Growing up in Soviet Belarus, Vitaly Herasevich didn't spend much time with computers. So when he entered medical school in Minsk in 1994, the school's network for tracking patient data proved an alluring novelty. Herasevich quickly learned some computer skills and for 5 years worked on the system, viewing it from both a programmer's and a clinician's perspectives and noting where it could be improved. As he completed medical school, a Ph.D. in cardiac physiology, and then a cardiology fellowship, Herasevich continued to develop his informatics expertise; he even published a Russian-language textbook on the subject.

That experience straddling informatics and medicine served Herasevich well when he arrived at the Mayo Clinic in Rochester, Minnesota, for a postdoctoral fellowship in 2006. He knew by then that he wanted to devote his career to improving information flow in hospitals, and saw plenty of opportunity to do so. In 2007 he won a career development award from Mayo's National Institutes of Health-supported Center for Translational Science Activities and set about building a sepsis sniffer, a computer program that monitors a patient's vital signs and sends out an alert if sepsis, which is an aggressive immune reaction to an infection, or other serious trouble is imminent.

The symptoms of sepsis -- such as fever and increased heart and breathing rates -- can sneak up on caregivers. Untreated, the condition can kill. "Just a few hours of notification can play a really big role in reducing mortality," says Herasevich, who is now an assistant professor of medicine at the Mayo Clinic. Herasevich's sepsis sniffer is a computer algorithm that taps the critical care unit's stream of real-time patient data and, if it detects indications of sepsis or other serious trouble, alerts nurses or physicians via pager, e-mail, or an alert board. Mayo is now studying whether the sniffer decreases mortality in the hospital's intensive care unit. "This is an exciting field because you can really impact patient outcomes," he says.

The field Herasevich works in is called medical informatics -- a broad term encompassing a wide swath of careers, from mining genetics databases for disease clues to maintaining electronic patient records in a clinic. Between those two extremes of basic and applied knowledge lie two nascent, rapidly expanding subfields: translational research informatics and clinical research informatics. The former involves helping to move new tests and treatments from the lab to the clinic, and the latter focuses on improving information flow during clinical studies. Leaders say the need for workers in both subfields is exploding.

"These areas are just solidifying as fields, and there's a lot of work out there to be done, a lot of opportunities," says Peter Embi, director of the University of Cincinnati Center for Health Informatics.

New training opportunities

The Department of Health and Human Services announced in November that it would give out $80 million from the American Recovery and Reinvestment Act to help colleges and universities develop health information technology courses. Already, some 40 to 50 master's degree and Ph.D.-level medical informatics programs operate in the United States, says William Hersh, chair of the Department of Medical Informatics and Clinical Epidemiology at Oregon Health & Science University in Portland. And there are a handful of informatics fellowships for people who want further training, such as those offered through the National Library of Medicine and the Veterans Administration.

The aims, goals, and structures of the graduate programs vary widely, but Hersh says all emphasize bridging information technology and medicine. "The technical people like that I know the clinical stuff, and vice versa," says Julie Eckstrand, a research informaticist at the Duke University Health System in Durham, North Carolina.

There is no traditional training path into medical informatics, notes Eric Perakslis, vice president for research and development informatics at Johnson & Johnson Pharmaceutical Research and Development L.L.C. in Titusville, New Jersey. In fact, if you talk to 10 people in medical informatics, you'll probably hear 10 different stories about how they got there. Physicians, nurses, pharmacists, and computer engineers all find their way into the field.

Improving clinical research


(Brittany G)

Quantitative Careers

Also in Science Careers this week:

Quantitative Biomedical Careers - Physicists, mathematicians, and others are finding new ways to apply quantitative skills to biomedical sciences.

Statistics Serving Biomedicine - Spanish statistician David Rossell supports other biomedical scientists while pursuing his own research.

Khamis Abu-Hasaballah began his career as a biomedical engineer, designing medical devices for a company. During the 1990s dot-com boom, he grew interested in information technology, and over the course of a decade graduated from job to job in the IT department of the University of Connecticut Health Center in Farmington while learning basic hospital systems, the ins and outs of electronic health records, and the process of clinical research. "I came into the field not knowing what I needed to know," he says.

In 2008, Abu-Hasaballah created a new position for himself as his hospital's first clinical research informatician, reporting to the chief information officer. The goal of the new position is basic but vital: to streamline research. For example, Abu-Hasaballah, who has since progressed to assistant vice president for research informatics at the hospital, is building a system to match hospital patients to clinical trials.

