August 8, 2022
Keio University School of Medicine
FRONTEO, Inc.
In Japan, where the population is aging rapidly, the number of people with dementia continues to increase, making dementia countermeasures one of the country's critical social issues. The diagnosis of dementia is typically made through a combination of medical history interviews, imaging tests, and multiple cognitive function tests that measure memory, calculation skills, and more. However, these tests present several challenges: they require a high level of expertise, necessitate training for the healthcare professionals who administer them, and are time-consuming.
Project Professor Taishiro Kishimoto and his team at the Hills Future Preventive Medicine and Wellness Joint Research Program, Keio University School of Medicine, together with FRONTEO, Inc., have developed a "conversational AI program for supporting dementia diagnosis" using natural language processing (NLP). This program enables AI to detect the possibility of dementia based on free-form conversations between elderly individuals and healthcare professionals.
In this study, the researchers successfully determined the presence of dementia from free-form conversations, without using specific tasks, with an accuracy of 0.90. This program is simple to use and can reduce the burden on both healthcare professionals and patients. Furthermore, it is expected to be put into practical use for applications such as screening tests, as it is a technology that can avoid the "learning effect"—a problem with conventional cognitive function tests where test accuracy declines because subjects remember the test content through repetition.
The results of this research were published in the journal Scientific Reports on August 3, 2022 (UK time).
For the full press release, please see below.