Abstract Brain aging is accompanied by progressive morphological and neurobiological changes, which are significantly accelerated in neurodegenerative diseases, such as Alzheimer’s disease. Detecting and differentiating these changes early is crucial for diagnosis, treatment planning, and therapeutic development. In this work, we present a computational multiphysics framework that couples protein biomarker propagation with tissue-level atrophy to distinguish between cognitively normal aging, mild cognitive impairment, and Alzheimer’s disease. Our model integrates a network-based simulation of amyloid beta and tau protein spread with a finite element model of brain mechanics to simulate longitudinal brain shape changes over 40 years. Notably, we observe that amyloid beta accumulation precedes tau-driven degeneration by over a decade, aligning with empirical biomarker studies. We also introduce several mechanomarkers which are quantitative metrics of brain morphology such as displacement, cortical thickness, curvature, and sulcal depth. They serve as quantitative measures of disease-specific deformation patterns. Our simulations predict that Alzheimer’s disease accelerates cerebral atrophy by about 12 years relative to normal aging, with early divergence in medial temporal and occipital regions. Our findings identify cortical thickness and area stretch as early and sensitive markers to distinguish between healthy and abnormal aging. Spatially, the supramarginal gyrus and entorhinal cortex should be considered as regions of early vulnerability. These results underscore the potential of physics-informed computational models to improve early detection of neurodegeneration and guide the development of region- and stage-specific diagnostic tools.
Jalalian et al. (2026) studied this question.