Stanford Medicine researchers have developed a scoring system using resting-state functional MRI scans that predicts Alzheimer’s disease and cognitive decline years before symptoms appear.
Published in Science Advances, the approach analyzes brain connectivity to identify both vulnerability and exceptional resistance to memory loss.
ARC Scoring Identifies Cognitive Risks Through Brain Patterns
- Early Detection: The new ARC scoring system evaluates brain activity patterns to spot future cognitive risks long before traditional memory tests register a change.
- Super-Ager Benchmarking: By comparing scans directly against individuals over 75 with exceptional memory, the model identifies protective brain networks.
- Preventive Potential: Researchers hope future interventions like cognitive training could eventually shift vulnerable brain activity toward healthier patterns.
Mapping Resilience Versus Vulnerability
Most conventional diagnostic frameworks compare Alzheimer’s patients against average older adults, a population that often harbors silent, preclinical pathology. To bypass this noise, the Stanford team trained their machine learning model using resting-state functional magnetic resonance imaging scans from two distinct cohorts aged 75 and older.
The study analyzed scans from 180 super-agers—individuals who maintain exceptional memory into old age—alongside 110 patients diagnosed with Alzheimer’s disease. This direct contrast of phenotypic extremes allowed computational models to isolate distinct functional connectivity patterns across prefrontal, temporal, insular, and parietal regions.
“The ARC score summarizes whether an individual’s brain connectivity more closely resembles that of SuperAgers, who maintain exceptional memory into old age, or individuals with Alzheimer’s disease,” Kanhao (Alex) Zhao told ScienceAlert.
From these differences, the team established the Alzheimer’s resilience-vulnerability connectome, or ARC score. The framework distinguished super-agers from Alzheimer’s patients with approximately 85% accuracy in initial testing.
| Cohort Group | Sample Size | Primary Baseline Characteristics |
|---|---|---|
| Super-Agers | 180 | Aged 75+, exceptional episodic memory, preserved cortical connectivity |
| Alzheimer’s Patients | 110 | Clinically diagnosed with Alzheimer’s disease, matched by age group |
| Cognitively Healthy Adults | 1,315 | Aged 55 to 70, no initial cognitive impairment across validation datasets |
Longitudinal Tracking of Asymptomatic Adults
Researchers then applied the ARC scoring system to more than 1,300 cognitively healthy adults aged 55 to 70 across multiple major research initiatives, including the Open Access Series of Imaging Studies and the Alzheimer’s Disease Neuroimaging Initiative. Although these participants showed no initial cognitive differences, their subsequent trajectories diverged significantly.
Over follow-up periods extending up to 12 years, individuals with high ARC scores—indicating brain activity resembling Alzheimer’s patients—experienced more rapid declines on standard tests of memory and executive function. They also showed greater amyloid-beta accumulation and a higher conversion rate to mild cognitive impairment or clinical Alzheimer’s disease.
“We used the ARC score to make a prediction of how likely this patient will convert into mild cognitive impairment or Alzheimer’s disease,” Zhao said. “Our model shows comparable or better prediction ability, and it can do so at an earlier time point. And if we combine our score with the existing biomarkers, the prediction performance improves,” Zhao said.
ARC Scoring Is Not a Clinical Diagnosis Tool
Researchers emphasize that the ARC scoring system is strictly an investigative tool for tracking cognitive risk and is not cleared for clinical diagnosis. Brain activity patterns can shift over time due to advancing age, brain injuries, or neuropsychiatric conditions, meaning an initial resilience score does not guarantee permanent protection.
Patients experiencing persistent memory lapses, confusion with routine tasks, or noticeable changes in executive function should consult a qualified neurologist for a comprehensive diagnostic evaluation.
Future Directions in Preventive Neurology
The discovery provides precise anatomical targets for future neuromodulation techniques and preventive clinical trials. While approaches like transcranial magnetic stimulation and targeted cognitive training have successfully altered brain activity in other contexts, researchers caution that neither intervention has yet been proven to alter the clinical course of Alzheimer’s disease.
Funding for the research was provided by the National Institutes of Health, a Rosenkranz Foundation-Knight Initiative program, an Alzheimer’s Association grant, and a Lehigh University CORE grant.
“There are two main components: One is to know which part of the brain to target, the other is how we should change it,” Yu Zhang, PhD, assistant professor of psychiatry and behavioral sciences at Stanford University, said.
“We want to address a major challenge in the Alzheimer’s field nowadays — most studies are looking at well-known Alzheimer’s disease biomarkers, such as amyloid-beta or tau,” said Yu Zhang, PhD, assistant professor of psychiatry and behavioral sciences at Stanford University. “What we really care about is defining early markers that can detect earlier signs of cognitive decline and potentially help us deliver timely interventions to people.”
References
- Zhao, K., et al. (2026). Alzheimer’s Resilience–Vulnerability Connectome predicts cognitive decline and clinical conversion. Science Advances.
- Research led by Stanford Medicine developed the Alzheimer’s Resilience-Vulnerability Connectome, or ARC score. National Institutes of Health.
