Duke Translational Medicine Institute Data

{{ story.headline }}

{{ story.subheading }}

{{ story.timestamp }}

A policeman is observing his night duty when he notices a man searching for something under a lamppost. He approaches the man and asks what the matter is, and the man informs him that he’s lost his keys. After several minutes of looking together, the policeman asks whether the man is sure he dropped them under the lamppost. The man replies, “No sir, but this is where the light is.”


This story, referred to as the “streetlight effect,” has been cited in the social sciences since the 1960s. It’s only very recently, however, that its relevance has been explored in the field of medicine.

Billions of dollars are funneled into medical research and technological innovation every year, yet the amount of people diagnosed with cancer has been on the rise, and over 600,000 Americans die annually from cardiovascular disease alone. Experts have been scratching their heads...despite endless efforts, groundbreaking discoveries, and promising treatments, why are so many patients still dying? What are we doing wrong?

Google’s high-reaching “secret” ideas lab, Google X, has introduced a possible answer with what it and The Wall Street Journal calls a “moonshot project”: the Baseline Study. Whereas most Google X endeavors have involved commercial products—like the self-driving car and the high-altitude Internet service balloons—Baseline is a partnership to conduct meaningful healthcare research that has typically been considered too expensive or ambitious. Its goal is to better understand the healthy body. In collaboration with renowned physicians and health professionals at Duke and Stanford, Google aims to spread the light of the lamppost in medicine by merging big data with traditional practices of research and physician-patient interaction.

But in the months since the July 2014 announcement by Google, there's been hype and skepticism around the elusive study. Critics question how Google will keep study participants anonymous when they're giving blood, tear and urine samples and sharing innermost details of their physical and mental state. And some question whether the study really qualifies as a "moonshot" considering the scope and scale of other scientific research over time.

But Dr. Michael Pencina, professor of biostatistics and bioinformatics at the Duke Clinical Research Institute and the lead statistician on the study, explains that the study is truly unique in process, methodology and technology. In conventional scientific research, hypotheses are formulated based on what’s already known, and so there may be some seemingly unrelated pieces in the puzzle that have long been ignored.

To unearth those pieces, Baseline borrows from the technical process of machine learning to draw associations between health patterns on a scale unlike ever before. If successful, this endeavor “has the potential to become the framing study in science for the upcoming generations,” Pencina says. 

And it’s happening right here in the Triangle.

Why Duke? 

There is some precedent for the Baseline Study at Google. In 2008, Google introduced an online portal for people to self-input their health data, calling it Google Health. But this endeavor largely failed. People didn't want to leave their personal medical information on the Internet— what would become of it? With growing concerns about internet safety and plans for Baseline taking shape, it was clear that Google needed partners that had clout—institutions that could define goals of the study, secure thousands of willing participants and use a formal review process to ensure safety and accuracy. And so Google turned to Stanford and Duke, two entities known for their success in clinical research.

There also were key connections to our region. Leading the new and improved Baseline effort at Google X is Dr. Andrew Conrad, a California cell biologist and manager of Google X’s Life Sciences team, who in 2005 worked with Dr. Robert Califf, Duke's vice chancellor for clinical and translational research and the founding director of the Duke Clinical Research Institute, and David Murdockthe billionaire chairman of Dole Food Company, to establish the North Carolina Research Campus, a collaboration between nine universities (including Duke, the University of North Carolina at Chapel Hill, and NC State). Murdock funded the Duke University-led MURDOCK Study, a community-based study in the city of Kannapolis, and Cabarrus county and its surrounding area. Califf serves as the principal investigator of the MURDOCK study and with Victoria Christian, executive manager of the study, he has led the development of Baseline.  

Dr. Robert Califf and Victoria Christian Duke Translational Medicine Institute
The MURDOCK study brings together an interdisciplinary, committed team of Duke clinicians, professors and scientists from various Duke entities, centrally led by the Duke Translational Medicine Institute. Several of those groups are now also involved with Baseline, locally led by Califf.

