The right side of the brain shows more signs of aging than the left, new study suggests

A recent study published in PNAS introduces a new artificial intelligence approach to map how different regions of the human brain age. By analyzing MRI scans in fine detail, the research provides evidence that aging does not happen evenly across the brain, with certain areas showing more advanced biological age in individuals with Alzheimer’s disease.

Doctors and scientists use Magnetic Resonance Imaging (MRI), a technique that uses strong magnetic fields to create detailed images of the body’s internal structures, to estimate a person’s “brain age.” This metric tells us whether a person’s brain looks older or younger than their actual chronological age. An older-looking brain is often a risk factor for cognitive decline and neurodegenerative conditions.

Usually, researchers calculate a single, global brain age. But the human brain is a complex organ made of distinct regions, each handling different tasks. For instance, the frontal lobe is involved in complex decision-making, and the hippocampus plays a role in memory. Because these regions perform differently, they also tend to age at different rates. A single brain age score can hide these regional differences, making it harder to detect early signs of localized damage.

“Our earlier work focused on global brain age, where AI estimates how old a person’s brain appears from an MRI,” study author Andrei Irimia, an associate professor of gerontology at the USC Leonard Davis School of Gerontology, told PsyPost. “That provides a useful summary, but it reduces the entire brain to a single number, making the result difficult to interpret biologically.”

“We therefore developed local brain age, which asks not only ‘How old does this brain look?’ but ‘How old does each part of this brain look?'” Irimia explained. “The resulting model transforms a structural MRI into a spatial map of local brain age.”

Free daily newsletter

The research, led by Nikhil N. Chaudhari with Irimia and colleagues at the Irimia Lab, aimed to create a more detailed map of brain aging. The authors wanted to measure biological aging at a fine-grained level, breaking the brain down into tiny three-dimensional pixels called voxels. By doing this, they hoped to pinpoint exactly where the brain is aging the fastest and how these specific regional changes relate to cognitive decline.

To build their localized map of brain aging, the scientists trained a deep learning computer model using MRI scans from 14,748 cognitively healthy adults ranging in age from 19 to 100. “Scale and interpretability were important aspects of the project,” Irimia noted. “We trained the model using 14,748 cognitively normal participants from multiple imaging cohorts, exposing it to substantial variation in age, anatomy, and imaging data.”

“Our goal was not simply to produce an AI model with good predictive accuracy, but to generate an output that neuroscientists can map directly back onto brain anatomy,” he said. “In that sense, local brain age is intended to provide a bridge between AI-based prediction and questions about the biology of human aging.”

After training the system, the research team tested it on a separate group of 1,985 individuals from the Alzheimer’s Disease Neuroimaging Initiative. This testing group included 1,102 adults with normal cognitive function, 354 with mild cognitive impairment, and 529 with Alzheimer’s disease.

When observing the cognitively healthy adults, the model’s estimates were off by about 5.9 years on average. It identified a distinct pattern of aging across the brain. The frontal and temporal lobes, which are heavily involved in higher-level thinking and memory, exhibited a more advanced biological age compared to regions at the back of the brain, such as the visual processing centers in the occipital lobe.

The researchers also looked closely at the brain’s folded outer layer, the cerebral cortex. They found that the sulci, which are the deep grooves between the folds, appeared biologically older than the gyri, the outward-facing ridges of the folds. On average, the local brain age gap (the difference between the estimated biological age and the person’s actual chronological age) was 0.40 years higher in the sulci. The team also observed that the right hemisphere of the brain generally showed an older biological age than the left hemisphere, a pattern that persisted when the researchers retrained the model on a sample balanced between left- and right-handed people.

“The left–right asymmetry was particularly interesting,” Irimia told PsyPost. “The model estimated systematically higher local brain ages in the right hemisphere than in the left, and this difference remained when we investigated whether handedness in the training data could account for it. We do not yet know what produces this asymmetry, but it raises an interesting question about whether structural aging may differ between the two hemispheres.”

