A new study found that artificial intelligence (AI) can use voice recordings to diagnose Alzheimer’s disease or cognitive decline.
Early stages of Alzheimer’s disease are hard to recognize. The recent developments in AI made it possible for the development of a new, time-saving method that detects Alzheimer’s disease early.
Subtle changes in speech and behavior have been linked with Alzheimer’s disease, but they are hard to detect early.
Dr. Hajjar is the lead author of the study and professor of neurology at UT Southwestern’s Peter O’Donnell Jr. Brain Institute. Dr. Hajjar said, “Our focus was on identifying subtle language and audio changes that are present in the very early stages of Alzheimer’s disease but not easily recognizable by family members or an individual’s primary care physician.”
This new study assessed 206 participants; 114 had mild cognitive decline and 92 had no cognitive decline. Researchers asked participants to describe a piece of art called “The Circus Procession.” This recording was 1-2 minutes long. The AI used this recording to detect Alzheimer’s disease.
Learn from top leaders in the field the skills you need today.
Dr. Hajjar stated, “The recorded descriptions of the picture provided us with an approximation of conversational abilities that we could study via artificial intelligence to determine speech motor control, idea density, grammatical complexity, and other speech features.”
The AI used advanced machine learning and natural language processing tools to assess speech patterns. These tools allowed the AI to determine what parts of speech predicted Alzheimer’s disease or cognitive decline. This allowed the researchers to develop “digital voice biomarkers.” The voice biomarkers that indicate cognitive decline included use of fewer content words, abstract nouns, and syntactic complexity.
So, how do we know that these voice biomarkers work? Researchers performed clinical tests with the patients and compared the tests to the AI’s findings. The tests included neuropsychological testing, MRI scans, and cerebral spinal fluid analysis.
The biomarkers taken from the clinical analyses gave the same diagnoses as the AI’s predictions!
The voice biomarkers were better associated with amyloid load than neuropsychological language tests. This shows that the AI could save valuable time for doctors. Traditional testing for Alzheimer’s disease or cognitive decline usually takes hours. Yet, in this study, the voice recordings used to diagnose Alzheimer’s disease were just 1-2 minutes long.
This new method is promising and would help with early diagnosis and treatment of Alzheimer’s disease. Dr. Hajjar stated, “If confirmed with larger studies, the use of artificial intelligence and machine learning to study vocal recordings could provide primary care providers with an easy-to-perform screening tool for at-risk individuals.”
- Seeley, D. (2023, April 17). AI Can Detect Early Signs of Alzheimer’s in How People Speak, New Study Shows. Dallas Innovates. https://dallasinnovates.com/ai-can-detect-early-signs-of-alzheimers-in-how-people-speak-new-study-shows/
- Hajjar, I., Okafor, M., Choi, J. D., Moore, E., Abrol, A., Calhoun, V. D., & Goldstein, F. C. (2023). Development of digital voice biomarkers and associations with cognition, cerebrospinal biomarkers, and neural representation in early Alzheimer’s disease. Alzheimer’s & dementia (Amst), 15(1), e12393. https://doi.org/10.1002/dad2.12393