Speech Recognition Software in Medical Practice

Speech recognition software has arrived! Once the material of science fiction, computers are now responding to the human voice. And at Dartmouth-Hitchcock, physicians and other providers are giving up wrist splints for microphones and bringing quality and detail back to the medical record. In Cardiology, OB-GYN, Vascular Surgery, Pulmonary, Endocrinology and Infectious Diseases, physicians are using Dragon NaturallySpeaking (DNS) to bring clinically-related text into documents at rates up to 160 words per minute and with accuracy rates in the range of 95-99%. And while this technology is not yet integrated with the Medical Center's Clinical Information System (CIS), users are finding that running voice recognition software along-side of CIS allows simple transfer of free text from DNS to CIS. It is anticipated that speech recognition will be an often utilized documentation method with the transition to Epic.

Is speech recognition software a practical tool for you? To be blunt, it is not for everyone. It requires a bit of computer savvy, good quality computer equipment, a relatively quiet environment, and perhaps most importantly the ability to enunciate words fairly distinctly. It also takes a few hours of commitment in getting to know the software.

At the moment, speech recognition software is not officially supported by the institution, but this is about to change. In addition to an IRB approved clinical trial of speech recognition software, the Departmental Systems committee recently authorized the hiring of a full time support FTE for speech recognition and plans go forth with a feasibility study among 30 or more users during the late spring/early summer of 2009. For now, though, there is not official support and in order to use this technology you will have to scrap a bit and use resources like this to get started.

Frequently asked questions:

• What does it do?  Speech recognition software can do three things:  1) turn your speech into words in a word-processing document 2) allow your voice to control functions of your computer and 3) simplify common multi-sequence tasks done on your pc into a single voice initiated process through the use of macros
  
  
• Will it work in Dartmouth-Hitchcock’s CIS?  Yes, buried with CIS is a program called TX Text which is the word processing window into which you type. Version 10 of Dragon works great in this application. For version 9 it is necessary to tweak a small file within Dragon.
  
  
• What do I need to get started?  There are 3 basic requirements: 
  
  
  1. A pc with enough speed and RAM:
  • Mimimum: Intel Pentium 1GZ and 1GB RAM
  • Recommended Intel P 2.4 or 1.6 dual core and 2 GB RAM
  2. Software (Dragon 10 Medical is recommended)
  3. A good microphone (the software is bundled with a cheap microphone but it is recommended that you upgrade to something better.
     
     
• Where should I buy the software?  If possible, you should go through your practice manager. Let your practice manager know that D-H has developed a channel to obtain the Medical version of Dragon at wholesale cost. For more information about this contact Jon Wahrenberger in Cardiology or visit the web site www.SpeechRecSolutions.com
  
  
• Is it accurate?  After a very brief training period (5-10 minutes) the technology has the capacity to be phenomenally accurate (nearly 100%).  In practice, such high levels of accuracy only occur when you dictate with very clear enunciation and when in an environment free of excessive contaminating noise.
  
  
• Where can I learn more?  Here are a few sources for more information:

• An overview of Dragon 10 Medical can be viewed here:  http://www.speechrecsolutions.com/assets/Dragon10Medical_Review.pdf
• An extensive guide to using Dragon 10 Medical can be viewed here:  http://www.speechrecsolutions.com/assets/Dragon10MedicalGuide.pdf

More Links on this Site: If you or your section/practice is contemplating the use of speech recognition software instead of transcription or provider hand-typing, you can learn more using the links below.

• Beginner's Guide: An overview of voice recognition technology prepared by DHMC users with an emphasis on how to get started, including purchasing software and requirement in terms of computer, sound card and microphone.
• Building or Buying Templates to Improve Your Efficiency
• Commands for Downloading: A variety of commands ready for importation into your copy of NaturallySpeaking.
• Tips for Improving Accuracy with Voice Recognition Products
• Speech Recognition Solutions: this is a commercial site put together by a DHMC user and contains lots free resources pertaining to speech recognition, as well as a speech recognition microphones and microphone accessories well below usual prices.
• Voice Rec Web Resources: A variety of web based resources to learn more about voice recognition software and its use in health care.