Healthcare a Seedbed for Analytics Research

This article appears in the September 2012 issue of HealthLeaders magazine.

The healthcare industry, which has long trailed other industries in its use of analytics, is developing into a seedbed for research on advanced analytics, including topics such as natural language processing, artificial intelligence, and genomics.

The healthcare analytics world last year took particular note of IBM's Watson project, a supercomputer able to answer questions posed in natural language that defeated human competitors on the TV quiz show Jeopardy! IBM then announced it was working with Columbia University and others to commercialize Watson for clinical analytics and decision support.

At HCA, Jonathan Perlin, MD, chief medical officer and president of the clinical and physician services group, says, "We're doing some advanced work in terms of looking at how we might use natural language processing to detect subtleties in data and, in the future, even better support for precision medicine and personalized care."

In March, Memorial Sloan-Kettering Cancer Center announced its partnership on the Watson research. "We're still in the early work, very actively working on this project with IBM," says Patricia Skarulis, vice president of information systems and CIO. "All I can tell you is they are devoting a lot of resources and we are devoting a lot of resources. Database people, analytics people, our senior physicians. It's been a fun project."

Indeed, the future for analytics in healthcare only gets bigger and more ambitious from here.

For one thing, the amount of data being collected for analysis is exploding. "We're bringing quite a bit of genomic data right now into our clinical warehouse," Skarulis says. With the consent of its patients, the organization is studying pathology information from colon cancer patients and certain lung cancer patients, currently importing results from 60 different tests.

"What makes our people very excited is to have the genomic data, and to have this pristine clinical data to be able to combine with it, we think, will be an extraordinarily useful resource tool," Skarulis says.

Researchers are tapping into even newer analytics technologies to analyze structured and unstructured healthcare data. At UW Health  in Madison, Wis., chief research information officer Umberto Tachinardi, MD, is creating an advanced data warehouse built in part on technology from  I2B2 (Informatics for Integrating Biology and the Bedside), an NIH-funded National Center for Biomedical Computing at Boston-based Partners HealthCare System.

UW Health represents the academic healthcare entities of the University of Wisconsin-Madison: UW Medical Foundation, the 566-bed UW Hospital and Clinics, UW School of Medicine and Public Health, American Family Children's Hospital, and UW Carbone Cancer Center.

The open-source software powering I2B2 is utilized by 60 large academic medical centers, including UW Health, which were awarded Clinical and Translational Science grants by the National Institutes of Health, Tachinardi says. I2B2 accelerates the process of copying parts of electronic health records into deidentified data sets so researchers can more easily identify interesting phenotypes within the data, he adds.

"Curating information is a very expensive part of our business, regardless of being a research or clinical organization," Tachinardi notes. For instance, simple data types such as gender become complicated given the growth in transgender population.

"In the research world we're already used to living with those complications. But the clinical world is just starting to learn about them right now," Tachinardi says.        

This article appears in the September 2012 issue of HealthLeaders magazine.

See Also:

How Mobile Technologies Fuel TeleHealth Advances