Fig. 5

Discriminative capacity of the genome-wide polygenic score and clinical risk factors in the testing dataset (\(n=\mathrm{15,776})\). The C statistic estimate of major osteoporotic fractures is first obtained with a baseline model of age, height, and weight using a multiple logistic regression model. Next, the C statistic was calculated after the additional inclusion of individual clinical risk factors; parental fracture, previous fracture, smoking, glucocorticoid use, rheumatoid arthritis, hip bone mineral density, spine bone mineral density, genetic risk score (1103 SNPs from GWAS), and genome-wide polygenic score (LDPred with ρ = 0.03)