Skip to main content
. 2018 Dec 20;44(5):898–906. doi: 10.1038/s41386-018-0298-z

Fig. 4.

Fig. 4

Difference in predicted age and chronological age (age delta) in a sample 1 and b sample 2. Random forest analysis estimates higher delta in chronological versus predicted brain age for participants with schizophrenia. a In the first sample (N = 175; 41 HC, 81 SSD, 53 BD), a model was trained on 78 cortical thickness features, 38 fractional anisotropy features, and 9 cognitive performance features from 50 non-overlapping HCs. The schizophrenia group brain age delta was 7.8 years. Predicted and chronological age differed significantly in schizophrenia (t(80) = 4.98, p < 0.001) but not in HCs (t(40) = −0.35, p = 0.73) and BD (t(52) = −0.13, p = 0.90) participants. Age deltas differed significantly across groups (F(2,172) = 9.94, p < 0.001), with post hoc t-tests showing the schizophrenia group delta to be larger than that of HCs (t(120) = −3.27, p = 0.001) and BD (t(132) = −3.73, p < 0.001). HCs and BD participants were statistically equivalent in their predicted age. b In the second sample (N = 122; 55 HC, 67 SSD) using a model trained on the same 78 cortical thickness features and 38 fractional anisotropy features from 25 non-overlapping HCs, the schizophrenia group brain age delta was 8.20 years. Predicted and chronological age differed significantly in schizophrenia (t(65) = 5.33, p < 0.001) but not in HCs (t(53) = 1.59, p = 0.12). Age delta was significantly higher (t(117.45) = −2.56, p = 0.01) in the SSD group compared with HC. Legend. Diagnoses: HC = healthy control, SSD = schizophrenia spectrum disorder, BD = bipolar disorder. Significance codes: NS = not significant, *p < .05, **p < .01, ***p < .001