Validation of neuroimaging-based brain age gap as a mediator between modifiable risk factors and cognition

Chang Le Chen, Ming Che Kuo, Pin Yu Chen, Yu Hung Tung, Yung Chin Hsu, Chi Wen Christina Huang, Wing P. Chan, Wen Yih Isaac Tseng

Research output: Contribution to journalArticlepeer-review

6 Citations (Scopus)


Neuroimaging-based brain age gap (BAG) is presumably a mediator linking modifiable risk factors to cognitive changes, but this has not been verified yet. To address this hypothesis, modality-specific brain age models were constructed and applied to a population-based cohort (N = 326) to estimate their BAG. Structural equation modeling was employed to investigate the mediation effect of BAG between modifiable risk factors (assessed by 2 cardiovascular risk scores) and cognitive functioning (examined by 4 cognitive assessments). The association between higher burden of modifiable risk factors and poorer cognitive functioning can be significantly mediated by a larger BAG (multimodal: p = 0.014, 40.8% mediation proportion; white matter-based: p = 0.023, 15.7% mediation proportion), which indicated an older brain. Subgroup analysis further revealed a steeper slope (p = 0.019) of association between cognitive functioning and multimodal BAG in the group of higher modifiable risks. The results confirm that BAG can serve as a mediating indicator linking risk loadings to cognitive functioning, implicating its potential in the management of cognitive aging and dementia.

Original languageEnglish
Pages (from-to)61-72
Number of pages12
JournalNeurobiology of Aging
Publication statusPublished - Jun 2022


  • Brain age gap
  • Cognitive aging
  • Machine learning
  • Mediation
  • Modifiable risk factor
  • Neuroimaging

ASJC Scopus subject areas

  • Neuroscience(all)
  • Ageing
  • Clinical Neurology
  • Developmental Biology
  • Geriatrics and Gerontology


Dive into the research topics of 'Validation of neuroimaging-based brain age gap as a mediator between modifiable risk factors and cognition'. Together they form a unique fingerprint.

Cite this