Two UNMC neurological sciences faculty joined a national team of other leading neurologists and dementia researchers in publishing new research to advance the use of artificial intelligence to diagnose challenging dementia cases.
Daniel Murman, MD, and Olga Taraschenko, MD, PhD, of the UNMC Department of Neurological Sciences, were co-authors of the journal article “AI-based differential diagnosis of dementia etiologies on multimodal data.”
The research, which was published in the journal Nature Medicine in July, developed a diagnostic computer model drawn from medical records from 51,269 participants in anonymous datasets that included exam findings, test results and magnetic resonance imaging scans. Those records were collected from nine medical databases covering both common and rare dementias, including Alzheimer’s dementia and Lewy body dementia.
The goal of the project was to develop a diagnostic tool that could assist practitioners who are rarely exposed to dementia to triage the patient referrals to subspecialty clinics.
According to the publication, a comparison of 100 randomly selected cases showed that “neurologist assessments augmented by our AI model exceeded neurologist-only evaluations by 26.25%.”
Drs. Murman and Taraschenko were among the neurologists who were presented those 100 cases for review and diagnosis.
The study, which built on earlier research involving the UNMC neurologists, said the computer model has “the potential to be integrated as a screening tool for dementia in clinical settings and drug trials.”
Dr. Taraschenko said: “It was exciting to be able to contribute to the training of this phenomenal tool by sharing our clinical wisdom.”
Dr. Taraschenko, an associate professor, chief of the Comprehensive Epilepsy Program and director of the Autoimmune Seizure Laboratory, said the project expanded on earlier work that created a machine learning framework to help distinguish people with normal cognition from those with Alzheimer’s dementia.
The latest AI algorithm, she said, allows the accurate prediction of a dementia diagnosis even if certain parts of the diagnostic workup are missing by applying the information from other complete cases.
In everyday clinical practice, the datasets are commonly “imperfect,” Dr. Taraschenko said.
In the future, Dr. Murman, professor and director of the Behavioral Neurology Division, said, this type of diagnostic software could be integrated into electronic medical records to assist providers in real time.
By combining clinical and test information, the UNMC neurologists said, physicians can diagnose a specific type of dementia or provide timely referrals to specialists.
awesome accomplishment
So neat! Congrats Drs. Taraschenko & Murman!
Thank you for your continued work in serving the community and pursuing the research.
Very neat research, Dr Taraschenko and Dr Murman!
Great work! I wonder if we can add AI retinal imaging to early diagnosis and screening, as we are working in that area…?