Researchers at Mayo Clinic are leveraging artificial intelligence (AI) and machine learning technology to analyze electroencephalogram (EEG) tests more swiftly and accurately, helping neurologists detect early indicators of dementia in data that is often overlooked.
EEGs, a well-established method that involves attaching multiple electrodes to the scalp to track brain activity, are primarily used to identify epilepsy. The results are interpreted by neurologists and specialists skilled at recognizing patterns in the complex waveforms produced by the test.
A recent study published in Brain Communications highlights how the Mayo Clinic Neurology AI Program (NAIP) utilizes AI to expedite the analysis process and alert specialists to unusual patterns that might be missed by the human eye. This technology has the potential to assist physicians in distinguishing various causes of cognitive decline, such as Alzheimer’s disease and Lewy body dementia. The findings suggest that EEGs, which are more accessible, affordable, and less invasive compared to other brain health assessments, may serve as a viable tool for early detection of cognitive issues in patients.
“These brain waves hold substantial medical information regarding brain health within the EEG,” explains Dr. David T. Jones, the study’s lead author, a neurologist, and director of NAIP. “It’s established that these waves tend to slow down and exhibit different characteristics in individuals facing cognitive difficulties. Our investigation aimed to determine if we could accurately quantify this type of slowing with AI support.”
For this study, researchers gathered EEG data from over 11,000 patients at Mayo Clinic throughout a decade. They employed machine learning and AI techniques to distill intricate brain wave patterns into six key features, training the model to eliminate irrelevant data and focus on patterns indicative of cognitive disorders such as Alzheimer’s.
“It was impressive how technology facilitated the rapid extraction of EEG patterns compared to traditional dementia evaluation methods like bedside cognitive assessments, fluid biomarkers, and brain imaging,” remarks Dr. Wentao Li, a co-first author of the study who participated in the project as a clinical behavioral neurology fellow at Mayo Clinic.
“Currently, we often quantify patterns in medical data based on expert assessments. But how do we confirm the presence of those patterns? It relies entirely on the expert’s judgment,” Dr. Jones states. “With AI and machine learning, we can observe details invisible to the expert and accurately quantify what they do see.”
Utilizing EEGs for identifying cognitive issues won’t necessarily replace other diagnostic tools like MRIs or PET scans, but AI technology may transform EEGs into a more cost-effective and accessible option for early diagnosis, particularly in areas lacking specialty care facilities, as noted by Dr. Jones.
“Early recognition of memory issues is crucial, even before they become apparent,” Dr. Jones asserts. “A timely and accurate diagnosis enables us to provide patients with the best possible care and treatment options. The approaches we’re exploring could present a more affordable means of identifying early stages of memory decline or dementia compared to current methods like spinal fluid analysis, brain glucose imaging, or cognitive tests.”
Further testing and validation of these tools will require several more years of research, according to Dr. Jones. Nonetheless, he believes that this research proves viable pathways exist for integrating clinical data into practice to support the development and innovation of new models, enhance existing evaluation methods, and expand this knowledge beyond Mayo Clinic.
“This initiative underscores the importance of collaborative efforts across disciplines to propel healthcare research that leverages technology,” states Dr. Yoga Varatharajah, co-first author of the paper and a NAIP research collaborator during this work.
Support for this research comes from various sources, including the Edson Family Fund, the Epilepsy Foundation of America, the Benjamin A. Miller Family Fellowship in Aging and Related Diseases, Mayo Clinic Neurology Artificial Intelligence Program, the National Science Foundation (Award No. IIS-2105233), and the National Institutes of Health, including grant UG3 NS123066.
