|
|
For other articles and previous issues click here. September 12, 2005 A
Survivor’s Guide to Speech Recognition Is the much-discussed technology ready to leave its imprint on the HIM industry? If so, what would be the aftereffects? Even in the midst of a boom in speech-recognition technologies, medical transcriptionists (MTs) aren’t expected to go the way of dinosaurs. Some long-used MT skills may become extinct as will some of the day-to-day duties. Many in the healthcare field anticipate that with increased use of speech-recognition technology, the services provided will be redistributed rather than replaced. Landscape Changes Nick van Terheyden, MD, chief medical officer at Philips Speech Recognition Systems, says it’s inevitable that the healthcare industry will go in this direction. “Speech-recognition technology has reached top-priority status,” he says. “It’s had a checkered past, but the software technology has grown exponentially.” According to Lea Sims, CMT, FAAMT, director of publications and communications at the American Association of Medical Transcription, “The biggest reason that speech recognition hasn’t been able to fully, or even reasonably, automate transcription is due to both the limitation of the software [at least to date] and the highly interpretive nature of narrative dictation that often requires the kind of human judgment that a speech-recognition system just cannot provide.” Another limitation over the years, Sims says, is the extensive amount of resource time required of the provider to train the software. According to Matt Revis, product marketing manager for dictation products at ScanSoft, the reason speech recognition hasn’t gained a stranglehold is twofold: 1) There was a “hypecycle” in the late ’90s and early 2000s about the promise of speech recognition. However, the software at that time wasn’t up to the challenge technically, leaving many people disappointed. Speech recognition was sold as a way to completely eliminate transcription rather than bring efficiencies to the transcription process. 2) Even if the technology had been perfect, it would have required physicians to change the way they work and document clinical encounters. Many were reluctant to do that. “The doctors need to know that the technology does improve patient care because records can be available almost immediately throughout the enterprise,” Revis says. “And the MTs need to know that speech recognition can dramatically improve their productivity and value as medical language specialists rather than typists.” Van Terheyden believes zero change in behavior is key to success in many institutions. “Choice is the critical factor in successful speech-recognition implementations,” he says. “All physicians document in different ways and have a different affinity for the various tools that help speed up documentation.” Sims believes speech-recognition technology will redistribute MTs rather than replace them. “Since most viable speech-recognition solutions involve back-end editing, there will continue to be an evolving role for an MT editor in a speech-recognition environment,” she explains. “The variability of both dictators and narrative dictation will necessitate a blended work environment where speech-recognition technology and MT editors work together to produce a chart-ready record. Given the increase in health encounter documentation demands and the shortage of MTs, speech-recognition technology will be a partnering solution by enabling MTs to be more productive and responsive to volume demands.” Van Terheyden agrees that speech recognition will not make the skills of MTs obsolete. “That was the overwhelming fear for many MTs but in reality the reverse is true,” he explains. “Transcription today is in high demand. Speech recognition is not a panacea. There are some physicians whose speech is well adapted to the technology but there are those poor speakers, or those with heavy accents, that will still need to utilize the MT.” With its traditional slow turnaround times, manual transcription will eventually be replaced, van Terheyden says, but the adept MT will develop additional skills and be able to offer value-added services to other departments in the healthcare facility. Board-certified dermatologist Matthew Doppelt, DO, says when his practice implemented an electronic medical record (EMR), it eliminated the need for an MT. He uses speech recognition within the structure of the EMR. eClinicalWorks Vice President Girish Kumar agrees that EMRs are replacing MTs and explains that speech recognition is becoming the catalyst for EMRs. He says speech recognition has become quite accurate and is now on par with what traditionalists expect in the tools they utilize. “In five years, there will be a dramatic impact on MTs. They will have to come up with different value propositions,” Kumar says. “EMR is the driving force behind speech technology,” he continues. “If it were just speech recognition for speech recognition’s sake, it wouldn’t have been adopted as quickly. EMRs drive the adoption of this technology.” What Skills Will an MT Need? Front-End vs. Back-End Speech
Recognition Back-end speech recognition refers to the process by which the provider dictates into a digital dictation system, where the voice is routed through a speech-recognition machine and the recognized draft document is routed along with the original voice file to the MT editor, who verifies the accuracy of the draft and finalizes the report. “MTs that migrate to this role will need to have exemplary interpretive and editorial skills,” Sims says. “They will need to be the cream of the crop, since they will be expected to be the final authority on the accuracy of the speech-recognized information and the final quality of the document.” Because most doctors are reluctant to use the front-end approach, Revis says what the industry is seeing is speech recognition being deployed in two ways. “One is the front-end technology and the other is the back-end where the dictation is done the same way it’s been done for the past 20 years,” he says. “The main difference is that now the dictation is being recorded and rerouted to a speech recognition technologist who gets a draft output along with the original recording. The MT looks at the draft, listens to the recording, and makes changes. The MT is acting more in an editorial capacity. A successful deployment of speech recognition must have both front-end and back-end components that work in an integrated way.” MTs will still play a role in the speech-recognition world, Kumar says. “If a doctor uses dictation rather than speech recognition, some MTs will be involved in background technology to create the notes into text,” he explains. “Other MTs will edit for accuracy.” An MT who still requires editing or quality assurance on a regular basis or often needs assistance in interpreting dictation will likely not be a candidate for this role. Likewise, an MT editor will need to be highly skilled in interpreting and transcribing difficult dictation, particularly in an acute care setting, where challenging dictators will possibly still require traditional transcription or extensive editing post speech recognition delivery. Revis