Researchers at Örebro University have created two powerful AI models that can detect dementia by analyzing the brain’s electrical activity.
These models can tell the difference between healthy individuals and those with dementia, including Alzheimer’s disease, with a high level of accuracy.
“Early diagnosis is crucial in order to be able to take proactive measures that slow down the progression of the disease and improve the patient’s quality of life,” says Muhammad Hanif, a researcher in informatics at Örebro University.
In their first study, titled “An explainable and efficient deep learning framework for EEG-based diagnosis of Alzheimer’s disease and frontotemporal dementia,” the team combined two advanced AI techniques—temporal convolutional networks and LSTM networks.
These methods process EEG (electroencephalogram) signals from the brain. The model was able to classify whether a person had Alzheimer’s, frontotemporal dementia, or was healthy with more than 80% accuracy.
This AI system is not only accurate but also explainable. It can show which parts of the brain’s electrical signal helped it reach a diagnosis, making it easier for doctors to trust and understand the results.
In the second study, “Privacy–preserving dementia classification from EEG via hybrid–fusion EEGNetv4 and federated learning,” researchers created a small AI model that protects privacy and is highly efficient.
It is less than one megabyte in size and uses federated learning—a technique that allows hospitals and clinics to train the AI together without sharing private patient data. Even with this extra privacy protection, the model achieved over 97% accuracy. This study appeared in Frontiers in Computational Neuroscience.
“Traditional machine learning models often lack transparency and are challenged by privacy concerns. Our study aims to address both issues,” says Hanif, who is also an associate senior lecturer at the university.
The models work by analyzing patterns in the brain’s electrical activity. EEG signals are broken down into different frequencies—such as alpha, beta, and gamma waves—and the AI can recognize changes that are linked to dementia. These changes include long-term signal shifts and subtle differences between various types of dementia.
Thanks to explainable AI, these models no longer work like a “black box.” Instead, they offer clear explanations for how they reach their conclusions, which makes them more useful in clinical settings.
The team believes this technology could lead to fast, low-cost, and privacy-safe tools for early dementia diagnosis. Because EEG is already simple and affordable, these AI tools could be used in regular clinics or even at home. They are small enough to run on portable devices, which could bring brain health screening into more communities.
“If solutions like this are fully implemented, it could ease the burden for everyone involved—patients, care staff, relatives and health care professionals,” Hanif notes.
The research was carried out in collaboration with institutions in the UK, Australia, Pakistan, and Saudi Arabia. The next steps include testing the AI on more people from different backgrounds, adding features from more types of dementia like vascular and Lewy body dementia, and continuing to use explainable AI while keeping patient data safe.
If you care about dementia, please read studies about the power of healthy fats for brain health, and wild blueberries can benefit your heart and brain..
For more health information, please see recent studies about how eating nuts can affect your cognitive ability, and brain foods nourish your mind to outsmart dementia.
The studies are published in Frontiers in Medicine and Frontiers in Computational Neuroscience.
Copyright © 2025 Knowridge Science Report. All rights reserved.
