Definition and Test-Retest Reliability of a Monitoring Method Integrating Accelerometric Actigraphy and Bluetooth Indoor Location Tracking Applied in a Long-Term Residential Unit for Persons With Dementia: Longitudinal Observational Study

Background: Dementia has an impact on the physical activities performed daily in a social context. Sleeping and resting, in general, are also affected by dementia. Monitoring techniques based on miniaturized wearable sensors and on sensorized environments allow for actigraphic recordings and location tracking. The availability of contemporaneous physical activities profile led to quantify, in the social actigraphy approach, the level of correlation between individuals living in the same environment. Objective: This study has two main objectives: (1) to define a methodology for actigraphic recordings, based on wearable accelerometers, and on location tracking, based on Bluetooth wearable technology, and to apply it in a well-defined social context, a long-term care residential unit for people with dementia; and (2) to quantify test-retest reliability of the indices obtained by the monitoring methodology. Methods: Persons with dementia living in the long-term care unit have been equipped with miniaturized wearable sensors, an accelerometer at their dominant wrist, and a Bluetooth beacon at their ankle for 7 days. The raw recordings allowed for computing indices related to physical activity intensities, to the occurrence of walking bouts, to the efficiency of sleep and waking phases, to social interactions between individuals, and to locations preferably occupied. The 7-day session was repeated at short (3 weeks) and long (3 months) terms in order to quantify the test-retest reliability of the indices. Results: Twenty-five persons with dementia were enrolled, 4 of them dropped out, and valid data were obtained, in the different sessions, from 19 to 21 individuals of the recruited group. Control data from 10 age-matched healthy participants were derived from published datasets. As a group, compared with age-matched healthy participants, persons with dementia showed a comparable duration of phases of no activity and of light activity (energy cost lower than 3 metabolic equivalents of tasks [METs]), a relevantly lower duration (-84.3%) of phases of moderate activity (energy cost ranging from 3 to 6 METs), and substantial absence (-100%) of phases of vigorous activity (larger than 6 METs); moreover, daytime and nighttime were characterized by comparable wake and sleep, respectively, efficiency; finally, as to the social interactions, persons with dementia showed a lower correlation of their motor activity profiles (-53.1%). The test-retest reliability was excellent for physical activity indices (intraclass correlation coefficients ranging from 0.76 to 0.98), good for social indices (0.65-0.67), excellent for sleep or wake efficiency (0.74-0.89), and fair for location tracking indices (0.37-0.78). Conclusions: The considered methodology, particularly concerning accelerometry, proved to be feasible, informative, and with a good to excellent test-retest reliability. Interestingly, the methodology clearly identified behaviors, such as wandering, in a minority of individuals inside this study's group of persons with dementia, thus supporting a possible clinical use of the methodology.