Sensemaking for mobile health

Mobile health (mHealth) leverages the power and ubiquity of mobile and cloud technologies to support patients and clinicians in monitoring and understanding symptoms, side effects and treatment outside the clinical setting; thereby closing the feedback loops of self-care, clinical-care, and personal-evidence-creation. However, to realize this promise, we must develop new data capture, processing and modeling techniques to convert the digital exhaust emitted by mobile phone use into behavioral biomarkers. This calls for a modular layered sensemaking framework in which low level state classifications of raw data (e.g., estimated activity states such as sitting, walking, driving from continuous accelerometer and location traces), are used to derive mid-level semantic features (e.g., total number of ambulatory minutes, number of hours spent out of house), that can then be mapped to particular behavioral biomarkers for specific diseases (e.g., chronic pain, GI disfunction, MS, fatigue, depression, etc). The techniques needed to derive these markers will range from simple functions to machine learning classifiers, and will need to fuse diverse data types, but all will need to cope with noisy, erratic data sources. We are working to build an open architecture and community to speed the rate and robustness of innovation in this space, both academic and commercial (http://openmhealth.org).