Fig. 4: The diagnostic ability of the identified PRS-associated frontal-temporal network. | Nature Communications

Fig. 4: The diagnostic ability of the identified PRS-associated frontal-temporal network.

From: Derivation and utility of schizophrenia polygenic risk associated multimodal MRI frontotemporal network

Fig. 4

a Group differences between SZ and HC of PRS-associated the positive and negative networks for independent BSNIP-1 (SZ: N = 178; HC: N = 220), COBRE (SZ: N = 100; HC: N = 90), fBIRN (SZ: N = 164; HC: N = 157) and MPRC (SZ: N = 164; HC: N = 157) cohorts, respectively. Two-tailed two-sample T test was used to calculate the group differences in (a). The minima, maxima and the mean were displayed in the box plots. b ROC curves of the classification results between SZ and HC for BSNIP-1, COBRE, fBIRN and MPRC cohorts, respectively. The classification accuracies were approximated as around 50% as a random distributed accuracy when treating site as categories (the more number of sites the lower classification accuracy, Supplementary Fig. 16), indicating that site was not a major confounding factor for the current SZ-HC classification analysis. AUC area under the curve, ACC accuracy.

Back to article page