Early detection of health changes among older adults is the key to maintaining health, independence, and function. Non-wearable sensors such as depth cameras, motion sensors (passive infrared, PIR) and bed sensors (based on ballistocardiography) are able to detect changes in gait activity and sleep, and have emerged as a possible solution for early detection of health changes. Since 2005, our interdisciplinary research team has investigated, developed and tested a state of the art sensor monitoring system for older adults at TigerPlace, a unique eldercare facility in Columbia, MO1.
Analyzing and acting upon sensor data remains a challenge for clinicians due to data variety (many sensor types) and velocity (continuous monitoring). To view the context of the health alerts sent by the monitoring system, clinicians currently use a secure interface to review multiple data displays that may take 7 minutes per alert. To save clinicians’ time and make alerts easier to interpret, we are investigating a new knowledge-generation methodology based on linguistic summaries as a tool to provide more meaningful and easier to interpret alerts to clinicians.
While certain nonspecific behaviors were shown to be linked to diseases in the elderly and non-wearable sensors can capture those behaviors, more specific information is needed for our monitoring system. As an initial step in creating linguistic summary alerts, our team conducted a survey of clinicians to determine which signs and behaviors captured by the monitoring system they find most relevant in evaluating and treating health conditions among older adults. The results of this survey will guide the development of linguistic summary methods.
Popescu, M., Craver, A., Phillips, L., Koopman, R., Alexander, G., Despins, L., & Rantz, M., November 7, 2017, “Linking resident behavior to health conditions in an Eldercare monitoring system“, AMIA 2017 Annual Symposium, Washington, DC, national. [paper] View the PDF