Privacy presents another formidable challenge in terms of how to keep patient data from falling into the wrong hands. In the U.S., health-care providers are subject to a set of stringent laws and regulations intended to keep patient data safe, most notably the Health Insurance Portability and Accountability Act (HIPAA). Other regulations vary from state to state, however, and it can be difficult to design EMR systems that comply with all the required policies. “Lawyers write these policies, and no one ever looks at them,” says Annie Antón, a professor of computer science at North Carolina State University. “It’s a huge problem.” Formalizing rules that were intended to be interpreted by courts, rather than by programmers, is also challenging. Whereas European Union law grants its citizens ownership of their medical data—and thus a great deal of control about what happens to it—U.S. laws are less clear-cut. HIPAA doesn’t address data ownership, and though it gives patients access to their records in most situations, it also grants doctors plenty of authority such as the right to share data with insurance companies, for example, and with other medical specialists.

A more basic privacy-related challenge stems from the complexity of health-care delivery. How is it determined which providers can access patient data? In general, only individuals with a direct medical need should be able to access files. Yet in an emergency-room setting, dozens of people may need to examine a single patient’s record, from doctors to dieticians to interns—and speed is often of the essence. At Partners HealthCare’s network of hospitals, emergency-room doctors can only access the records of patients who have already visited that hospital, and only view (but not modify) them. “It’s an imperfect system,” admits Cynthia Bero, CIO at Partners Community Healthcare, “but we have to balance providing the best care with preserving patient privacy.” Other EMR systems enable hospitals to grant and revoke patient data access on a daily, and sometimes hourly, basis.

From a technological perspective, there are three different architectures: centralized models, where data is stored and held in a single database; federated models, where it is distributed across a

network; and a hybrid of the two. Each model has advantages and disadvantages. Centralized systems are generally regarded as more secure and easy to manage—and hence better for research purposes. Federated systems, however, may be easier for small practices to deploy, and to facilitate the aggregation of data across different networks. But they also force healthcare providers to rely on a third party to guard against data loss and breaches of privacy. It’s not yet clear, experts say, exactly how architectures will evolve in the future.

 

clinicians’ Daily Workflow To be truly useful, of course, EMRs must fit seamlessly into clinicians’ daily workflow. Yet this, too, presents a challenge. From a physician’s perspective, it’s faster to scribble a prescription on a piece of paper than it is to log onto an EMR system and electronically enter the data. (Renewing a prescription, on the other hand, is much quicker with an EMR system.) HCI specialists like Jason Saleem, an investigator at the Regenstrief Institute, are therefore working to develop systems that are easier and more efficient to use. “By understanding doctors’ workflow, we can design better EMRs,” he explains. One lesson is that it may not be possible, or even desirable, for EMRs to fully replace paper. In a study published in the International Journal of Medical Informatics, Saleem observed that index cards and Post-It notes often serve as important extensions of EMRs when it comes to tasks such as remembering an impor-

standards for
the collection,
exchange, and
retrieval of
electronic medical
information vary
widely from system
to system.
Milestones
Computer
Science
Awards

The John d. and Catherine T. Macarthur Foundation and other organizations recently honored members of the computer science community for their innovative research.

macaRthuR founDation “Genius” a WaRD Maneesh agrawala, associate professor in the department of electrical engineering and computer sciences at the university of California, Berkeley, was among the 24 fellows for 2009 selected by the Macarthur Foundation. agrawala will receive $500,000 over the next five years. in announcing his selection, the foundation noted that agrawala is working “at the intersection of visualization, human-computer interaction, and computers graphics” and is “designing visual interfaces that enhance our ability to understand large quantities of complex information.”

nsf caReeR a WaRD
National science Foundation
awarded Fengyan Li, assistant
professor of mathematical
sciences at rensselaer
Polytechnic institute, a Faculty
early Career development award.
Li will use the five-year, $582, 112
award to design, analyze,
and implement computer
algorithms for solving complex
mathematical problems arising
in sciences and engineering.

comPuteR-aiDeD VeRification a WaRD The 2009 Computer-aided Verification (CaV) award, which includes a $10,000 prize and citation, was presented at the 21st annual CaV conference in grenoble, France, to seven individuals who have made major advances in creating high-performance Boolean satisfiability solvers. They are Conor F. Madigan, Kateeva, inc.; sharad Malik, Princeton university; Joao P. Marques-silva, university College dublin, ireland; Matthew W. Moskewicz, university of California, Berkeley; Karem a. sakallah, university of Michigan; Lintao Zhang, Microsoft research; and Ying Zhao, Wuxi Capital group.