Breakthrough AI system detects early signs of memory loss during sleep

During the quiet hours of sleep, your brain is hard at work. It sends out tiny electrical signals that ripple across networks deep in your mind.

A new AI tool reads brainwaves during sleep to predict future memory problems years before symptoms appear.

A new AI tool reads brainwaves during sleep to predict future memory problems years before symptoms appear. (CREDIT: Shutterstock)

During the quiet hours of sleep, your brain is hard at work. It sends out tiny electrical signals that ripple across networks deep in your mind. For years, scientists knew these patterns told stories about your health—but now, new research reveals they might also predict if your memory will fade in the future.

Researchers from Mass General Brigham have created an artificial intelligence tool that reads these electrical signals, known as brain waves. They used a technique called electroencephalography, or EEG, which records the brain’s activity through sensors placed on the scalp during sleep. The results from this research, published in a leading journal, show that the AI can spot early signs of memory loss—years before symptoms begin.

This development could be a game changer in the fight against diseases like Alzheimer’s. Early detection gives doctors and patients valuable time. With the right lifestyle changes and medical care, it may be possible to slow down or even prevent further decline.

EEG machine. (CREDIT: Baburov)

Tiny Waves, Big Clues

The study focused on data from women over the age of 65. These women took part in an earlier sleep and health study and had normal thinking ability at the time. As part of that research, they spent a night hooked up to an EEG machine, which recorded their brain activity as they slept. They also completed a series of mental tests.

Five years later, they took the same mental tests again. By then, 96 of the 281 women had developed noticeable memory and thinking problems, or cognitive impairment.

Researchers used advanced tools, including AI and information theory techniques, to go back and study the brain wave data from the original sleep sessions. They found that the women who later showed signs of memory trouble had subtle differences in their brain waves—especially in a type of wave called gamma waves, which are active during deep sleep.

“The brain leaves little signs of trouble long before any symptoms appear,” said Shahab Haghayegh, a scientist who helped lead the research. “With advanced analysis and AI, we can find these signals in EEG recordings from just one night.”

The AI tool correctly predicted who would develop memory issues in about 85% of cases. Its overall accuracy was 77%. That’s a strong result for a test done before any signs of disease showed up.

Why Early Detection Matters

Memory problems don't usually start overnight. In most cases, the brain begins to change years before you notice anything is wrong. Doctors call this the "preclinical" phase of cognitive decline. That’s when treatments could make the biggest difference.

Right now, treatments for diseases like Alzheimer’s work best in the early stages. But if you wait until symptoms appear, it may already be too late for many options to help. That’s why finding warning signs early is so important.

Diagram shows how sleep EEG signals and brainwave patterns are analyzed with AI to predict future cognitive decline risk. (CREDIT: Shahab Haghayegh, et al.)

“The new, FDA-approved treatments for Alzheimer’s disease are effective at the earlier stages of dementia, but not the more advanced stages,” said Dr. Kun Hu, another lead researcher on the study. “Interventions that are performed even earlier—before someone shows signs of cognitive decline—are likely to be even more effective.”

This kind of testing could give people more time to act. They might adjust their diet, increase exercise, or improve sleep habits—lifestyle changes that are linked to better brain health. Doctors could also start treatments earlier, with a better chance of success.

A Simple Test With Big Potential

What makes this discovery even more exciting is that it comes from a very simple test. EEGs are painless, safe, and already used in hospitals around the world. The researchers believe this kind of brain scan could become part of regular health checkups, especially for older adults.

The test used data from only one night of sleep, but the team says future research could do even more. “EEG data from multiple nights of sleep could be even more predictive of future cognitive impairment,” said Hu.


Like these kind of stories? Get The Brighter Side of News' newsletter


The idea is that the more information the AI tool has, the better it can see patterns. Think of it like a photo. One snapshot shows a moment. But a full album gives a clearer story.

The researchers hope that in the future, people could wear special EEG headbands at home while they sleep. These devices could quietly gather brain wave data over several nights, without the need for a hospital visit. That would make early testing much easier and more affordable.

Looking Toward the Future

This study also opens the door to another bold idea: What if we could change these brain wave patterns before memory problems even begin?

Haghayegh and Hu are already planning a clinical trial to test this. They want to see if giving the brain a small electrical pulse during sleep can adjust the wave patterns that the AI tool identified. If it works, it might actually slow down or prevent cognitive decline.

Charts show how different brainwave frequencies during sleep stages vary in people who later develop cognitive decline. (CREDIT: Shahab Haghayegh, et al.)

“We’re not just looking at prediction,” said Haghayegh. “We’re asking if we can change the future.”

Still, there are some limits to the current findings. The study only looked at older women, and it used EEG data from a single night. The researchers say they now need to test their tool on more diverse groups, including men and younger people, to confirm the results.

They also want to see if the same changes in gamma wave activity show up in other people who develop memory problems. That would help prove that the brain wave signals they found are truly linked to future cognitive decline.

But even with these early results, the study marks a major step forward.

“What makes this research particularly significant is how we can identify those at risk using a simple overnight EEG recording,” said Haghayegh. “This could completely change how we approach dementia prevention.”

Combining spectral and multivariate EEG features yields the highest accuracy for predicting future neurodegeneration from sleep data. (CREDIT: Shahab Haghayegh, et al.)

A New Way to See the Brain

Science has long searched for better ways to understand how the brain ages. MRI scans can show damage after it happens. Blood tests can give clues about disease markers. But until now, few tools could spot the quiet, early signs of decline while the brain still seems healthy.

This AI-powered EEG tool may be the answer. It gives doctors a view into how the brain functions in real time—and in a natural state of sleep.

It’s non-invasive. It’s easy to use. And it might give people the head start they need to keep their minds sharp for years to come.

Note: The article above provided above by The Brighter Side of News.


Like these kind of feel good stories? Get The Brighter Side of News' newsletter.


Mac Oliveau
Mac OliveauScience & Technology Writer

Mac Oliveau
Science & Technology Writer | AI and Robotics Reporter

Mac Oliveau is a Los Angeles–based science and technology journalist for The Brighter Side of News, an online publication focused on uplifting, transformative stories from around the globe. Passionate about spotlighting groundbreaking discoveries and innovations, Mac covers a broad spectrum of topics—from medical breakthroughs and artificial intelligence to green tech and archeology. With a talent for making complex science clear and compelling, they connect readers to the advancements shaping a brighter, more hopeful future.