Multi-time scale dynamic effective brain networks reveal accelerated brain aging in individuals with major depressive...

Estimating brain age, a promising biomarker for evaluating brain health, continues to present significant challenges in terms of accuracy. This study investigates the potential of multi-time scale dynamic effective brain networks (MTS-DEBN) to enhance the prediction of brain age and to identify atypical aging patterns associated with major depressive disorder (MDD) using resting-state functional magnetic resonance imaging (rs-fMRI).

Rs-fMRI data were collected from 80 healthy controls (HC) and 80 MDD patients, including subgroups in current phases (n = 46) and remitted phases (n = 34). Time-series signals were extracted from 116 brain regions to construct dynamic effective networks across four temporal scales, utilizing a coarse-graining algorithm, with an integrated feature set (ALL) created. A support vector regression model was trained using data from the HC group to estimate brain age. The optimal model identified was applied to predict brain age in the MDD groups. Model performance was assessed through mean absolute error (MAE). The brain age gap (BAG) was compared between groups.

The features ALL achieved the highest prediction accuracy in HCs (MAE = 3.64 years). The mean BAG was 1.96 years for HCs, 4.56 years for current MDD, and 3.16 years for remitted MDD. Post hoc tests with Bonferroni correction showed significantly higher BAG in current MDD compared to HC (t = 4.85, p < 0.001) and in remitted MDD compared to HC (t = 2.72, p = 0.009), but no significant difference between current and remitted MDD groups. No significant correlations were found between BAG and depression duration or HAMD scores.

MTS-DEBN significantly improves brain age prediction accuracy and reveals accelerated brain aging in both current and remitted MDD patients. These findings support the use of MTS-DEBN as a sensitive biomarker for tracking brain aging dynamics and treatment effects in neuropsychiatric disorders.