Embi helped build similar software at Cincinnati, which alerts clinical trials intake coordinators when eligible patients pop up in the hospital's data warehouse. Duke's Eckstrand also streamlines clinical trials. She entered the field after spending 20 years as a research pharmacist on new-drug studies. In 2006, Eckstrand saw an opportunity to deploy her clinical research skills in a broader manner, designing systems that automatically move patient data from bedside monitors into research databases. Eliminating manual data entry for clinical trials is a huge time and money saver, Eckstrand says.

With the tools Eckstrand helped build, physicians and others involved in trials can watch data accumulate in real time, monitor volunteers remotely, and receive alerts about adverse events. "Everyone is keenly aware of how expensive drugs are, and one reason is because clinical trials are hugely inefficient," she says. "There's real tangible value to what we do."

Dealing with data


(Lucile Packard Children's Hospital)
Atul Butte

At academic medical centers, research informatics offices often operate in a service role, says physician and informatics professor Atul Butte of Stanford University School of Medicine in Palo Alto, California. Clinical departments and study investigators act as the customers. "Everyone [doing research] needs help organizing and analyzing data," he says. But for individuals who specialize in translational research informatics, like he does, Butte also sees expanding opportunities "to step out of the service role into a more scientific role."

At Johnson & Johnson, the informatics shop that Perakslis helps run focuses almost exclusively on translational informatics. Perhaslis says the goal is to integrate data from lab-dish experiments all the way through end-stage clinical trials. If, say, an unexpected adverse event pops up in a human trial, the informaticians can then sift animal, lab, and human data to investigate the possibility that certain gene variants or other factors predispose patients to the adverse event.

Given the huge quantities of genomic and proteomics data now being collected by research hospitals, professionals skilled in sifting that data for important signals can generate new knowledge of their own. "There are so many open questions [in medicine] and so much data out there" to be mined, Butte says. Approaching a biomedical problem from the informatics point of view -- for example, Butte is studying which gene variants might predispose children to type 2 diabetes -- allows a researcher to survey the available data without preconceived ideas. In contrast, most lab-based medical researchers focus on a few genes or a particular disease model, blinding themselves to other possibilities, Butte says.

Butte majored in computer science as an undergraduate and then attended medical school, later finishing a Ph.D. in medical engineering. His code-jockey skills "help tremendously" when designing translational studies, he says. For instance, the National Center for Biotechnology Information's public repository of gene-expression data, GEO, now contains 392,000 microarray experiments. "It's the hardest thing to get usable data out of these repositories. At the very least, you need to be able to write some code, write some scripts," Butte says.

Knowledge of biostatistics, database operation, and computer programming are also important skills, Butte and others say. Butte notes that for a career in translational research informatics, knowledge of genetics and proteomics is just as important as the computer skills, especially as data-management software matures and becomes standardized.

A diplomat's negotiating skills and a marketer's verbal prowess prove handy, too; Perakslis and his colleagues spend a lot of time selling and explaining the new resources the team develops. "You need top-notch communications," he says. "The best folks have to be out there making friends."

"I end up brokering a lot of agreements between different departments and finding compromises," Abu-Hasaballah says. "Being a people person really goes a long way."

And the best people in his field will also possess a blend of medical knowledge and computer savvy, he says. "You get to be right at the fulcrum of a lot of different fields, which is exciting."

Selected Informatics Resources

Organizations

- American Medical Informatics Association (AMIA)

- Healthcare Information and Management Systems Society

- International Medical Informatics Association

Training and fellowships

- University-based Biomedical Informatics Research Training Programs from the National Library of Medicine (NLM)

- Degree Programs, Certificate Programs, Fellowships, and Short Courses (from AMIA)

- Potential Funding Sources (from AMIA)

- AMIA's semester-long training course for health professionals

- NIH Pathway to Independence (PI) Award (K99/R00) through NLM

- Veterans Administration Advanced Fellowship Program in Medical Informatics

Blogs

- Informatics Professor -- by William Hersh, chair of the Department of Medical Informatics and Clinical Epidemiology at Oregon Health & Science University

- Health IT Buzz - by David Blumenthal, the national coordinator for health information technology at the Department of Health and Human Services

- On Informatics - multiauthor informatics blog from the Oregon Health & Science University

- Clinfowiki - Clinical Informatics Wiki

Brian Vastag is a science journalist in Washington, D.C.

10.1126/science.caredit.a1000009