Big data is reshaping every aspect of the health care business model. Tools like AI that connect the clinic to the patient have the potential to advance precision medicine and reduce the burden of chronic disease, which accounts for more than 80% of total U.S. health costs.

Despite the optimism, health care institutions have been slow converts to the revolution. Obstacles include a culture of professional autonomy in medicine and the logistics challenge of turning vast amounts of raw, disaggregated data into the knowledge to act.

“It takes more than an app to get that right balance between people and the machine, but if you do, you’ve nailed the future of health care,” says Oscar Marroquin, a cardiologist who oversees the development of UPMC’s large suite of information assets as Chief Clinical Analytics Officer in the Health Services Division.

Marroquin runs a team of data scientists, software engineers, and visualization specialists with a simple mandate: to back clinical decisions with evidence relatable to the whole person, not just the disease.

Never Miss a Beat!

Get Healthy Tips Sent to Your Phone!

Message and data rates may apply. Text the word STOP to opt out and HELP for help. Click here to view the privacy and terms.

“We don’t see data as just a research tool but as continuous learning support in patient care,” Marroquin says. “It’s more the bedside than the bench.”

Marroquin’s group is fully funded by UPMC and is embedded within its clinical network, eliminating the distractions of vying for outside research grants common among data crunchers in rival hospital systems.

Among its many projects, Marroquin’s group is using data prospectively to map what’s missing in the patient journey from diagnosis to recovery. With its own insurance plan, UPMC can curate the huge data sets necessary for conducting broader population health assessments linked to “well care” and prevention.

”In medicine we’ve been trained to lump people into fixed therapeutic categories,” Marroquin says, “but we now know that the phenotype of every individual consists of seemingly endless shades of gray. These include social determinants of health status, such as where you live, family background, occupational stress, and mobility issues. Using data from UPMC’s network of acute care and outpatient facilities, we can add these variables to the patient chart and then build highly accurate predictive models to identify those most at risk for adverse outcomes.”

Identifying the risk factors behind costly postoperative readmissions in UPMC’s network of 40 hospitals has been Marroquin’s priority. Pilot programs in 10 hospitals were conducted in 2018, evaluating millions of discharge records to target patients at highest risk for readmission. The pilot resulted in a 35% to 50% reduction in readmissions compared to prior rates, enough to start deploying the model for use by physicians at the point of care.

Marroquin sees more improvements to come as the data on individual patients deepens. “It is bringing us to the day when the art of medicine is validated by the hard evidence that can prevent disease, not just treat it,” he says.


Scientific American asked four UPMC leaders about how technology, science, and adjusted incentives are transforming the patient experience, and the future of medicine. This article is one of four interviews.

About UPMC

Headquartered in Pittsburgh, UPMC is a world-renowned health care provider and insurer. We operate 40 hospitals and 800 doctors’ offices and outpatient centers, with locations throughout Pennsylvania, Maryland, New York, West Virginia, and internationally. We employ 4,900 physicians, and we are leaders in clinical care, groundbreaking research, and treatment breakthroughs. U.S. News & World Report consistently ranks UPMC Presbyterian Shadyside as one of the nation’s best hospitals in many specialties and ranks UPMC Children’s Hospital of Pittsburgh on its Honor Roll of America’s Best Children’s Hospitals. We are dedicated to providing Life Changing Medicine to our communities.