Estimating cognitive function in elderly people using information from outdoor walking

Abstract

The rapid increase in the aging population of Japan is becoming a serious social concern, and the number of elderly individuals with dementia is also increasing. The Ministry of Health, Labour and Welfare of Japan has reported that the elderly account for 4.62 million individuals, of which approximately 4 million have mild cognitive impairment (MCI). However, monofunctional disorders, such as those in individuals with MCI, can be treated, with patients recovering 44% of their abilities 2 years after treatment, thereby suggesting that early detection and treatment of dementia is important. It has been reported that individuals who walk slowly or have experienced a significant decline in walking speed with age have a higher risk of developing dementia. In this study, to study movement in individuals aged ≥ 60 years, we focused on walking, a basic activity of daily living. We proposed and evaluated novel methods to estimate cognitive function. Acceleration and angular velocity sensors were attached to the waists of 20 elderly participants who were asked to walk outdoors ordinarily for 5-10 min, during which acceleration and angular velocity were measured. The similarity, standard deviation, and period of the stride were determined from the acceleration waveform and angular velocity waveform during walking. These were used as independent variables, and multiple regression analysis was performed using the Mini-Mental State Examination (MMSE) score as a dependent variable. An MMSE score estimation equation was constructed. The relationship between the estimation formula and the actual test value was R2 = 0.773 (P <0.01), which was good. As a result of cross-validation, the root mean square (RMS) error is low and the error is neither fixed nor proportional. Using the body acceleration and angular velocity information when walking outdoors, we built a very accurate formula for estimating the MMSE score.