says that for an MT to thrive in the new environment, he or she will need a strong knowledge of not only medical terminology but editing skills as well. “MTs have to work differently, but speech recognition will make them much more productive if they embrace the technology,” he says. “We are seeing dramatic productivity increases for MTs editing on our speech-recognition solution.” Doppelt is a fan of front-end speech recognition. “When I have a difficult case that doesn’t fit one of my EMR templates, I use speech recognition in a front-end fashion (instead of typing) to enhance the readability of my notes,” he says. Doppelt says the new system has allowed him to improve the overall quality of documentation to his patients. “The process of getting a note back from the MT, reviewing it, correcting it, and sending it to the referring physician would sometimes take a week,” he explains. “Now I can have the note done and off to their primary care doctor before the patient even leaves the office. This is not just a convenience, it enhances patient care.” Getting Physician/Clinician Buy-In “If there are two minutes of extra effort on the doctors’ part, they will balk at the technology,” van Terheyden explains. “It has to be an almost seamless transition in order to be effective and accepted.” There are many doctors, Revis believes, who like front-end speech-recognition projects. “The front-end products bridge the gap from traditional dictation to the EMR and what we’re saying to clinicians is that you can’t continue traditional dictation and get that info into an EMR,” he says. “That message is resonating not only with the doctors but with those making the EMRs.” “Think basic for physicians, since many doctors are computer phobic,” Kumar says. “Those who are used to scribbling notes on a file will have a hard time changing. However, utilizing speech recognition will help ease the transition. The trending line is to have new doctors utilize the technology as soon as they come on board.” If a facility chooses a physician to be a champion for the new technology, more of the staff will be inclined to emulate that behavior, Kumar says. Van Terheyden explains that some departments are better suited to speech-recognition technology. “Radiology is a great area for speech recognition installation as it uses a finite grammar to describe a procedure,” he says. “On a chest x-ray, for example, if I see 150 of those a day and of that number 80% are normal, the standard components of a normal chest x-ray report could be called up using speech recognition and autotext so the radiologist doesn’t have to repeat the same thing over and over.” Cost Trade-Offs “To get the doctors to buy into speech recognition, it has to integrate almost seamlessly, in the least disruptive manner as possible, into the doctor’s daily routine,” Revis says. Van Terheyden says speech recognition’s true value lies in its multifaceted approach. Information can be available instantaneously and if there are six or seven teams in a facility that need access to that technology, it’s there. Doppelt says the switch to an EMR and speech recognition resulted in a return on investment that is much greater than the $1,800 per month he saved on dictation. “I don’t have staff running around trying to find a chart and making sure all the notes are filed,” he says. “I also used to have a room in my office that was devoted strictly to medical record retention—that room has been converted to a patient exam room. With the combination of EMR and speech recognition not only have I realized significant cost savings but I have also improved documentation of the patient encounter.” In a clinical setting in a small practice, Revis
explains, the doctor is paying his own transcription bills and that’s
a compelling reason to adopt speech recognition because the practitioner
could potentially realize a savings of $10,000 to $15,000 annually. “The first scenario is the healthcare facility likes the technology and doesn’t want to wait for MTs to turn around the records or because the hospital is making the switch to an EMR and they are forcing the doctors to use speech recognition as a way to bridge the gap,” Revis says. “Or it could be because the administration is looking to cut costs and transcription is where they decide to cut them.” This, according to Sims, is how speech recognition has migrated to a back-door scenario where it is often the MT service owner who invests in the technology as a productivity tool. A physician or facility that has invested in speech recognition has done so both to increase productivity and reduce transcription overhead. It works well in scenarios where the dictating staff are retooled to maximize dictation recognition. “The better the speech-recognition machine recognizes, the more accurate the draft is. The more accurate the draft is, the more productive the MT editor can be,” Sims notes. “Increasing the productivity of these MT editors will result in needing far fewer MTs to meet documentation demand, and the facility meets its goal of lowering transcription overhead.” Speech recognition can help improve quality of care
through better documentation, quicker turnaround times, and cost
and time savings, according to van Terheyden. Will Your Physicians Thrive?
“Poor dictators will not migrate easily and will find that they have spent a lot of money on a technology solution that still requires the same amount of work on the back end from an editor,” she says. “They will essentially pay more for the final product that way.” In even the past three or four years, Revis says the technology has improved tremendously, but he agrees that there is still much that needs to be done to bring the message to market. “No matter how good the technology though, we still face the challenge of getting buy-in from the doctors,” he says. “Our solution has been to offer doctors options. Allow the physician to use front-end, back-end, or a combination of the two approaches. Though physicians are said to be slow to adopt new technologies, our experience is that if we can make it productive and convenient, there is strong and enthusiastic pickup.” Doppelt began using speech recognition within his EMR in October 2004 to enhance both readability and speed. Originally, he used an MT service with back-end editing. “I would dictate into a digital recorder [and] upload it to a Web site where it would be transcribed and then sent back for my review,” he recalls. “Although the notes looked good, this service was costing me almost $1,800 per month.” Many of the software programs come with the capability to analyze prepared documents and pick up words that are frequently used by physicians and clinicians in specialty practices. According to the experts, speech-recognition technology is light years ahead of where it was even a handful of years ago. Many products feature voice recognition that offers accuracy rates 30% higher than their recent counterparts. The advances also include out-of-the-box software that is a preconfigured model of a medical specialty. — Robbi Hess, a journalist for more than 20 years, is a writer/editor for a weekly newspaper and monthly business magazine in western New York. |
 |
 |