0.833	0.818	0.667
Centre-Age-Sex-BMI-Lipids-SBP	0.145	0.994	0.889	0.778	0.570
Centre-Age-Sex-BMI-Lipids-SBP-Proteins	0.455	0.988	0.926	0.845	0.721
BMI-Lipids-SBP	0.164	0.994	0.900	0.782	0.579
BMI-Lipids-SBP-Proteins	0.455	0.988	0.926	0.845	0.721
RISC → ULSAM
Proteins	0.299	0.984	0.862	0.807	0.642
BMI-Lipids-SBP	0.075	0.996	0.875	0.764	0.536
BMI-Lipids-SBP-Proteins	0.226	0.986	0.840	0.793	0.606
ULSAM → RISC
Proteins	0.310	0.942	0.641	0.803	0.626
BMI-Lipids-SBP	0.213	0.969	0.699	0.787	0.591
BMI-Lipids-SBP-Proteins	0.479	0.915	0.653	0.840	0.697

Sensitivity = TP/(TP+FN); specificity = TN/(TN+FP); PPV = (sensitivity×prevalence)/((sensitivity×prevalence)+((1–specificity)×(1–prevalence))); NPV = (specificity×(1–prevalence))/(((1–sensitivity)×prevalence)+((specificity)×(1–prevalence))); balanced accuracy = (sensitivity+specificity)/2

FN, false negative; FP, false positive; NPV, negative predictive value; PPV, positive predictive value; TN, true negative; TP, true positive

RISC → RISC, models trained in RISC and tested in RISC; ULSAM → ULSAM, models trained in ULSAM and tested in ULSAM; RISC → ULSAM, models trained in RISC and tested in ULSAM; ULSAM → RISC, models trained in ULSAM and tested in RISC