With strong connections to big decision-makers in Silicon Valley and a proven track record with similar research ventures, Duke provided the talent and resources to make Baseline’s vision happen.

How the study works.

In order to draw a comprehensive picture of human health and the transition to disease, detailed health information is collected directly from study participants over a five-year period. The data gathered ranges from common laboratory metrics like blood pressure to updates on audiology. Participants maintain close communication with doctors and study coordinators on a day-to-day basis using unique tools like an online “participant portal,” developed by Google. Updates are also collected via Google’s other latest biomedical technology, including glucose-measuring contact lenses and a physical accelerometer to measure physical activity.

Updates are then stored in a private database program, which uses pattern identification software to draw associations. In the case that a participant becomes ill, clinicians can look at past information to identify these associations and critical transitions that may have heralded the illness.

Duke Baseline Study Researchers
Drs. Michael Pencina and Erich Huang are Duke University researchers and professors working on the Google X Baseline Study.


“It could be that big events like heart attacks and stroke are dated by the change in a parameter we haven’t yet identified in classical medicine,” Pencina says. Baseline seeks those parameters, otherwise known as “biomarkers,” using its immense computing capabilities on the provided health records. These biomarkers will inform the development of better diagnostics and therapeutics, and diseases could be prevented at the sign of certain “signals” before they become life-threatening.

“Right now, we aren’t using information to its max value. Electronic health records still look like old paper ones. What if you had a system that could alert a doctor if a measurement they’re inputting is drastically different from a patient’s previous measurements?” said Erich Huang of Duke’s Department of Biostatistics and Bioinformatics. Through its mission of compiling data over time, Baseline may allow for such ideas to become possibilities.

Along with the collection of molecular and genomic data, Baseline will look at factors never before incorporated into highly scientific research like one’s social environment, family structure, career, psychological well-being and other sociobehavioral factors. More importantly, the study increases the potential for innovative partnerships across disciplines, such as behavioral sciences, informatics, ethics, and nutrition— both at Duke and beyond.

Critical, are the right participants.

The success of the study requires community participation. Duke’s experience with the MURDOCK Study gives it a unique position in terms of credibility. By doing on-the-ground recruiting (with the help of community ambassadors, study volunteers, and student interns) and giving participants early access to some of Google’s newest gadgets, Duke aims to enroll thousands of participants from diverse backgrounds into Baseline over the next three years. Stanford is also conducting pilot tests at its affiliated sites.

Google Baseline Study Researchers 2
Drs. Ashley Dunham and Kristin Newby are Duke University researchers working on the Google X Baseline Study.

“We’re giving our participants an opportunity to be partners in this endeavor and hoping it will incentivize them to be more engaged in their health,” says Dr. Ashley Dunham, director of operations for population health initiatives at DTMI. Dunham claims that the most important thing is to establish trust and assure participants that their data is recorded and stored confidentially.

What it means for the future of medicine and health.

By collecting and integrating layers and layers of critical data, Baseline could reveal knowledge of health and disease on an unprecedented scale, the team members say.

“The scope of this project has never been attempted in the past,” explains the principal Investigator of the study and Duke cardiology professor, Dr. Kristin Newby, who manages Duke’s efforts in Baseline. “Not only does it cost a lot and take a lot of computing power, but it requires the melding of many disciplines and skill sets, partners with varied but complementary motivations, operational expertise, technologies that are just now coming of age and converging, and the trust and imagination of willing research participants over the years.”

Medicine has seen radical growth in recent times, yet many problems are still begging to be solved. Google X, Duke and Stanford propose that perhaps we haven’t been asking all the right questions. 

Perhaps we need to update our definitions of disease. Perhaps, for so long, we’ve only been looking under the lamppost. And by leveraging technology to expand our field of illumination, we may finally have the chance of finding our keys.