Stanford Brain Score Flags Alzheimer's Risk Early

Inside the brains of some cognitively normal adults, firing patterns foretell Alzheimer's disease years before memory loss starts, Stanford Medicine researchers reported on November 5, 2026. The findings were published November 4 in Science Advances. The score they built predicted decline before plaques or tangles appeared.

The team started at the extremes. They analyzed resting brain activity in adults 75 and older at two ends of the cognitive spectrum: people diagnosed with Alzheimer's and so-called super agers, whose memory matches people decades younger. The contrast between those groups became the measuring stick.

That measuring stick is the Alzheimer's resilience-vulnerability connectome, or ARC score. It distinguished super agers from Alzheimer's patients with roughly 85 percent accuracy. The more Alzheimer's-like a person's activity looked, the worse their odds ran over the following years.

The researchers then applied the score to more than 1,300 people aged 55 to 70 who showed no signs of cognitive impairment. Follow-up ran as long as 12 years in the longest datasets, though most tracking lasted five to six years. The score still sorted the group by who declined.

The comparison is the headline inside the result. The ARC score predicted cognitive decline more accurately than demographics, amyloid-beta and tau biomarkers, and genetics. Those markers are the field's standard early-warning tools, and a ten-minute resting scan outperformed them in this analysis.

People whose patterns leaned Alzheimer's-like were more likely to lose ground on cognitive tests, develop amyloid-beta plaques and receive an Alzheimer's diagnosis. People whose patterns leaned super-ager-like were more likely to avoid all three outcomes. The score reads a spectrum, not a yes-or-no verdict.

The authors are careful about the limit. Senior author Yu Zhang, an assistant professor of psychiatry and behavioral sciences, noted that an ARC score is a snapshot, not destiny. The team says the score shouldn't be considered diagnostic and is better used to monitor cognitive status over time.

The study drew funding from the National Institutes of Health, a joint Rosenkranz Foundation and Knight Initiative program, an Alzheimer's Association grant and a Lehigh University grant. Whether a score like this can guide treatment is the open question. For now it's a sharper forecast, and forecasts only matter if earlier warning changes what a patient can do.

The datasets behind the score came from established aging research, including the Alzheimer's Disease Neuroimaging Initiative, the Harvard Aging Brain Study and the Open Access Series of Imaging Studies. Pooling them is what gave the team follow-up windows long enough to test a prediction instead of a correlation. A biomarker that only works inside one clinic's scanner wouldn't travel, and the authors built the test across cohorts from the start.