Recent work has demonstrated that sensors placed in the home environment can be used to develop algorithms that infer context, such as activities of daily living. Context-aware applications can then be created that may help people stay healthy, active, and safe in their homes as they age. Although automatic detection of context from home sensors is an active research area, some critical aspects of the “human-centric” side of creating and deploying these home health systems related to sensor installation, algorithm training, and error recovery have been largely ignored by the research community. Using an example of a medication adherence system, and motivated by some pilot experiments in which we had subjects self-install sensors in a home, we set out twelve questions that we encourage context-aware application developers (as well as machine learning researchers providing the context detection algorithms) to ask themselves as they go forward with their research. We argue that these human-centric questions directly impact not only important aspects of human–computer interface design but also the appropriateness of the selection of specific context detection algorithms.
Context, computing, home, health technology, activity recognition, pattern recognition.