Look who's talking

Voice recognition in healthcare might have been unspeakable a decade ago, but advances in technology have made manual transcribing a thing of the past for many physicians. Medical Times investigates the systems that everyone is talking about.

  • E-Mail
By  Vernon Baxter Published  March 9, 2008

Voice recognition in healthcare might have been unspeakable a decade ago, but advances in technology have made manual transcribing a thing of the past for many physicians. Medical Times investigates the systems that everyone is talking about.

In October 1997, Bill Gates, chairman of Microsoft and one of the world's richest individuals, predicted that "in this 10-year time frame...we will have perfected speech recognition and speech output".

Gates may have been a little optimistic with his prediction, but voice-recognition is finally making inroads into healthcare. With their fast-paced environments, hospitals and clinics have been reluctant testing grounds for speech-recognition.

If industry experts are to be believed, speech-recognition software is starting to find its voice in the healthcare industry.

But with the increasing presence of electronic medical records in clinics, the potential benefits of a voice-activated interface have never been greater.

"There is a trend in hospitals to try and reduce their transcription fixed costs and management is also trying to push the move towards EMR, which now forces doctors to type," explains Koen Schoof, senior product manager, dictation and healthcare, EMEA, Nuance Communication.

"A lot of doctors don't like to type, don't type that well, or work in situations where they are on the move the whole time.

"Voice recognition software can be used on virtually all windows programmes, including EMRS, where doctors currently lose a lot of productivity by typing into it."

Ironically, it is the fear of a further dip in productivity that has proved the greatest barrier for physicians considering speech-recognition.

The lure of hands-free dictation has so far been soured by an onerous training programme and inaccurate transcriptions. But if industry experts are to be believed, speech-recognition software is starting to find its voice in the healthcare industry.

Talking it up

"Speech recognition has come a long way," insists James McPherson, project and client manager of the UK-based Voice Technologies.

"When you look back even six or seven years ago, you would almost have to dictate like a robot for the system to understand what you were saying."

Programmes were previously reliant on a long training process, where physicians would have to read out prescribed texts for a minimum of one hour. For busy physicians, the investment was rarely worth the monotony of the exercise, and the results underwhelmed most.

Now the recommended training time is just over two minutes, according to McPherson.

"After that, consultants can expect to use the system and expect to get quite a high recognition rate from it straight away." The software's ability to constantly adjust itself with every change has been hailed as a crucial breakthrough, he adds.

"Previously if you wanted the recognition to get any better you would have to go into a separate training process, whereas now you can continue to use the application and it is basically learning in the background."

An obvious concern for the medical field is the system's error rate. Transcription error is a notorious problem in hospitals, and has potentially fatal implications.

For many, trusting software to dictate prescriptions is a technological bridge too far. McPherson stresses that every physician has a responsibility to check that what it is in each report is accurate and correct, no matter which method was used to document it.

"It is really down to a physician to see that the errors are not there, so I don't really see a way round that," he reasons.

"It is the same when you record a dictation and send it to a secretary - if the consultant doesn't read it properly and then verifies the report, then the same could happen."

Talk is cheap

Whether doctors like it or not, voice recognition software may become an economic necessity.

The sheer volume of information the healthcare industry generates requires that either doctors takes time out of their schedules, or administrators foot the transcription bill for medical secretaries.

Over the last 15 years, healthcare has gone from handwritten notes, to micro cassettes for secretaries to transcribe, to the cusp of centralised digital dictation with a digital repository, according to Costa Mandilaras, president of Canada-based Crescendo Systems.

"What we will see over the next few years is physicians dictating to PCs and doing their changes in real time," he claims. Voice recognition is now at the stage that it makes economic sense for most providers to implement, he adds.

"There are limited resources for documentation and the faster you can get it done the faster you can get paid - if you do everything at one time with voice technology then you will really start to see the benefits."

From a technological standpoint, the next stage for voice-recognition is artificial intelligence. James McPherson has been at the forefront of the voice-recognition research for the past 11 years and feels the systems are ready to move past simple recording.

"The next stage of evolution will be for the system to almost know what you are talking about," he says.

"The system itself should be able to understand the text that is being dictated and put it into order itself - it is not at that point right now, but I think in the future it could become intelligent enough to understand what type of dictation is being made."

Following a voice

Introducing voice-recognition software into a healthcare facility is not an easy task. Management usually faces resistance from skeptical physicians and secretaries who think their job might be at threat.

A savvy move is to appoint a voice-recognition ‘champion', according to Costa Mandilaras. "You have to start with a band leader within a department - it will help with any resistance," he states.

"All it takes is one person to use it and not get the results they expect and it will be bad-mouthed."

It may be tough to implement now, but voice recognition could lead to massive rewards in terms of productivity. Bill Gates might have been off the mark in 1997, but you wouldn't bet against his prediction ringing true in the next 10 years.

Add a Comment

Your display name This field is mandatory

Your e-mail address This field is mandatory (Your e-mail address won't be published)

Security code