Brain scan patterns can predict Alzheimer’s disease up to twelve years before standard cognitive tests show any measurable decline, according to a study published on November 4, 2026, in Science Advances.
Researchers at Stanford University developed an artificial intelligence model that analyzes resting-state functional magnetic resonance imaging scans to measure how different brain regions communicate while a patient lies still. WELT reported that the method evaluates functional connectivity to establish an Alzheimer’s Resilience-Vulnerability Connectome score.
The system contrasts two extreme groups as biological benchmarks: Alzheimer’s patients and super-agers, defined as adults over the age of 75 who maintain the memory function of a 45-year-old. The model was trained on 290 individuals and tested on an independent group of 143 additional individuals, distinguishing the two groups with an accuracy of approximately 80 percent.
When applied to more than 1,300 cognitively healthy adults aged 55 to 70, the tool identified participants whose brain patterns resembled the vulnerability end of the scale. Ad-hoc-news.de reported that these individuals experienced faster cognitive decline, higher amyloid-beta accumulation, and an elevated risk of developing Alzheimer’s over a monitoring period lasting up to twelve years.
The Stanford team designed the screening tool to bypass the biological diversity found in normal older populations. Standard control groups often display early, undetected neurodegenerative changes that blur research findings.
By comparing 180 super-agers against 110 Alzheimer’s patients, the AI model mapped a clear baseline.
At the initial screening, adults who later showed rapid cognitive decline exhibited no differences in classic amyloid and tau protein biomarkers compared to resilient participants. The communication network analysis flagged biological brain aging before traditional diagnostic markers appeared.
“Our model shows comparable or better predictive power compared to established biomarkers, and at an earlier stage,” said Kanhao Zhao, first author of the study, in a statement released by Stanford Medicine.
The scoring system remains a research tool rather than a diagnostic test for individual patients. Researchers emphasize that the scan results reflect a biological snapshot rather than a fixed medical outcome.
“An ARC-score is just a snapshot, not a fate,” said Yu Zhang, senior author of the study, in a Stanford Medicine announcement.
The research team plans to launch clinical trials to determine whether targeted cognitive training can shift brain connectivity patterns toward the resilience profile seen in super-agers. Investigators also identified two candidate genes, CLYBL and FRMD6, linked to the observed neural patterns, though their exact roles in inflammation and synapse formation require confirmation in larger studies.
The model measures resting-state functional connectivity across brain networks responsible for memory, focus, and emotion.
The scoring system compares patient brain activity against established data from super-agers over age 75 who retain exceptional memory function.
The ARC-score is not yet approved for clinical use, and researchers are currently planning interventional trials to test its practical applications.
Related reading