Lifestreams: a modular sense-making toolset for identifying important patterns from everyday life

Smartphones can capture diverse spatio-temporal data about an individual; including both intermittent self-report, and continuous passive data collection from onboard sensors and applications. The resulting personal data streams can support powerful inference about the user's state, behavior, well-being and environment. However making sense and acting on these multi-dimensional, heterogeneous data streams requires iterative and intensive exploration of the datasets, and development of customized analysis techniques that are appropriate for a particular health domain. Lifestreams is a modular and extensible open-source data analysis stack designed to facilitate the exploration and evaluation of personal data stream sense-making. Lifestreams analysis modules include: feature extraction from raw data; feature selection; pattern and trend inference; and interactive visualization. The system was iteratively designed during a 6-month pilot in which 44 young mothers used an open-source participatory mHealth platform to record both self-report and passive data about their diet, stress and exercise. Feedback as participants and the study coordinator attempted to use the Lifestreams dashboard to make sense of their data collected during this intensive study were critical inputs into the design process. In order to explore the generality and extensibility of Lifestreams pipeline, it was then applied to two additional studies with different datasets, including a continuous stream of audio data, self-report data, and mobile system analytics. In all three studies, Lifestreams' integrated analysis pipeline was able to identify key behaviors and trends in the data that were not otherwise identified by participants.