An MRI study found older regional brain-age estimates on the right. Learn what the findings mean and why the method remains a research tool.
On October 2, 2026, PsyPost reported on a PNAS study in which researchers used artificial intelligence to estimate age across individual brain regions. The model returned older local age estimates on the right side of the brain than on the left.
The study was titled “Deep learning maps local brain aging in relation to cognition across human adulthood.” According to PsyPost’s coverage, the researchers used structural MRI scans to map apparent age across brain regions instead of producing a single estimate for the whole brain.
That change in approach was central to the work. A whole-brain estimate combines information into one number. A regional map lets researchers examine where age estimates differ and whether particular patterns relate to cognition.
Study coauthor Andrei Irimia described the question as moving from “How old does this brain look?” to “How old does each part of this brain look?”
The researchers trained their deep-learning model on scans from 14,748 cognitively healthy adults aged 19 to 100. This gave the model examples spanning much of human adulthood.
They then tested it on 1,985 people in an Alzheimer’s Disease Neuroimaging Initiative dataset. According to PsyPost, that group included:
These groups allowed the researchers to examine regional estimates in people with different levels of cognitive health. They also compared local brain age with cognitive screening and everyday-functioning scores.
PsyPost reported several patterns beyond the right-left difference. Frontal and temporal areas tended to appear older than areas toward the back of the brain.
The brain’s surface also showed differences. Grooves, called sulci, appeared older than ridges called gyri. The average gap between those estimates was 0.40 years.
The right hemisphere returned systematically older local age estimates than the left. The pattern remained after researchers balanced the training sample for left- and right-handed participants.
Together, these findings describe an uneven map rather than a single aging pattern across the brain. The regional approach made those differences visible within the model’s estimates.
The Alzheimer’s group showed larger brain-age gaps in several structures. PsyPost reported that gaps in the pallidum and putamen were approximately 3.5 and 3.6 years larger than in cognitively healthy adults. The hippocampus appeared more than three years older on average.
The study also examined whether older regional estimates went with poorer performance. Among participants with Alzheimer’s, older local brain age in areas including the right pallidum was associated with worse cognitive screening and everyday-functioning scores.
Those associations were weaker in mild cognitive impairment. They were largely absent in cognitively healthy adults.
This distinction matters when reading the findings. The relationship between regional estimates and functioning differed across the participant groups. The study did not produce the same cognitive pattern in everyone it examined.
On October 5, 2026, USC announced a $3.8 million National Institute on Aging grant for related work using AI to forecast regional brain changes over time.
According to the USC Leonard Davis School of Gerontology announcement, the project will compare AI-generated forecasts with follow-up MRI scans. It will also examine cognitive measures, genetics and blood-based markers.
The research question therefore moves beyond mapping differences in existing scans. The team plans to test whether forecasts match later changes and whether those patterns relate to cognitive impairment or Alzheimer’s risk.
Irimia described the goal as understanding where unusual aging occurs and how quickly change happens. The project will also examine whether AI can anticipate what comes next.
For readers, the immediate value is a better way to interpret brain-aging headlines. Instead of focusing only on whether a brain appears “older,” ask which regions were examined and which participants showed the reported pattern.
The link with functioning deserves equal attention. In this study, associations with cognitive screening and everyday functioning were strongest in the Alzheimer’s group. Keeping that context attached to the finding helps prevent a result from becoming a broad claim about all older adults.
For a personal memory concern, the useful focus remains what is happening in daily life. A qualified healthcare professional can put those concerns in context and discuss whether an assessment is appropriate. A discussion of actual changes serves a different purpose from reading about an experimental imaging result.
Readers sorting through that distinction can turn to a guide to cognitive changes after 60. The aim is to frame questions about memory clearly rather than attach a research label to oneself.
For researchers, the regional maps provide specific patterns to investigate. For clinicians, the future question is whether those patterns add useful information about thinking and everyday function. USC’s follow-up project is directed toward testing that connection over time.
No such conclusion is established by this finding. PsyPost reports that the right-left difference is a model-based result across the study group. The researchers do not yet know what produces it.
An older apparent age in an MRI model is not the same finding as a directly measured faster rate of aging. It also does not establish that the right side of every person’s brain is older. The appropriate reading is narrower: this model detected a consistent difference that needs further explanation.
In the healthy validation group, the model’s estimates were off by about 5.9 years on average, according to PsyPost. That error is a necessary part of interpreting the results.
An average error does not describe the accuracy of every individual estimate. Readers should not treat a regional age score as a precise reading of their brain’s condition. A detailed map can still contain uncertainty.
PsyPost’s coverage describes local brain age as a research tool rather than a clinical test. Irimia cautioned that an older local estimate alone does not indicate Alzheimer’s disease or establish that someone will develop it.
The differences reported in the Alzheimer’s group are group comparisons. They are not personal diagnostic thresholds. The hippocampus finding therefore cannot be turned into a rule for deciding whether an individual has the disease.
Related research comparing healthy aging with Alzheimer’s brain networks provides another topic for reading about imaging findings. It should remain separate from decisions about an individual’s diagnosis or care.
The study used research-grade MRI scans. PsyPost reports uncertainty about how the model would perform on lower-resolution scans used in ordinary clinical settings. That question matters before a research method can be applied more broadly.
The study also did not test whether lifestyle changes, supplements or treatments could reverse a local brain-age estimate. It supplies no specific lifestyle prescription. Personal treatment decisions should remain with qualified care rather than follow a hypothetical regional score.
The USC grant announcement describes a research aim, not a validated forecasting service. Its planned comparisons are future tests of the approach rather than completed evidence of clinical usefulness.
The study’s lasting contribution is a more specific question about cognitive aging. Instead of asking only whether an entire brain appears older, researchers can examine where differences occur and how those patterns relate to functioning.
The next meaningful step is to connect such maps with change over time. USC’s planned comparisons with later scans and cognitive measures address that question directly.
For cognitive health, the value of a regional map will rest on the understanding it adds. More detailed images are worthwhile when they help explain meaningful differences in thinking and daily life. That is the long-term goal to keep in view as this research develops.
Deciding whether a brain-age headline belongs in a personal health discussion calls for the evidence translation at the heart of FitBrainLab. Its plain-language reporting addresses confusion about normal brain aging by distinguishing established knowledge from early findings, helping older readers assess memory research without fear or exaggerated health claims.
Follow FitBrainLab for research-led insights on memory, focus, brain aging, nutrition and mental fitness after 60. Stay connected for new articles, practical guidance and ideas for a sharper, more engaged life.
Build habits that support memory, focus and a curious, connected life.
Read the Blog
