AgeNet-SHAP: An explainable AI approach for optimally mapping multivariate regional brain age and clinical severity patterns in Alzheimer’s disease

Age is a significant risk factor for mild cognitive impairment (MCI) and Alzheimer’s disease (AD) and identifying brain age patterns is critical for comprehending the normal aging and MCI/AD processes. Prior studies have widely established the univariate relationships between brain regions and age, while multivariate associations remain largely unexplored. Herein, various artificial intelligence (AI) models were employed to perform brain age prediction using an MRI dataset (n=668). Then the optimal AI model was integrated with the Shapley additive explanations (SHAP) feature importance technique to identify the significant multivariate brain regions involved in this prediction. Our results indicated that the deep learning model (referred to as AgeNet) tremendously outperformed the conventional machine learning models for brain age prediction, and AgeNet integrated with SHAP (referred to as AgeNet-SHAP) identified all ground-truth perturbed regions as key predictors of brain age in semi-simulation, proved the validity of our methodology. In the experimental dataset, compared to cognitively normal (CN) participants, MCI exhibited moderate differences in brain regions, whereas AD had highly robust and widely distributed regional differences. The individualized AgeNet-SHAP regional features further showed associations with clinical severity scores in the AD continuum. These results collectively facilitate data-driven predictive modelling approaches for disease progression, diagnostics, prognostics, and personalized medicine efforts.