Vol. 14 •Issue 16 • Page 26
Is Speech Recognition All Talk?
Does new MT technology generate money or just a lot of noise?
The latest health care press releases boast, “Speech Recognition Software Increases Transcription Productivity by 150 Percent,” and “New MT Technology Improves Turnaround Time.” Hook, line and sinker, right?
Well that’s what “they’d” like you to believe. But in reality, it’s just not that simple, and even though there are successful speech recognition programs out there, medical transcriptionists (MTs) and health care facilities seem to be bombarded with noise and not a lot of detail.
ADVANCE has decided to filter out the noise, sifting through the many success stories to find examples that can shed some light on medical transcription technology now and in the future.
“I think it is most important for health care facilities to understand their current documentation process and future documentation plans. Savings can only be gained when the current costs and work flows are fully understood and the length of line and all of the extraneous costs to transcription services are identified,” stated Jefferson Howe, MSA, CMT, transcription manager, Maine Medical Center, Portland, ME, and president of the American Association for Medical Transcription (AAMT).
“Many times people sweep into ventures and projects without understanding what the true outcome should be and that the ultimate goal should be accurate, timely and efficient documentation,” Howe continued.
Thorough research and continuous communication are essential in implementing new medical transcription technology. And the first step is deciding what kind of technology will help your health care facility’s return on investment (ROI).
Don’t Listen to the Hype
“One hundred percent accuracy through automation is never going to happen,” said Howe. “MTs transcribe for physicians and health care providers whose primary goal is patient care. [Doctors] get tired; they spend 12 hours in surgery and then emerge from the operating room to dictate. These dictators are bound to misspeak, and need corrections or edits. Aside from that, there are too many colloquialisms in English that a recognition engine would need to overcome.”
Using word expansion or speech recognition programs is a must to produce quality work in less time, but an MT’s trained eye will always be needed.
It is assumed by most vendors that MTs already work with some type of word expansion program, whether it is Microsoft’s AutoCorrect, SpeedType or Smartype.
But according to Betsy Ertel, president of SpeedType, “You would be surprised at how many don’t use an expander program at all, because they like to ‘type’.” Her question to those MTs who refuse to stay in step with society is, “Why would you want to type all the letters if the technology is out there? You could type two letters and produce 13 letters. It prevents carpel tunnel syndrome and preserves your fingers. The use of macros improves quality assurance.”
So the issue to be addressed is not should an expansion program be purchased, it’s should you advance your expansion system or venture into the speech recognition world.
Julie Stegman, publisher, Medical Language.and.Reference/STEDMAN’s, Lippincott Williams & Wilkins, feels that, “It would be smarter for an MT or a health care facility to look into enhancing their expansion program rather than waiting for speech recognition to advance.”
Detective Skills Needed
The main reason for Stegman’s advice is because word expansion programs are proven to increase productivity instantly.
“It will pay you back the first day you use it, saving three hours in a seven-hour day or five hours in a 10-hour day,” Ertel stated.
And in terms of cost, it’s really hard to beat. “It’s $195 for Stedman’s Smartype—a one-time charge,” Stegman pointed out.
So if speech recognition could increase productivity, improve ergonomics and produce full ROI, just like word expanders, would there even be a debate?
No, but trying to find out if that’s possible isn’t always easy.
Right now Chicago’s Mercy Hospital is proclaiming success with Agfa HealthCare’s TalkStation® program. The hospital’s turnaround time is less than 20 hours and the implementation process went smoothly. But there are a few drawbacks.
It’s currently only used in the radiology department because repetitive fields of medicine are easier for the speech recognition software, it’s a huge investment and it’s a front-end speech recognition program.
“The radiologist can read the exam, dictate the results, review and modify the report if necessary, and then sign it electronically, right at any workstation,” explained Phyllis Bachand, information technology applications manager at Mercy-Chicago.
It’s working seamlessly in Mercy-Chicago’s radiology department, but what about the rest of the hospital?
