Advances in Voice Recognition Software Win Over Doctors
Most of us still don't have a computer that talks when we listen. But the newest voice recognition software is gaining support from skeptical doctors and helping in the complex effort to make electronic medical records electronic.
CIO — Voice recognition software has earned a reputation for being a gee-whiz technology that has yet to change the everyday life of most business users. However, recent improvements in the technology have convinced more doctors to make voice recognition part of their daily routines—and part of the complex, ongoing effort to make medical records electronic.
"We are moving fast in the emergency room and the voice interface makes us more productive and accurate," says Dr. Brian Zimmerman, an emergency room physician at Miami Valley Hospital in Dayton, Ohio. In 2006, Miami Valley instituted Nuance's Dragon NaturallySpeaking voice recognition software with its electronic medical record (EMR) system to reduce the complexity of paper records.
The software not only accelerates processes and saves money, but also allows doctors to make more informed diagnoses, says Mikki Clancy, the hospital's VP and CIO. Using Dragon saved the hospital $1.4 million in transcription costs in 2007 and has improved records to include more notes from doctors, which improves patient care, Clancy says.
As the voice recognition software continues to improve its accuracy, it can more easily be implemented into a physician's hectic workflow, Zimmerman adds.
In one recent advance, Nuance, the leading company for voice recognition software, has enhanced its Dictaphone Dragon NaturallySpeaking Medical version 9 to have 80 different vocabularies, each specific to a division of medicine (such as radiology), to increase accuracy, says Peter Mahoney, VP and general manager of desktop dictation business units for Nuance's Dictaphone division. That unit has some 500 internal speech scientists working to advance the software's accuracy.
Another recent improvement involves increasing accuracy by calibrating the software to "understand" accents, which greatly helps doctors who don't speak English as their native language. The software's accuracy rate is currently 98 percent and rising, according to the company.
For doctors, transitioning from manila folder to computer screen continues to be a complex endeavor. Without speech recognition software, hospitals typically need to hire someone or use an outside service to transcribe all the doctor's notes into the records system, which creates costs and a significant time lag between patient data being captured and made available. Dr. Douglas Golding, the medical director and chief of healthcare informatics at Lifetime Health Medical Group, a group of primary care doctors serving more than 100,000 patients in Buffalo and Rochester, N.Y., switched from a system of transcription to Dragon's software in 2006.
