Monitoring Everyday Functioning in Normal Aging and Mild Cognitive Impairment in a Smart Environment: An Evaluation of Direct Observation and Data Mining Techniques
Research on everyday functioning has revealed the importance of evaluating activities of daily living (IADL) as an ecologically valid measure of everyday performance. Development of smart environment technologies has allowed researchers to use “real world” scenarios to evaluate everyday functioning. This interdisciplinary study of clinical psychology and computer science disciplines investigated the use of extracted sensor data to track and assess everyday performance of participants within a smart environment. Participants were 28 older adults who completed eight IADL activities (e.g., sweeping and dusting, cooking) in a smart home on the Washington State University campus. Researchers coded these activities and derived a total direction observation score for each participant. Sensor data (e.g., motion sensors, door sensors) gathered from the smart home during the sessions were analyzed using a neural networks machine learning model to determine an overall functional score for each participant. In addition, participants completed the Everyday Problems Test (EPT), a paper and pencil-based measure of everyday functioning as part of a larger battery of cognitive tests. Pearson correlations revealed strong agreement between the sensor data and direct observation scores, as well as significant correlations between direct observation, sensor data, and paper and pencil-based measures of cognition and everyday performance (EPT). Extracted sensor data from a smart environment can provide revealing information about everyday functioning of older adults and may be useful in monitoring changes in everyday activity performance. These smart environments could be used to promote independent living in older adults and delay the need for placement in assisted living facilities.