AI system predicts dementia by identifying brain age
A group of researchers led by Ajou University Medical Center (AUMC) and the University of Pittsburgh School of Medicine said it was possible to predict dementia in advance by looking at the brain age.
Samsung Medical Center, Inha University Hospital, Pusan National University Hospital, and Chonnam National University Hospital also participated in the study.
According to AUMC, elderly dementia due to aging has emerged as a serious social problem worldwide, but the development of drugs or non-pharmacological methods that show clear effects is still unavailable.
As a result, hospitals are focusing on correcting risk factors such as physical disease and lifestyle and using early drug treatment before the onset of dementia.
Public health centers, hospitals, and communities in Korea are conducting dementia screening tests.
However, conventional dementia screening tests have limitations in screening the various socio-demographic factors related to dementia, and there was a high unmet need for developing a new predictive tool, the researchers said.
To resolve this issue, the research team, led by Professors Hong Chang-hyung, Son Sang-joon, and Roh Hyung-woong at AUMC, and Helmet T. Karim and Howard Aizenstein at the University of Pittsburgh, predicted brain age by analyzing brain magnetic resonance imaging (MRI) data of 650 Koreans obtained through the "Chronic Cerebrovascular Disease Biobank" of the Korea Biobank Project using artificial intelligence.
The team used an AI model developed through joint research by Professors Son and Karim.
The two professors used the learning results for age prediction through the volume change pattern of each part of the brain based on brain MRI data of a person without dementia.
As a result, if the brain age predicted by the AI appeared to be higher than the actual age, the probability of that person developing dementia after two years was about 150 percent higher than those with a matching brain age.
"This study is significant because it confirmed the possibility of early prediction of dementia progression through brain image analysis through AI," Professor Son said. "I hope that this study will help open a new chapter in early detection of dementia, as the number of dementia patients is expected to increase rapidly along with a super-aging society."
The research, titled "Independent replication of advanced brain age in mild cognitive impairment and dementia: detection of future cognitive dysfunction," was published in Molecular Psychiatry on Dec. 8.