How AI Is Changing the Pace of Discovery for Millions Living With Dementia

The Consortium for Biomedical Research and AI in Neurodegeneration, known as C-BRAIN, is working with NVIDIA to apply open source AI tools to Alzheimer's research, according to an NVIDIA blog post dated October 27, 2026. Launched in November 2025 and led by Dr. Randall Bateman of Washington University School of Medicine in St. Louis, the group includes 18 member institutions from academia, philanthropy, pharma and research.

NVIDIA frames the problem as one of synthesis: 57 million people live with dementia, and although annual NIH funding has grown sevenfold in 15 years to $3.9 billion, no treatment yet stops, prevents or reverses Alzheimer's. C-BRAIN targets what it calls "dark data," including unpublished findings, negative results and proprietary datasets. A hybrid centralized and federated setup lets models learn from institutional data without it leaving those institutions. The consortium runs 64 NVIDIA data center GPUs on premises plus Microsoft Azure cloud nodes, and is incorporating NVIDIA's BioNeMo platform and the MONAI framework.

Two open source agentic tools already exist. OpenScientist, built with Lawrence Berkeley National Laboratory, generates hypotheses and runs analyses semi-autonomously. InsightEngine, built with 387 Labs, tests hypotheses against the literature and checks claims against evidence to catch hallucinations. NVIDIA cites published case studies in which OpenScientist finished in minutes, for under $10 per run, analyses scientists estimated would take weeks to months. It suggests that across more than 200 research groups, the savings could exceed one million hours a year at scale.

C-BRAIN next aims to build a multimodal biological foundation model and, by year three, identify new causal mechanisms of brain aging and Alzheimer's.