AI-Derived Brain Age from MRI Predicts Future Dementia Risk

Dementia develops gradually and can ultimately compromise independent living, making timely risk stratification important for counseling and care planning. This is particularly relevant for patients with subjective cognitive decline, who report memory problems despite normal cognitive testing. To improve early risk assessment, researchers have evaluated an artificial intelligence approach that estimates “brain age” from MRI to predict future dementia risk.

Investigators at the Alzheimer Center of Amsterdam University Medical Centers (Amsterdam UMC) assessed an MRI-derived metric called brain-predicted age difference (brain-PAD), which compares a model-estimated brain age with a person’s chronological age. Using advanced artificial intelligence models, including brainageR and cNeuro, they generated automated brain age estimates from structural MRI. The team then calculated brain-PAD to determine whether this measure added clinically useful information beyond standard visual MRI rating scales.

The study drew on patients enrolled between 2002 and 2024 in the Amsterdam Dementia Cohort and the Subjective Cognitive Impairment Cohort (SCIENCe) project. From these records, the researchers analyzed 799 individuals, including 412 with subjective cognitive decline and 387 with mild cognitive impairment. Baseline data included structural MRI, a standard cognitive test score, and visual MRI ratings, and patients were followed over time for progression to dementia.

Across the cohort, each additional year that a brain appeared older than a person’s actual age was associated with a 6% increase in dementia risk. After accounting for standard MRI visual ratings, brain-PAD still contributed an independent 4% increase in risk for every additional year. In subjective cognitive decline, each extra year of brain age raised the relative risk of progression by 9%, and a threshold emerged: if the brain appeared at least 2.6 years younger than chronological age, patients had a 91% to 99% chance of remaining dementia-free over the next two to 10 years. In mild cognitive impairment, brain age added little beyond what visual ratings already showed. The findings were published in Neurology on August 10, 2026.

The authors note that additional studies are needed before routine deployment in clinics. Even so, MRI-derived brain-PAD offers a sensitive, automated measure that could complement existing visual scales for earlier risk stratification. Such integration could help clinicians identify higher-risk patients sooner, align follow-up intensity with risk, and provide families with clearer expectations and options, including clinical trial consideration.

Related Links
Alzheimer Center Amsterdam UMC