Twenty years ago, technologists predicted that medical transcription would be replaced by speech recognition in 3 to 5 years. Five years later, they uttered the same prediction. And that's the way it's been ever since-a rolling 3- to 5-year window. So I'm a little reticent to make a forecast on this front. But I will anyway. In 3 to 5 years, the need for traditional transcription will be a small fraction of what it is today.

The market erosion will be brought about by three factors. First, many physicians will migrate to direct entry into a PC, via template or other point-and-click methodology, as a result of the inexorable march of electronic health records (EHRs). Second, more physicians will embrace front-end speech recognition, dealing with the extra work of correcting errors to save transcription costs and to have an immediately available final report. And third, back-end speech recognition will get considerable traction, as it becomes obvious that it no longer makes sense, in most circumstances, to transcribe the report entirely-when technology can save more than 90 percent of your keystrokes, typically resulting in an overall doubling of productivity.

Of those three factors, only the last one utilizes the professional expertise of MTs. So it seems wise for MTs to begin exploring the opportunities in this arena. Given the decreasing supply of worms, it will surely pay to be an early bird. Nonetheless, this scary erosion of the transcription market will not occur overnight. And it will be somewhat balanced by the fact that many handwriting physicians will transition to dictation. But the time to make a move is soon . if not now.

There are four ways to get into this game:

·        Acquire a complete dictation system that incorporates speech recognition.

·        Integrate a speech-recognition server into your existing platform.

·        Sign up with an application service provider (ASP) that provides a total dictation/speech workflow.

·        Transmit your voice files to a speech-rec ASP, which will send back draft reports.

The first two require a substantial capital investment. For larger transcription companies and health care organizations, the second option will be the lowest-cost route to go, once the purchase price is amortized over a few years. But for smaller organizations and individual MTs, one of the ASP approaches is more sensible. Anyone who is interested in getting the names of some suppliers of any of these solutions is welcome to e-mail me their needs.

Editing drafts from speech-recognition systems, although it draws on your medical-language skills, is very different than typing out the entire report. Some slow typists will be excellent editors. And some very fast typists might find that the technology is not that helpful for them. The key to productive editing seems to be the ability to multitask. The most successful editors are able to make a correction while they continue to playback the audio and watch for other errors in the text. And they can also read ahead of the voice, to identify and correct some of the more obvious errors coming up. Of course, not all physicians are qualified for speech recognition. Those that are dysfluent, mumble and/or keep changing their minds will have to wait for further advances in the technology. And they'll continue to plague the MTs that haven't become editors. But for those dictators that qualify-and the MTs who adapt gracefully to editing-it's a very pretty picture!

Joe Weber, the inventor of Smartype, is CEO of Lexicore, provider of software and consulting related to speech and language technology for health care. He can be reached at joeweber@alum.mit.edu.