Distribution Bias in Brain Age Research: Towards age-specific interpretation of Brain Age Gaps

Biol Psychiatry Cogn Neurosci Neuroimaging. 2026 Mar 16:S2451-9022(26)00078-9. doi: 10.1016/j.bpsc.2026.03.005. Online ahead of print.

ABSTRACT

BACKGROUND: The brain age biomarker estimates biological age from brain structure and is discussed as a potential screening tool for clinically relevant brain aging patterns in individuals. For brain age estimates to be of clinical utility, they must be meaningful for individual patients and free from systematic bias. Here, we investigate how biases from training data age-skewness, which we call distribution bias, impact the reliability and biological interpretability of this promising biomarker.

METHODS: Using Monte Carlo simulations with data from 9,305 individuals and external validation in neuropsychiatric cohorts (1,345 individuals), we trained 100 brain age models for each of four differently age-skewed training distributions, respectively. For each model, we evaluated predictive performance, conducted standard group-level analyses for different neurodegenerative and psychiatric diseases and evaluated clinical utility of the prediction as an individual risk marker.

RESULTS: Training data age-distribution significantly influenced model predictions, causing substantial fluctuations in predicted brain ages across the aging continuum. Statistical analyses revealed that these fluctuations impact effect sizes and statistical significance across all diseases. Moreover, we found limited effectiveness of the brain age gap as an individual risk marker and different levels of disease-associated brain age across the aging continuum.

CONCLUSIONS: Skewed training data age distributions significantly impact brain age model predictions and may compromise scientific results. Based on our findings, we want to raise awareness for distribution bias and propose age-wise interpretation of brain age gaps as a practical solution for robust research and meaningful clinical application.

PMID:41850583 | DOI:10.1016/j.bpsc.2026.03.005