Making Applied Research More Sensible

Burt Kaliski | Jun 10, 2013

Technology has been defined as "the application of practical sciences to industry or commerce." In its best form, it's a synthesis: the convergence of what's feasible in theory with what's viable in practice.  

Deborah Estrin, Verisign Labs' most recent distinguished speaker, recounted the point in her career as an accomplished networking researcher at UCLA where she recognized that her contributions to theory had extended much further than to practice. At that point, she decided to increase her efforts to make her ideas real -- even if, as she noted, research papers on applications were much harder to publish in academic conferences than papers on theory. 

That change in approach would ultimately lead Deborah to pioneering work on networked sensing technology, the use of mobile and wireless systems to collect and analyze data about people and the physical world they interact with. 

Networked sensing has many potential applications, depending on what one wants to measure. Is it something about the environment? About a data center? Perhaps about a person's health? 

Research into networked sensing, Deborah pointed out, is highly application-dependent. The problem is not how to sense in general, but how to sense for a particular purpose, to solve a specific problem. The application drives the research: it's what's needed, not what's new that matters. 

The application dependence also makes networked sensing a good example of multidisciplinary research, involving both networking researchers and application researchers. That type of collaboration, while vital, can take longer than research in a single domain. Researchers have to "take turns," she explained, improving their understanding of both the measurement tools, and what they're trying to measure. 

Deborah and her team's research started with distributed sensing, where devices at fixed locations measure properties of their environment, and often very specific properties for each device. The research eventually transitioned to mobile sensing, consistent with the advent of more capable mobile phones in the mid-2000s. Fixed locations were replaced with continually moving ones as the devices traveled with their users. Specific properties were replaced with more general ones as the devices became even more capable.

Mobile health, it turned out, would be a "killer app" for mobile sensing. With the instrumentation available via mobile phones, Deborah and team recognized that it would be possible to understand people's health-related behaviors, such as their level of physical activity. The measurements could also help assess the effectiveness of treatments. In essence, mobile sensing enabled an ongoing personalized data-driven health check. Chronic disease management particularly benefits from such a continuous check, as individuals often face what Deborah called an "optimization problem" for the use of pain medication and other treatments: too little and they're worn down by the effects of the disease; too much and they're weakened by the impact of the treatment. Mobile sensing can make it easy for a patient to assemble concrete feedback on the use of medication, both explicitly (e.g., by entering data into an application) and implicitly (via automated sensing of activity level), in order to find the optimal doses and times (or to know when to switch to something else). 

It's clear from Deborah's survey of her research, which she continues as a faculty member at the new Cornell Tech campus in New York City and as a co-founder of the non-profit Open mHealth, that the Internet is again enabling something transformational. Traditional anecdotal, unreliable feedback on a person's health is being replaced with real data that health professionals -- and individuals themselves -- can analyze in order to act more effectively. Indeed, mobile sensing offers a new form of evidence of health care effectiveness. Rather than long-term studies of a random population -- which as Deborah noted are only available for limited combinations of treatments -- mobile sensing can provide immediate feedback on a given individual and his or her specific interventions. 

Deborah concluded her captivating lecture with a compelling personal testimony on what could be possible if an individual's health data were more readily available to health care professionals -- a good motivation for continued research in this area. The full recording of her is available below. 

What other applications of mobile sensing do you think will make a difference in the quality of our lives?