A technology that can screen for mild cognitive impairment (MCI), a pre-dementia stage, with 85% accuracy using artificial intelligence (AI) has been developed by domestic researchers.

The research team led by Dr. Park Young-jin of the Auditory Cognitive Brain Function Diagnosis Research Team at the Korea Electrotechnology Research Institute (KERI) announced on the 29th that they had successfully conducted a demonstration targeting the elderly using wearable devices and AI speech and brain wave analysis technology, and had completed domestic and international patent applications.

Mild cognitive impairment is a condition in which memory and cognitive function decline but daily life is possible, and there is a high possibility of developing dementia.

On the other hand, the current screening method requires a visit to a dementia care center and a written and questionnaire test, which makes it difficult to access and enable early screening.

To overcome these limitations, the KERI research team developed a diagnostic technology that involves wearing a neckband-type wearable device and performing simple problem-solving and speaking tasks while looking at a monitor screen.

Voice and brainwave data collected during the examination process are transmitted to the cloud and analyzed by multimodal AI.

In particular, considering the pronunciation characteristics, dialects, and hearing impairment of the elderly in Korea, we developed our own voice recognition system with an accuracy of over 97%, and increased reliability by combining brainwave data to address areas where speech alone was insufficient.

The research team is located in Seoul & MiddoA pilot study was conducted on 90 elderly individuals (25 with mild cognitive impairment and 65 healthy individuals) at a dementia support center and welfare center in the Ansan area. The results showed a sensitivity of 72%, a specificity of 90.8%, and an overall accuracy of 85%. This is a very high level for an early detection technology targeting elderly individuals who are still able to perform daily activities.

Above all, the screening process is simple, requiring only an average of 15 questions to screen patients, and repeated testing is possible without time or location constraints. The research team anticipates that this technology will be installed in everyday settings such as senior welfare facilities, public health centers, and government offices, enabling "early diagnosis of dementia in everyday life."

Dr. Park Young-jin emphasized, “The annual dementia progression rate of normal people is 115%, and the 5th Comprehensive Dementia Management Plan (2026-2030), which will be announced soon, should include an early screening and treatment strategy based on advanced science and technology.”

Going forward, the research team plans to expand the scope of diagnosis, accumulate and enhance AI data, and establish a local government-linked treatment system. Through these efforts, the team aims to contribute to lowering the prevalence of dementia and alleviating the national health burden.

According to the Ministry of Health and Welfare, the number of dementia patients and mild cognitive impairment patients in Korea is projected to reach 970,000 and 2.98 million by 2025, respectively. By 2030, these numbers are projected to exceed 1.21 million and 3.68 million, respectively, indicating that the social impact of AI-based early diagnosis technology is expected to grow even further.