The findings were published in Frontiers in Medicine and Frontiers in Computational Neuroscience.
By integrating temporal convolutional networks with LSTM networks, researchers developed an AI program that analyzes EEG signals to accurately differentiate between healthy and sick individuals.
Can Distinguish Healthy From Sick With 80 % Certainty
The approach attained above 80 % accuracy when differentiating Alzheimer's, frontotemporal dementia, and healthy cohorts. An interpretable AI method was employed to reveal EEG signal components influencing diagnosis, aiding clinicians in understanding the system's reasoning.
In their second study, a compact, efficient AI model (under 1MB) was created to protect patient confidentiality. Federated learning enables collaborative AI training across healthcare providers without data sharing. This privacy-preserving model still reached over 97 % accuracy.
Traditional machine learning models often lack transparency and are challenged by privacy concerns. Our study aims to address both issues.
Muhammad Hanif, Associate Senior Lecturer, Informatics, Örebro University
AI Detects Patterns in the Brain’s Electrical Signals
Researchers successfully integrated diverse brain signal analysis techniques. AI, by separating EEG signals into alpha, beta, and gamma frequencies, detects dementia-related patterns.
The algorithms are capable of spotting enduring signal alterations and discerning slight diagnostic variations. Furthermore, the interpretable AI method guarantees the system is not a "black box" anymore; it explicitly reveals the rationale underpinning its choices.
Research indicates AI's potential as a fast, affordable, and secure tool for early dementia detection. EEG, a basic and economical technique used in primary care, paired with AI models on portable devices, enables broader healthcare applications, spanning specialist clinics to in-home assessments.
The AI Test Could Be Used at Home in the Future
Early diagnosis is essential for implementing proactive measures that slow disease progression and improve quality of life. If solutions like this are fully implemented, it could ease the burden for everyone involved – patients, care staff, relatives, and healthcare professionals.
Muhammad Hanif, Associate Senior Lecturer, Informatics, Örebro University
“We plan to continue the research by expanding to larger and more diverse datasets, exploring more EEG features, and including other types of dementia such as vascular dementia and Lewy body dementia. At the same time, we will use explainable AI and ensure strict protection of patient data,” concludes Muhammad Hanif.
Journal References:
Khan, W., et al. (2025). An explainable and efficient deep learning framework for EEG-based diagnosis of Alzheimer’s disease and frontotemporal dementia. Frontiers in Medicine. DOI:10.3389/fmed.2025.1590201. https://www.frontiersin.org/journals/medicine/articles/10.3389/fmed.2025.1590201/full.
Umair, M., et al. (2025). Privacy–preserving dementia classification from EEG via hybrid–fusion EEGNetv4 and federated learning. Frontiers in Computational Neuroscience. DOI:10.3389/fncom.2025.1617883. https://www.frontiersin.org/journals/computational-neuroscience/articles/10.3389/fncom.2025.1617883/full.