To Type or Talk? Will cloud-based voice-recognition applications put an end to doctors typing at the bedside?

Voice recognition is a technology that has been bubbling below the surface for several decades. The inclusion of the Siri voice recognition system in the iPhone 4S seems to many a signal that the time has come for voice recognition to be adopted on a wide-scale basis. Physicians have been using voice recognition for some time now, but many have seen it as the least-worst option for docs who have never learned to type and don’t have the time or inclination to do so. As voice recognition gets better, however, there may be a point where even doctors who are fluent typers will switch over to dictation using voice recognition software.

One company that I talked to at HIMSS has taken advantage of a cloud-based API from Nuance (the makers of the Dragon suite of voice-recognition software) to come up with a new iPad based information system for use in emergency departments called SparrowEDIS. Designed by emergency medicine physician Dr Brian Phelps and his team at Montrue Technologies, the system has just won the Nuance 2012 Mobile Clinician Voice Challenge.

The software is, at heart, a unique iPad-based user-interface that enables doctors to enter narrative data into the hospital’s EHR system by first communicating with Nuance’s Healthcare 360 cloud-computing system (to translate the dictation into text) and then, through the hospital’s EHR interface engine, sending the text into the EHR. On top of this functionality, users can also check prescription interactions, order new prescriptions and share discharge instructions. The team have put together a video that shows off the system and the thinking behind it:

I can see lots of doctors being attracted to this system. Doctors who can’t type quickly and work in environments where they are required to be on the move are obviously the most likely to benefit. However, I can see that even doctors who can type will start to see the benefits of voice recognition in terms of maintaining eye contact and rapport with patients as they perform their history taking and examination. Hopefully, Dr Phelps and his team will be publishing some results from their pilot and future trials to see whether or not this type of applications marks “the end of typing” or whether the younger generation of doctors, who can already type quickly, prefer to type than talk.

What do you think? Is mobile voice recognition the future of medical narrative data-entry or is it just a transitioning tool for doctors who can’t type and don’t want to learn?

Views: 635


You need to be a member of Health Informatics Forum to add comments!

Join Health Informatics Forum

Comment by Bob Hoyt on March 4, 2012 at 8:11am

Speech recognition continues to improve and be adopted by physicians as one of the main methods to input data during a patient encounter. We published results of a relatively large speech recognition implementation in a military treatment facility (MTF). While it was well received overall, we had a failure rate of about 30%. This was due to a multitude of factors (inadequate training, perception of low value, etc) that would likely be true of any technology adoption. Physicians do complain that they hate to proofread and make corrections, which is why some institutions have a server based speech recognition solution so transcriptions can "clean up obvious errors and insert the text into the EHR.  The good news is that the technology continues to improve (but not the price)....Bob

Comment by Brian Phelps on March 1, 2012 at 8:57am

Thanks John, that's exactly what we do with Sparrow EDIS. Our award was for our implementation of speech, so that's what all the focus has been on. But we only use speech for the history of present illness and medical decision making—the two areas that are not possible to template, and which are always unique to the patient. Everything else, including CDS, med reconciliation, prescribing, order entry, and discharge planning, uses tappable templates. Dictation of the HPI tends to happen as the physician is moving to the next patient, and dictation of MDM tends to happen once results have returned. It is possible to achieve a balance between structured and unstructured data with a combination of taps and speech, never using a QWERTY keyboard.

Comment by Chris Paton on February 28, 2012 at 9:17am

Hi John,

Thanks for the comment. I hadn't thought about the fact that they might not be looking so closely at the tools - it's a good point. Being able to look up from the screen does mean that they could be looking at the patient while they are taking the note though.

At the end of the day, we need some solid empirical evidence about which methods of data input work best from a cost-effective patient care point of view. The results may be surprising for the reason you mention, or many other reasons that we haven't thought of.



Comment by John Sharp on February 28, 2012 at 9:09am

I think doctors should use templates and smart data elements which make data entry easier rather than depending on text to speech technology. If they are dictating, they are not necessarily looking at the patient record or using clinical decision support tools.


Advertise on the Health Informatics Forum


Advertise on the Health Informatics Forum

New Media Medicine

This website is published by New Media Medicine Ltd.

New Media Medicine develops new media solutions for the healthcare industry.

As well as the Health Informatics Forum, we operate a discussion forum for medical students and doctors:

We also operate the eLearning Talk social network for elearning developers.

© 2014   Created by Chris Paton.   Powered by

Badges  |  Report an Issue  |  Terms of Service