Bjoern Aune, U.S. sales director, Philips Speech Recognition, feels that there is another option.
Read Between the Lines
An improvement to front-end speech recognition is back-end speech recognition. “It means no change in the physicians’ working habits, which is a significant argument when it comes to accepting new technology. Physicians continue to concentrate on their core competencies, while MTs have the experience and skills to professionally finalize documents,” Aune stated.
Back-end speech recognition software produces a draft of the dictated report and transmits a synchronized voice-and-text document to the MT, who then becomes an editor. The MT listens and reads, correcting any recognition errors or dictation mistakes, according to Joe Weber, CEO, Lexicore.
Currently Aiken Medical Center in South Carolina is experiencing the benefits Aune described. Aiken is using Philips’ Speech-Magic™ and is pleased with the results. They went from 48 hours turnaround time to 1.5 hours. SpeechMagic is still only fully integrated via Winscribe’s digital dictation workflow into Aiken’s Radiology Management System, but expansion into other departments is possible.
But as the benefits improve, the cost could still be an issue.
“Usually most vendors have an up-front cost to buy the software. With some of them charging up to 4 cents a line, it can get very expensive,” Weber explained as his biggest concern about the new technology.
There is a continued cost for Speech-Magic to maintain the software.
And the cost is even more when you buy a total solution. A total solution means a facility has to go onto the vendor’s platform. “That’s fine if you want to replace your dictation and transcription systems. But if you’re happy with those, then it’s a waste of money to switch. It might also lead to a suboptimal workflow,” stated Weber.
Some facilities are happy with their current workflow and don’t want to rock the boat by adopting an entirely new platform and adjusting to a new system.
But Aune insists that Philip’s is solving this issue by having SpeechMagic embedded into medical dictation and transcription workflows such as Dolbey & Company’s Fusion Text or Winscribe’s Internet Author. Instead of introducing a new platform, hospitals can upgrade their existing systems to speech recognition and preserve existing workflows.
One back-end speech recognition program at Maine Medical Center has found greater success. “eScription Inc.’s Edit Script™ is being used throughout the hospital, in standard dictated reports in 40 work types including outpatient mental health notes, operative notes, history and physical, consultations, summaries and a variety of clinic reports,” raved Howe.
More impressive is the 75 percent productivity improvement, which allowed additional work to be tackled that was once outsourced. Seventy percent of dictation is now done by MTs editing draft transcription. MTs do both editing and transcription throughout the day so it breaks up the physical demands of the job, and its projected savings in 2004 is $213,000 in reduced outsourced costs!
But remember that fine print? It’s still a total solution. Maine Medical Center went onto eScription’s platform, which is a significant investment, but bearable for the biggest hospital in the state. It’s the bigger corporations and facilities that are best positioned to implement this technology right now.
So Is It all Just Noise?
“Speech recognition is highly sophisticated technology. Recognizing the speech of dictating physicians is enormously challenging, and it’s surprising the technology can do as well as it does. It’s not perfect, but it’s amazing how accurate the current technology has become,” explained Weber.
There have been unbelievable strides made in the speech recognition arena, but for the technology to substantially penetrate the market, it must handle general clinical dictation in a cost-effective manner, which Weber insists is within his company’s grasp.
Lexicore’s back-end speech recognition solution is coming soon and is designed to plug into an existing dictation/transcription workflow and cost less than a penny a line. Weber explains that it will be more affordable because the existing workflow is not altered.
“I recommend that health care organizations and medical transcription companies immediately implement the expander program that they determine will maximize their MT’s speed. At the same time, they should run an experiment to see what the productivity increase and ROI will actually be if they install back-end speech recognition,” Weber advised.
And Howe emphasized that MTs play a vital role in the patient health care process. “A back-end speech recognition process can provide another layer of protection to risk management issues that can arise from erroneous dictation.”
But when it comes down to it, Stegman reminded, “I know speech recognition will definitely have a role in the future, but until then–advance with the technology we have now.”
Tricia Cassidy is an assistant editor at ADVANCE.