Lead researcher Park Young-jin and the auditory cognitive brain function diagnostics research team at the Korea Electrotechnology Research Institute (KERI) developed an "artificial intelligence (AI)-based speech and brainwave analysis technology" that can conveniently screen for mild cognitive impairment using wearable devices in daily life.
Mild cognitive impairment is a pre-dementia stage in which memory or cognitive function is diminished but daily life remains manageable. If high-risk groups for mild cognitive impairment can be identified early in the community and treated proactively, the number of people with dementia and its prevalence associated with entry into a super-aged society can be reduced, easing the burden on national finances and significantly cutting personal and social expense.
In response, the research team unveiled a technology that can identify high-risk groups for mild cognitive impairment early by simply wearing a wearable device and answering prompts that elicit speech. Test subjects wear an easy-to-use neckband-type wearable device shaped like earphones and perform five speech/EEG collection tasks using audio- and screen-based prompts via a monitor. These include describing pictures, everyday Q&A, and explaining procedures.
Speech and brainwave data collected by the device are stored in real time in the cloud, and multimodal AI technology determines whether someone is in a high-risk group for mild cognitive impairment. The AI was jointly developed by the Korea Electrotechnology Research Institute (KERI) and the Electronics and Telecommunications Research Institute (ETRI) through training on the voices and text data of older adults in Korea.
In a demonstration involving 90 older adults, sensitivity—the rate of correctly identifying patients as positive—was 72%, and specificity—the rate of correctly identifying people without the disease as negative—was 90.8%. The overall screening accuracy for mild cognitive impairment based on this was 85%.
Lead researcher Park Young-jin said, "Among healthy people 65 and older, the annual progression rate to dementia is only 1–2%, but among patients with mild cognitive impairment, 10–15% progress to dementia each year, and over six years as many as 80% progress to dementia," adding, "The soon-to-be-announced "5th Comprehensive Dementia Management Plan" should include national-level technical support and plans to lower dementia prevalence, such as screening high-risk groups for mild cognitive impairment using advanced science and technology and promoting proactive treatment."
The research team has filed domestic and international patent applications for the technology and plans to expand the target population for diagnosis, accumulate AI data, and advance the technology based on collaboration with community welfare centers and Dementia Relief Centers, as well as establish a treatment system linked with local governments.