My magnificent obsession

I’m obsessed with speech recognition technology, but I am not alone. As Claire Betis, marketing manager at Crescendo Systems Corp. noted recently, speech recognition technology is finally fulfilling the promises we’ve been long awaiting.

A line from Military Medical Technology used to say it all when it came to speech recognition. “A simple sentence — much less one with medical lingo like “granulomatous” — came out looking like a kid had played Mad Libs with it.”
Although, there has been vast improvements in the technology there still remain problems. Crescendo, in fact, created a blog to address some of the problems. The whole point of speech recognition is to save time, and that is still the biggest challenge.

Doctors have experienced plenty of troubles with dictation. The technology doesn’t particularly like the use of “the” or “a” because it confuses it with a pause or an “um” and will just delete it. (Ironically speech technology is always advertised as deleting pauses and noises from the environment as if it is the best function). It also has trouble with heavy accents and people who may talk too fast and slur their words together.

There is also the debate about front-end and back-end solutions. Many like using front-end because they can edit their reports themselves. Yes, it takes time, but they are comforted by the fact that their report is accurate. Others like the back-end solution that calls on an editor to make corrections. However, sometimes the process of getting a note back from the editor, reviewing it and sending it back to the physician for review can take a week.

No doubt speech recognition that is incorporated into an EMR is very efficient and saves time. It is like talking into a file and then having it file itself, a file that can be found by the right person exactly when needed. But, again, nothing is ever as easy as it sounds.

There is also a little thing called cost that goes with this technology. A back-end service can cost around $1,800 a month to maintain. Speech software costs somewhere around $900 and recorders around $150 each. Not to mention costs associated with computer upgrades or a computer! Doctors tend to have a “fear of technology.” The biggest problem that speech recognition may face is buy-in by doctors.

- Molly Merrill