New AI Model Improves Dementia Detection for Minorities

A new formula improves diagnostic accuracy for minority patients to help ensure equitable dementia care across the U.S.

Updated on Oct. 20, 2026 in Alzheimer’s

Isometric editorial illustration of a translucent geometric brain model, visualizing medical data analysis for dementia screening. — Researchers at UT San Antonio developed the RegAlign AI formula to increase diagnostic accuracy for dementia among minority patient populations. AI Illustration. Upload story photo >

Live Poll

Should medical AI tools be required to meet strict accuracy standards for all ethnic groups?

Researchers at the UT San Antonio Health Science Center have developed a new AI formula, RegAlign, to improve dementia detection accuracy for minority patients. Published on June 26, the model significantly narrows the diagnostic accuracy gap between different demographic groups.

Why it matters

Original AI diagnostic tools often showed lower accuracy for minority patients, potentially delaying care. This development helps ensure that machine-learning-based health tools can provide reliable dementia screenings across diverse populations.

This study utilized 9,007 MRI scans to train the RegAlign formula, which measures the volume of 144 brain regions to detect Alzheimer's. Researchers balanced a test set of 852 scans to confirm that accuracy gains in minority populations were independent of sample size.

The players

UT San Antonio Health Science Center

A research institution focused on advancing medical diagnostics and health equity.

The details

The RegAlign formula functions by assigning greater statistical weight to data from minority patients during the AI training process. By adjusting how the model interprets brain region volumes based on ethnicity, the tool reduces the systematic bias that previously led to higher rates of missed dementia cases. The model specifically targets volumetric changes across 144 distinct brain regions to provide more accurate clinical assessments.

June 26, 2026: Study published in Nature Communications.
October 6, 2026: Findings presented at the neurepiomics conference.

Health Landscape

This development represents a critical shift in medical AI, moving from generalized models to those that account for demographic-specific biological variability. It sits within a growing effort to eliminate algorithmic bias that has historically limited the efficacy of diagnostic tools for minority populations.

If you are concerned about cognitive changes, talk with your doctor about the most accurate diagnostic tools available for your demographic profile. This research highlights the importance of asking your physician whether diagnostic tests are validated for your specific background.

The takeaway

AI tools are becoming more equitable, improving the accuracy of dementia detection for a wider range of patients. Stay informed about these advancements by discussing with your primary care provider how screening technologies are evolving to better support diverse health needs.