When comparing the groups, the researchers found that patients with Alzheimer’s disease and mild cognitive impairment had older local brain ages in specific regions known to be affected early in neurodegeneration. In Alzheimer’s patients, subcortical structures like the pallidum and putamen, which help regulate movement and some types of learning, showed brain age gaps about 3.5 to 3.6 years larger than those of cognitively healthy adults. The hippocampus, a seahorse-shaped structure important for forming new memories, appeared more than three years older on average in the Alzheimer’s group compared to the healthy adults.

“The main message is that the brain does not appear to age uniformly,” Irimia said. “Different regions can exhibit different aging patterns within the same person, and these regional differences become more pronounced in cognitive impairment. This gives us a way to move beyond a single estimate of overall brain age and investigate where aging-related structural differences occur.”

The team also analyzed how these localized aging gaps related to performance on cognitive tests. In patients with Alzheimer’s disease, having an older local brain age in regions like the right pallidum strongly correlated with worse scores on general cognitive screening tests and a measure of everyday functioning. These associations were much weaker in people with mild cognitive impairment and largely absent in healthy adults.

The findings are in line with research covered by PsyPost in May 2026, which demonstrated that mapping regional structural changes across the brain with MRI can track neurodegeneration across the progression from healthy aging to Alzheimer’s disease. The new study differs slightly in its approach. The previous research compared scans to an established Alzheimer’s structural vulnerability blueprint, whereas the current model estimates localized biological brain aging at the voxel level without relying on a pre-existing disease map.

As with all research, there are a few things to keep in mind. The artificial intelligence model was trained using research-grade MRI scans, which are typically much higher quality than the standard scans taken in everyday clinical settings. It is unknown if the system would perform as accurately on the lower-resolution images typically produced by hospital equipment.

Because of this and other factors, the tool is not yet ready for everyday medical use. “I would view local brain age primarily as a research tool at this stage rather than a clinical test,” Irimia noted. “Its importance is that it provides spatial information that is lost when brain aging is summarized with a single number. The next question is whether these regional patterns can eventually help us understand differences in cognitive aging or predict clinically meaningful trajectories, which will require further longitudinal validation.”

Additionally, the model was trained exclusively on the brains of cognitively healthy adults. The researchers suggest that future versions of the software might benefit from training on the brains of individuals diagnosed with Alzheimer’s disease. Including those cases could help the system become more sensitive to specific disease-related changes.

Finally, the study analyzed images captured at a single point in time, with limited follow-up data. Because of this, it is not possible to say whether having an older local brain age directly predicts that a healthy individual will eventually develop cognitive impairment. Tracking individuals over many years with repeated scans is a necessary next step to determine if these local age gaps can serve as an early warning system for dementia.

Irimia echoed this caution against overinterpreting a single scan. “An ‘older’ local brain age does not by itself mean that someone has Alzheimer’s disease or will develop it,” he clarified. “The method identifies structural MRI patterns associated with aging; determining whether it can predict an individual’s future disease trajectory will require longitudinal validation. The longitudinal examples in this study are encouraging, but we regard them as supportive rather than definitive validation.”

Looking forward, the team hopes to uncover the root causes of these regional differences. “A major next step is understanding what local brain age represents biologically,” Irimia told PsyPost. “We want to relate these spatial aging patterns to vascular and cardiovascular health, metabolic factors, lifestyle, and other biomarkers to understand why some regions appear to age faster than others. Longitudinal studies will be particularly important for determining whether changes in local brain age can track aging trajectories and identify factors associated with healthier brain aging.”

The study, “Deep learning maps local brain aging in relation to cognition across human adulthood,” was authored by Nikhil N. Chaudhari, Owen M. Vega Huerta, Samayan Bhattacharya, Nahian F. Chowdhury, Andrei Irimia, and the Alzheimer’s Disease Neuroimaging Initiative.