Practical investigation of the performance of robust logistic regression to predict the genetic risk of hypertension

Table 4 Concordance, sensitivity, specificity, clinical net benefit, and overall AUCs.

Probability cutoff	Standard logistic regression				Robust logistic regression
	Concordance N (%)	Sensitivity	Specificity	Clinical net benefit	Concordance N (%)	Sensitivity	Specificity	Clinical net benefit
0.0	43 (33.1)	1.00	0.00	0.33	43 (33.1)	1.00	0.00	0.33
0.1	79 (60.8)	0.95	0.44	0.27	82 (63.1)	0.88	0.51	0.26
0.2	90 (69.2)	0.86	0.61	0.22	97 (74.6)	0.86	0.69	0.23
0.3	98 (75.4)	0.81	0.72	0.19	99 (76.2)	0.81	0.74	0.19
0.4	98 (75.4)	0.70	0.78	0.13	102 (78.5)	0.72	0.82	0.16
0.5	101 (77.7)	0.60	0.86	0.11	107 (82.3)	0.67	0.90	0.15
0.6	97 (74.6)	0.40	0.92	0.05	102 (78.5)	0.51	0.92	0.09
0.7	99 (76.2)	0.35	0.97	0.06	100 (76.9)	0.42	0.94	0.05
0.8	93 (71.5)	0.19	0.98	0.00	97 (74.6)	0.30	0.97	0.01
0.9	91 (70.0)	0.12	0.99	−0.03	93 (71.5)	0.19	0.98	−0.08
1.0	87 (66.9)	0.00	1.00	-	87 (66.9)	0.00	1.00	-
AUC	0.835				0.830

These characteristics rely on the age-genotype models for standard and robust logistic regression estimated based on leave-one-out cross-validation.

ISSN: 1753-6561