Computing Applications News

Making Sense of Real-Time Behavior

Data captured by sensors worn on the human body and analyzed in near real-time could transform our understanding of human behavior, health, and society.
  1. Introduction
  2. Popular Clubs and Events
  3. Author
  4. Footnotes
  5. Figures
Alex Pentland and researchers at MIT's Media Lab
Alex Pentland, right, and a team of researchers with reality-mining devices at Massachusetts Institute of Technology's Media Lab.

As the proliferation of smartphones and networks puts a mobile sensor in many people’s pockets, more intimate wearable sensors are playing a leading role in understanding human behavior to an extent that has not previously been possible.

The sensors can be integrated into a variety of wearable items, such as badges, plastic tags on lanyards, and sticky plasters, but their commonality is in collecting fine-grained data that can be computed to visualize real-time behavior. With in-depth behavioral knowledge, applications can be developed that make the best use of behavioral patterns or monitor behavior to ensure and potentially improve human well being.

Alex “Sandy” Pentland, a professor at Massachusetts Institute of Technology’s Media Lab who specializes in computational social science, suggests real-time collection and analysis of data about people, a discipline he calls “reality mining,” could transform the way we understand ourselves and society (although he adds the caveat that reality mining will raise the bar on data privacy and ownership).

Pentland started research on reality mining in the mid-1990s, leading a team that used large wearable devices to track human behavior. “Reality mining is about understanding people and situations very rapidly,” he explains. “As computational social scientists we can build a better model of how people behave than psychologists or sociologists because we can actually see what people are doing.”

Pentland suggests that one of reality mining’s uses could be to reduce human errors in hospitals. Every day, hospital staff would wear sensor-equipped name badges that capture their tone of voice, body language, and location data. The real-time data would be sent to a central server and analyzed to detect patterns of mistakes, perhaps identifying stress in the tone of voice as a precursor to mistakes.

A similar solution can be used to improve employee productivity. Pentland points out that as corporations strive for greater productivity they often structure staff time to reduce personal conversations. “At more than a dozen companies where we have used wearable sensors, social interaction has been found to be an important element in productivity,” says Pentland. “Staff who socialize trade information about how to do their jobs better. People who are cut off aren’t wise in the ways of the company and don’t have social support.”

Having discovered the business benefits of socializing, companies typically reorganize, structuring rest periods and coffee breaks to support social interaction. Data from this kind of reality mining can also be analyzed to evaluate office layout and recognize desirable changes, such as moving employees or knocking down walls to improve communication.

Pentland acknowledges that such data-capture schemes could face opposition from staff, but says that most buy in if their boss is included and they can view the data. Pentland goes one step further by giving staff and managers control over the data. “It’s spying that people don’t like,” he says. “If their information is open to them and they have control of it, people are generally much happier.”

Back to Top

Popular Clubs and Events

Beyond Pentland’s work, reality mining is reaching the market through companies such as Hitachi Consulting, which is using smart badges on client projects similar to those undertaken by Pentland, and through MIT Media Lab spin-offs. The latter includes Sense Networks, which analyzes mobile location data from cellphones, taxis, cameras, and GPS devices to create a real-time map of popular clubs and events to provide answers to consumers’ questions, such as “Where is everyone this evening?”

Another spin-off is Cogito Health, which is developing Pentland’s research to extract meaning from speech behavior. One application that uses Cogito’s vocal signaling platform is a depression-monitoring service that automatically processes care-management telephone calls and allows healthcare professionals to proactively identify patients who may need support for clinical depression.

An MIT Media Lab spin-off is nTag, a company founded in 2002 that fell into bankruptcy before being acquired by Alliance Tech in March 2009.

Alliance Tech focuses on marketing intelligence at events such as trade shows and conferences. It offers event organizers sensor-laden handheld devices that can be used by event participants. The devices are worn on lanyards and support not only real-time attendee tracking, but also social networking. Conference attendees, for example, can exchange contact information by pressing a button on their nTag device and later access that information on the event’s Web site, or they can be alerted by an nTag vibration if they are close to someone who, according to preselected criteria, they want to meet. For event organizers, real-time data capture and analysis means changes can be made on the fly during an event, the success of the event can be measured, and paperwork can be reduced by using nTag for activities such as attendee feedback.

