Table 3.

Brain age prediction.

Model Performance—Training Sample (Cam-CAN)
Model	R2	RMSE	MAE	R2nestedCV	RMSEnestedCV	MAEnestedCV
DTI	0.82 ± 0.03	7.81 ± 0.48	6.17 ± 0.40	0.82 ± 0.03	7.75 ± 0.62	6.13 ± 0.52
T1	0.81 ± 0.03	7.97 ± 0.50	6.26 ± 0.42	0.81 ± 0.02	8.07 ± 0.29	6.39 ± 0.21
Model Performance—Test Sample (MTBI)
Model	R2	RMSE	MAE	R2 corr	RMSE corr	MAE corr
DTI	0.58	11.89	9.76	0.76	9.01	7.48
T1	0.56	10.35	8.63	0.78	7.65	6.20

R², root mean square error (RMSE), and mean absolute error (MAE) for the age prediction models within the training sample (the Cambridge Centre for Ageing and Neuroscience, Cam-CAN), and when applied to the test sample (MTBI patients). For the model validation within the training sample, means and standard deviations are provided based on 5-fold cross validations with 100 repetitions. NestedCV indicates the values based on nested cross-validation for hyperparameter optimization. For the test sample, R², RMSE, and MAE are provided before and after age-bias correction (described in Section 2.5).