The National Institutes of Health has committed another $12.6 million to Artificial Intelligence for Alzheimer's Disease (AI4AD), pushing its total investment in the project to $30.7 million. The next phase, AI4AD2, is led by Paul M. Thompson, PhD, associate director of the USC Mark and Mary Stevens Neuroimaging and Informatics Institute at the Keck School of Medicine of USC, and will focus on using artificial intelligence to identify the biological drivers of Alzheimer's and related dementias, sharpen predictions about how the disease will progress, and point toward more precise treatments.
The consortium brings together 10 investigators and 23 co-investigators across 10 institutions, working toward four connected research goals. Their raw material is scale: whole-genome sequencing, brain imaging, cognitive testing, and other biological data, analyzed together rather than in isolation. That approach follows from the first AI4AD effort, launched in 2020, which built tools to spot Alzheimer's-related patterns in brain scans and demonstrated that machine learning could connect what shows up on imaging to a person's underlying genetic risk.
Thompson frames the central problem as one of mixture. Individual brains rarely carry a single pathology; Alzheimer's changes may sit alongside vascular disease and alterations more typical of Parkinson's, each advancing at its own pace. That heterogeneity is a large part of why dementia has resisted treatment, since a drug aimed at one process may do little about the others at work in the same patient.
AI4AD2's answer is what Thompson calls genome-guided drug discovery: using genetic data to find compounds aimed at specific forms of dementia, including rarer subtypes that have drawn less pharmaceutical attention. The practical payoff is still some distance off, since this is discovery-stage work rather than a therapy in testing, but it reflects a shift away from treating dementia as one disease toward sorting patients by the biology actually driving their decline.