Integrating objective health measurement using sensors, devices and pervasive computing in large-scale surveys

While large-scale population studies provide a wealth of insight and knowledge about the health and wellbeing of the aging population, they typically rely on self-report which has been found to be unreliable, especially among older adults. In addition, the assessment strategy usually occurs sporadically, spaced years apart to reduce patient and investigator burden. Finally, the data itself is not fully ecologically relevant being prone to test situation biases. To overcome these shortcomings of self-report and procedural limitations many new developments using pervasive computing and continuous remote sensing strategies, incorporating high dimensional (“big data”) analytics show great promise for transforming health data capture and follow-up. By assessing health and behavior continuously, objectively and longitudinally, it becomes possible to generate more robust models on the inter-relationships between health and behavior. This review describes the various behaviors and parameters that can be collected via continuous assessment and the devices and assessment strategies that are used to capture key behaviors. Using the framework of wellbeing, we review strategies to assess behaviors that fall into three key categories of wellbeing. These include physical and physiological function, cognitive and intellectual wellbeing, and social behaviors and function. Thus, specific behaviors that can be assessed objectively and more continuously include body composition or weight (an example of a basic physiologic measure), medication adherence (an example of an everyday cognitive-functional task as well as an important medical outcome), and time out-of-home (and example of a measure of social engagement with the world). Devices and assessment strategies that are used to capture these key behaviors include an array of wearable devices, in-home sensor platforms, internet based surveys, computer tracking software, and “smart” devices. We review the applicability of these data collection methods to Health and Retirement Study and give suggestions for future avenues of research.