Embedded assessment offers a basic strategy for addressing a broad range of health issues, across stages of illness and at different points in the lifespan. The prototypes shared in this paper illustrate key capabilities of this approach: continuous sensing of objective and subjective health indicators, intuitive feedback offered when and where it is most needed, and facilitation of activities with preventive value. Future products using this approach will most likely include a more comprehensive set of sensing and feedback applications to address a range of health concerns. For example, the cardiovascular sensing and mood reporting in mobile heart health project would be logically combined with the interpersonal measures from the social health platform. The addition of other noninvasive measures (e.g., of glucose and hormonal levels) would allow such systems to adapt to the changing health needs that individuals experience at different points in their lives.

Significant design advances are necessary to develop compelling products from the exploratory prototypes discussed in this paper. Hardware configurations must of course extend from the standalone PC to a range of mobile and wearable interfaces. These systems will require more sensitivity to both geographic and social contexts. Computing-intensive solutions will be required to interpret continuous streams of behavioral and biological data gathered by peripherals.

Perhaps even greater challenges lie in interaction design. We need to display health information in ways that mirror the mental schemas that people use to make sense of health concerns and the rest of their lives. That is, they need to reflect not isolated biological metrics but complex interplays of emotion, cognition, social interaction, and physiology. In addition, far more intuitive interfaces are required for collecting and reflecting health data. Input modalities, ranging from passive physiological sensing to gesturing, will depend on individual preference and contexts. Similar differences apply for feedback: a succinct text message might be ideal in some situations, but in others, musical feedback or a physical nudge will be more effective. Research is needed to develop these basic interaction modalities and to determine their adaptation to individuals, context, and moment-to-moment variability in health status. Advances in data visualization and interaction design will increase the odds that technologies will stick. Ultimately, we want to develop objects that people not only use, but love—ones that invite close attachment as individuals initiate and maintain the often difficult changes required to improve their health.