A new editorial in Nature Biotech poses the idea of the "digital phenotype" as a catch-all term for the trail of relevant health data people leave behind in their interactions with the internet, social media, and technology, which has largely untapped potential for the early detection of various conditions.
CareMore Chief Medical Officer Dr. Sachin Jain submitted the paper along with Harvard Medical School colleagues doctors Brian Powers, Jared Hawkins, and John Brownstein, whose Computational Epidemiology Group has done a lot of work in the area of digital phenotyping already.
“People are increasingly leaving a footprint of their health status through technology, including social media, forums, online communities, wearable technologies and mobile devices,” Jain said in a statement. “This information all has clinical value to a physician. The challenge is being aware of it and knowing how to access and interpret it.”
Back when Jain was at Merck, his innovation group funded some of Brownstein's work on HealthMap. They were already working on the notion of the digital phenotype at that time.
"Merck is funding the research, but I think there’s also a broader digital health agenda that we’re really kind of pursuing and this is a big part of it, which is around this notion of defining what I call digital phenotypes,” Jain told MobiHealthNews last March. “We think many diseases will actually have a phenotype that presents through patient use of technology."
The concept builds on Richard Dawkins' idea of an extended phenotype, which refers to how animal behaviors and interactions with their environments should play into how we describe and classify them. According to Jain and his colleagues, a burgeoning dataset includes data from health wearables, social media postings, and even data from people's mobile phones about how and how often they text and use apps. The latter has formed the basis of Ginger.io's research, which contends that changes in people's smartphone usage behavior can presage changes in a mental health condition like depression of bipolar disorder.
The idea of the digital phenotype can be used in a variety of ways, the paper goes on to say. It can be predictive of diseases for individuals, but it can also predict trends at the population level using something like Google Flu Trends or the Boston Children's insomnia study. It has also shown value in providing feedback about the efficacy of new treatments.
"For example, by tracking digital phenotypes among members of an online disease community, researchers were able to demonstrate the lack of efficacy of lithium in slowing disease progression in individuals with amyotrophic lateral sclerosis," the paper reads, referring to a PatientsLikeMe study from 2011. "These findings were later replicated in several slower and more expensive randomized controlled trials."
The challenge going forward will be incorporating this data into clinical practice in a way that is useful, doesn't overwhelm physicians, and maintains patient safety and privacy.
“Our digital breadcrumbs, whether from our online discussions or the health tracking devices we wear, have tremendous potential to inform clinical decision making,” Brownstein said in a statement. “The next step is to figure out how these data integrate into traditional electronic medical records.”