A research team in South Korea has published the design for a three year study, called DETECT, that will follow community dwelling older adults already diagnosed with mild cognitive impairment, combining wrist actigraphy, sweat patch biochemical sampling and indoor environmental sensors with annual home visits, according to the protocol published in BMC Geriatrics on June 9, 2026. The stated aim is to identify factors associated with the transition from mild cognitive impairment to dementia and to build machine learning based prediction models from that combined data, rather than relying only on clinic based testing.
Mild cognitive impairment is widely treated in the research literature as a high risk state that can precede dementia, and the protocol argues that prior studies have leaned on single clinic visits rather than continuous, lived context measurement. The home based instruments described include a wearable sweat patch, with participants instructed on how to keep it in place through bathing and showering, intended to capture biochemical signals outside a clinic setting over the follow up period. The published protocol also lists a structured table of variables spanning biological, psychological and social domains, including standardized instruments such as the Fried frailty phenotype questionnaire, that will be collected at each annual visit.
What has been published so far is a protocol, not results. No data on how many participants converted to dementia, nor on how accurately any model predicted that transition, has been reported, because the three year follow up period described in the paper has not yet concluded as of this writing. Whether combining actigraphy, sweat patch chemistry and indoor sensor data actually improves prediction of dementia transition beyond standard cognitive and clinical assessment is the question the study is designed to test, not a finding it has already produced.