The nTag solution uses proprietary hardware, but Alliance Tech CEO Art Borrego says future products will move away from such technology. “Why invest in more hardware when we could now use what is in people’s pockets?” Borrego asks. “Our next generation technology will use smartphones and a micro browser.”

As smartphones become the sensor for many reality-mining applications, wearable sensors are likely to prevail in the medical field where the automated analysis of real-time data captured by sensors could prove transformational.

Chris Toumazou, CEO of Toumaz and a professor at the Institute of Biomedical Engineering at Imperial College, London, has developed the technology behind a digital plaster concept that is being used in patient trials at St. Mary’s Hospital in London. Realizing the constraints of existing wireless technologies, such as Bluetooth and Zigbee, particularly their high-power requirements and that any device built using them would be bulky, obtrusive, and possess a short operating life, Toumazou set out to create a low-power technology that could capture data from a body, even if it was ambulatory.

Digital plasters will allow “patients to leave the hospital, but continue to be monitored by healthcare professionals,” says Chris Toumazou.

“We have commercialized ultra low-power wireless systems and signal processing, putting them together on a single chip,” says Toumazou. “We call this the Sensium technology platform to which sensors can be attached. This is the basis of the digital plaster.”

The digital plaster technology, which has been developed by Toumaz but is expected to be licensed and commercialized, includes a plaster or patch that sticks to the body and captures vital sign data from patients. The data is forwarded to a hospital information system, where it is analyzed, interpreted, and delivered to a nurse or doctor.

“Because of the economies of scale of semiconductors we can drive down cost and make digital plasters disposable, avoiding problems such as infection, the need to sterilize plasters, or the need to recharge their batteries,” explains Toumazou. “Ultimately, this technology will allow patients to leave the hospital, but continue to be monitored by healthcare professionals.”

Continuous ambulatory monitoring could also provide important insights into human health as the vital sign data captured in real time from patients could be correlated to show trends in patient health that cannot be readily understood using traditional methods of data capture that are bulky and have limited portability. Correlation could also provide an extra dimension to alerts, as the analysis of multiple, vital sign data could predict adverse patient effects that may occur in hours or even days.

As trials of the digital plaster continue at St. Mary’s Hospital, further applications of the Sensium technology are anticipated, with Toumazou and his team at the Institute of Biomedical Engineering currently working with the technology to create a diabetes management system that, like the digital plaster technology, is expected to be licensed to a commercial partner.

*  Further Reading

Burdett, A., Lakhanpal, A., McPartland, R., Nunn, C., McDonagh, D., Silveira, M.H.
Key considerations and experience using the ultra-low-power Sensium platform in body sensor networks. Sixth International Workshop on Wearable and Implantable Body Sensor Networks, Berkeley, CA, June 3–5, 2009.

Olguin Olguin, D., Pentland, A.
Sensible organizations: a sensor-based system for organizational design and engineering. International Workshop on Organizational Design and Engineering, Lisbon, Portugal, December 11–12, 2009.

Olguin Olguin, D., Gloor, P., Pentland, A.
Wearable Sensors for Pervasive Healthcare Management. Third International Conference on Pervasive Computing Technologies for Healthcare, London, U.K., April 2009.

Pappas, C.
Three problems, one solution: attendee surveillance. Corporate Event Magazine, Summer 2009.

Pentland, A.
Honest Signals: How They Shape our World. MIT Press, Cambridge, MA, 2008.

Wong, A.C.W., McDonagh, D., Omeni, O., Nunn, C., Hernandez-Silveira, M., Burdett, A.J.
Sensium: An ultra-low-power wireless body sensor network platform: design & application challenges. Proceedings of the Annual International Conference of IEEE Engineering in Medicine and Biology Society, Minneapolis, MN, September 3–6, 2009.

Back to Top

Back to Top

Back to Top


UF1 Figure. Alex Pentland, right, and a team of researchers with reality-mining devices at Massachusetts Institute of Technology’s Media Lab.

Back to top

Join the Discussion (0)

Become a Member or Sign In to Post a Comment

The Latest from CACM

Shape the Future of Computing

ACM encourages its members to take a direct hand in shaping the future of the association. There are more ways than ever to get involved.

Get Involved

Communications of the ACM (CACM) is now a fully Open Access publication.

By opening CACM to the world, we hope to increase engagement among the broader computer science community and encourage non-members to discover the rich resources ACM has to offer.

Learn More