MIT Media Lab Human Dynamics

Friends and Family Open Sensing Network

ID3 is also collaborating with the MIT Media Lab Human Dynamics Group (http://hd.media.mit.edu) on the Funf project (http://funf.org) to develop and test trust frameworks for the trusted sharing of mobile phone data. It is our shared belief that mobile platforms provide an enormous amount of valuable data about people’s health, affinities, purchasing behavior, and mobility, but at the same time raise significant challenges about preserving data. We are working closely with Funf researchers to address this problem so that people can have control over how their mobile data are to be shared.

Analysis for Mobile Health Monitoring

Reality Analysis, a term coined by Sandy Pentland, involves collecting and analyzing the trail of data left behind as we go through our daily lives. A DARPA-funded project applies Reality Analysis to psychological health monitoring of soldiers back from the battlefield, in an effort to detect early signs of PTSD and depression. A key component of this project is a collaboration between MIT and Cogito Health, a company launched off Human Dynamics’ research.

The project targets development of enabling Trust Framework technology capabilities and governance mechanisms, in support of a mobile health use case that can provide multiple layers of trust, through the use of multiple, mutually reinforcing mechanisms. These include contractual governance and enforcement through binding legal agreements, anonymized IDs that cannot be traced back to device IDs, user selection of sharing levels, privacy technologies for anonymized aggregation, use of secure OAuth tokens for separation of authentication from service delivery, and encrypted storage of sensitive data.

Drawing on Funf sensor data collected from Android phones, data sets are scored and rolled up into DSM-IV constructs for PTSD and depression. In one visualization, a “health triangle” shows Sociability, Activity, and Focus. If the participant agrees to share his or her data anonymously into an aggregate data pool, a comparison is made against the average participant.