Electronic Health Records: Will Natural Language Processing Be the Key?

With the majority of physicians still resistant to electronic health records, the fate of the Obama Administration's health IT program may depend on a technological breakthrough that makes it easier for doctors to use EHRs. While it's unclear whether such a breakthrough is on the horizon, IBM (IBM) and Nuance Communications (NUAN), which makes speech recognition software, claim that they have found a way to convert dictated text into discrete, searchable data. If doctors could document their notes in an EHR as easily as they now dictate them, it would be a game changer for both physicians and health IT vendors.

IBM and Nuance have collaborated recently on research that has helped make Nuance's Dragon Natural Speaking software faster and more accurate. The companies have just announced a new partnership under which they will co-develop and Nuance will market a "clinical language understanding" (CLU) product. This fall, Nuance will begin beta-testing the product in healthcare systems and physician practices, and it plans to begin selling it to EHR vendors and other parties next year. Not coincidentally, 2011 is the year in which physicians will have their first opportunity to qualify for government EHR incentives if they can show "meaningful use" of their systems.

"Clinical language understanding" is just another moniker for "natural language processing" (NLP), which connotes the ability of computers to understand and make use of human language. In applications marketed by 3M (MMM), CodeRyte, and A-Life, NLP is already being used to help hospital billers improve their coding speed and accuracy. But going from billing to clinical applications is a quantum leap.

Obviously, NLP has enormous commercial potential, not only in healthcare, but also in other industries. IBM, Microsoft (MSFT), and Philips (PHG), among other companies, have been researching this area for a long time. Progress has been slow, because language is so complex. Medical terminology is even more challenging, partly because many different terms are used for the same concept.

But Janet Dillione, executive vice president and general manager of Nuance's healthcare division, tells BNET, "The technology is now there. It's smart enough and distinct enough that we can embed this in the physician workflow."

As Dillione explains it, CLU applies statistical algorithms to the relationships between words and phrases so that they can be understand for purposes of clinical documentation. Another part of the application maps multiple terms for a particular medical concept to a standard terminology. The program is good enough, she says, to capture the nuances (no pun intended) of a "history of present illness," one of the most challenging parts of a patient exam to document in an EHR.

IBM's interest in working with Nuance on CLU is apparently related to Nuance's relationships with leading EHR vendors that have integrated Dragon into their applications. According to Dillione, more than 160,000 physicians use Dragon to dictate their notes, and many of those have EHRs that are interfaced with the speech recognition software.

Nuance is already discussing the new CLU product with EHR vendors, Dillione adds. One reason they're interested, she says, is that the application can help doctors abstract data from their dictation to populate a particular type of patient care summary. Doctors must use this summary, known as the Continuity of Care document, to exchange data with other providers in order to show meaningful use.

Whether or not the CLU software will work as advertised remains to be seen. But if it does, it could surmount a significant obstacle to physician adoption of EHRs.

Image supplied courtesy of Wikipedia.
Related: