Fig. 4

Calibration curves for the training dataset (left) and validation dataset (right), assessing the agreement between predicted and actual probabilities of depression. The “Ideal” line represents perfect calibration where predicted probabilities exactly match actual outcomes. The “Apparent” line shows the observed calibration, while the “Bias-corrected” line accounts for overfitting through internal validation. The close alignment of the “Bias-corrected” line with the “Ideal” line indicates that the model exhibits good calibration in both datasets