UW–Madison AI tools map Alzheimer's changes across 6.3M cells | The Locally Times

Published 10/2/2026, 9:14:38 AM UTC · The Locally Times

Researchers in Daifeng Wang's lab built PASCode and iBrainMap using brain cells from nearly 1,500 autopsy donors.

Researchers at the University of Wisconsin–Madison have developed two AI-driven bioinformatics tools, PASCode and iBrainMap, that link cellular and genetic changes in the brain to Alzheimer's disease symptoms. The tools were built on a dataset of more than 6.3 million cells from the brains of nearly 1,500 autopsy donors, according to a September 29, 2026 university news article. ## What is known The tools were designed in the lab of Daifeng Wang, a UW–Madison professor of biostatistics, medical informatics and computer sciences and an investigator at the university's Waisman Center. PASCode was published in Nature Medicine and iBrainMap in Nature Communications, and both use an AI model called graph neural networks. The tools aim to identify specific genes and cell populations involved in disease progression and cognitive resilience, and to create gene roadmaps for individuals with the condition. PASCode helps researchers identify which cellular changes and alterations in gene expression may be linked to specific Alzheimer's symptoms, information the university says could be used to design new treatments. ## Background The work is part of a group of Alzheimer's findings by the PsychAD Consortium, a National Institutes of Health-supported effort led by Wang and Panos Roussos, a professor of genetics and genomics sciences at Mount Sinai Medical School. The consortium includes researchers around the country working to identify biomarkers and therapeutic targets for Alzheimer's disease and neuropsychiatric symptoms. The university article states that Alzheimer's is difficult to study because it presents and progresses differently in individual patients, making it hard to determine which biological changes relate to specific symptoms. The article describes the potential to improve diagnosis and pinpoint therapeutic targets as a possibility, not a measured outcome.