Milan, July 21th 2026 – Alzheimer’s disease is not only about the accumulation of toxic proteins in the brain. A growing number of studies indicate that the health of cerebral blood vessels may also play a decisive role in the onset and progression of the disease. A reduction in blood flow, vascular damage and the accumulation of amyloid beta, one of the key proteins associated with Alzheimer’s, in fact feed into one another in a pathological cycle.
This hypothesis has led to a new “whole-brain” mathematical and computational model developed at the MOX Laboratory of the Department of Mathematics of Politecnico di Milano . The model is designed to describe in an integrated way the interaction between the spread of amyloid beta and the functioning of the cerebral vascular network. The aim is to provide a tool capable of simulating, on the scale of the whole organ, how small biological or vascular alterations can evolve over time and contribute to neurodegeneration . The model and the accompanying scientific study have been published in the prestigious scientific journal Computer Methods in Applied Mechanics and Engineering .
The model integrates two scales of analysis . On the one hand, it describes the dynamics of the production, transformation, diffusion and elimination of the healthy and pathological forms of amyloid beta. On the other, it describes blood flow through a “macroscopic” description of arteries, capillaries and veins, treating brain tissue as a porous medium perfused by blood vessels, through a macroscopic compartmental model. The two components are then connected to represent the possible mechanism of mutual reinforcement between protein accumulation and vascular dysfunction.
The simulations show a particularly relevant result: the brain can evolve into different states depending on the initial conditions. Small localised amounts of amyloid beta can be eliminated, allowing the system to return to a healthy state. Conversely, larger amounts can trigger a self-sustained spread of the pathology at brain scale.
The model also makes it possible to explore the effect of localised vascular damage , such as a reduction in blood flow due to a lesion or to a stroke-like event. In these scenarios, the research suggests that hypoperfusion, that is, an insufficient supply of blood to brain tissue, can make the brain more vulnerable: levels of amyloid beta that under normal conditions would be efficiently compensated can become sufficient to trigger broader pathological progression.
«This work stems from the collaboration between Politecnico di Milano and Stanford University and the University of Oxford » explains Mattia Corti , first author of the study and postdoc at the MOX Laboratory, who carries out his research as part of the activities of the ERC Synergy Grant project “NEMESIS – NEw Generation MEthods for Numerical SImulationS”. «By integrating expertise in mathematical and numerical modelling, mechanical engineering and high-performance scientific computing, we were able to obtain these results, which represent a first step towards understanding how mechanisms known at the microscopic scale can have a global impact in neurodegenerative brain diseases and artificial intelligence».
«For some years now, at the MOX Laboratory we have launched the BraiNum project with the aim of developing a mathematical and computational model that describes the physiological and pathological functioning of the brain and the central nervous system» adds Paola Antonietti , Head of the MOX Laboratory, principal investigator of BraiNum and NEMESIS projects and faculty member in the Department of Mathematics. «To this end, we integrate the potential and flexibility of mathematical modelling, numerical analysis, high-performance scientific computing and artificial intelligence».
The work is part of the so-called “vascular hypothesis” of Alzheimer’s, according to which initial damage to the blood vessels can favour the development or worsening of the accumulation and spread of the amyloid beta protein involved in Alzheimer’s disease. The contribution of the new computational model developed is therefore not diagnostic or therapeutic in a direct sense, but concerns the possibility of understanding and studying in a controlled way scenarios that are difficult to observe experimentally, generating new hypotheses on the mechanisms that favour the onset and progression of the disease.
By linking biological mechanisms, detailed brain geometries reconstructed from “patient-specific” clinical images and advanced, next-generation computational models, the model paves the way for increasingly realistic simulations of the evolution of neurodegeneration . In perspective, tools of this kind may contribute to identifying risk conditions, to better understanding the role of the vascular component in Alzheimer’s and to guiding new prevention and intervention strategies.
Computer Methods in Applied Mechanics and Engineering