Imagine a future where a simple, non-invasive test could accurately diagnose different types of dementia, revolutionizing how we care for millions of affected individuals. But here's where it gets controversial: what if this breakthrough relies on artificial intelligence, a technology often met with skepticism in healthcare? Engineers at Florida Atlantic University (FAU) have done just that, developing an AI model that uses EEG brainwave analysis to distinguish between Alzheimer’s disease (AD) and frontotemporal dementia (FTD) with remarkable precision. And this is the part most people miss: their approach not only identifies the type of dementia but also gauges its severity, offering a comprehensive tool for clinicians.
Dementia, a group of disorders that progressively impair memory, thinking, and daily functioning, affects millions worldwide. By 2025, an estimated 7.2 million Americans aged 65 and older will have Alzheimer’s, the most common form. Frontotemporal dementia, though rarer, is the second leading cause of early-onset dementia, often striking individuals in their 40s to 60s. While both diseases damage the brain, they do so differently: AD primarily affects memory and spatial awareness, while FTD targets behavior, personality, and language. This overlap in symptoms frequently leads to misdiagnosis, making accurate differentiation not just a scientific challenge but a clinical necessity.
Traditional diagnostic tools like MRI and PET scans are effective but costly, time-consuming, and require specialized equipment. Electroencephalography (EEG), on the other hand, is portable, non-invasive, and affordable, measuring brain activity through sensors across various frequency bands. However, EEG signals are often noisy and highly variable between individuals, complicating analysis. Even with machine learning, distinguishing between AD and FTD has remained elusive—until now.
FAU researchers from the College of Engineering and Computer Science have created a deep learning model that analyzes both frequency- and time-based brain activity patterns to detect and evaluate AD and FTD. Published in Biomedical Signal Processing and Control, their study reveals that slow delta brain waves are a key biomarker for both conditions, particularly in the frontal and central brain regions. In AD, brain activity disruption is more widespread, affecting additional regions and frequency bands like beta, indicating more extensive damage. These findings explain why AD is often easier to detect than FTD.
The model achieved over 90% accuracy in distinguishing individuals with dementia from cognitively normal participants. It also predicted disease severity with relative errors of less than 35% for AD and 15.5% for FTD. By employing feature selection, the researchers improved the model’s specificity—its ability to correctly identify healthy individuals—from 26% to 65%. Their two-stage approach—first detecting healthy individuals, then differentiating between AD and FTD—achieved 84% accuracy, ranking among the best EEG-based methods to date.
But here's the controversial part: the model combines convolutional neural networks and attention-based LSTMs, technologies some critics argue are 'black boxes' in healthcare. However, the researchers addressed this by using Grad-CAM to visualize which brain signals influenced the model’s decisions, providing transparency for clinicians. This approach not only identifies dementia type and severity but also offers insights into how brain activity evolves and which regions and frequencies drive diagnosis—something traditional tools rarely capture.
“Our study’s novelty lies in extracting both spatial and temporal information from EEG signals using deep learning,” said Tuan Vo, the study’s first author and a doctoral student at FAU. “This allows us to detect subtle brainwave patterns linked to Alzheimer’s and frontotemporal dementia that might otherwise be missed. Our model doesn’t just diagnose—it estimates severity, giving clinicians a fuller picture of each patient’s condition.”
The findings also highlight that AD tends to be more severe, impacting a broader range of brain areas and resulting in lower cognitive scores, while FTD’s effects are more localized to the frontal and temporal lobes. These insights align with previous neuroimaging studies but add new depth by demonstrating how these patterns appear in EEG data—an inexpensive, noninvasive tool.
“Alzheimer’s disrupts brain activity more broadly, especially in the frontal, parietal, and temporal regions, while frontotemporal dementia primarily affects the frontal and central areas,” explained Hanqi Zhuang, co-author and professor at FAU. “This difference explains why Alzheimer’s is often easier to detect. However, our work shows that careful feature selection can significantly improve FTD differentiation.”
Overall, the study demonstrates how deep learning can streamline dementia diagnosis by combining detection and severity assessment into one system, reducing lengthy evaluations and providing clinicians with real-time tools to track disease progression. But here’s a thought-provoking question: As AI becomes more integrated into healthcare, how do we balance its potential for transformative breakthroughs with ethical concerns about transparency and patient trust?
“This work exemplifies how merging engineering, AI, and neuroscience can tackle major health challenges,” said Stella Batalama, dean of FAU’s College of Engineering and Computer Science. “With millions affected by dementia, breakthroughs like this pave the way for earlier detection, personalized care, and interventions that can truly improve lives.”
The study’s co-authors include Ali K. Ibrahim, an assistant professor, and Chiron Bang, a doctoral student, both from FAU’s Department of Electrical Engineering and Computer Science.
About FAU’s College of Engineering and Computer Science:
FAU’s College of Engineering and Computer Science is globally recognized for its innovative research and education in fields like computer science, AI, electrical engineering, and biomedical engineering. Supported by grants from the NSF, NIH, and other agencies, the college offers cutting-edge degree programs, including Florida’s first Master of Science in AI and new degrees in data science and analytics. For more information, visit eng.fau.edu.
About Florida Atlantic University:
Serving over 32,000 students across six campuses in Southeast Florida, FAU is a leader in research and social mobility. Recognized as an R1 institution and an Opportunity College, FAU is ranked among the Top 100 Public Universities by U.S. News & World Report and praised for its role in fostering upward mobility. Learn more at www.fau.edu.
