Archive for the tag 'Dragon'

How to Train Your Dragon

As physicians migrate to the Electronic Health Record, there are many new systems and processes they have to learn and adapt to.  One of these systems is voice recognition software, such as Dragon Medical 10.  I have worked with some physicians recently who were implementing a new EHR in their office, as well as transitioning from a transcription service to Dragon voice recognition.  This introduced some new challenges which I hope to shed some light on in this article.

Dragon Medical has the ability to ‘Type as you Talk’, which allows the user to dictate blocks of text and see this appear in their note in the EHR.  This has a huge benefit to the provider by allowing them to review and finalize their documentation for the visit immediately, rather than waiting a few days to receive the note back from a transcription service.  We discovered that there are some steps that you can take to improve the performance and/or accuracy of Dragon.  Here are a few to note:

  1. Spend the time training Dragon to recognize your voice.  During this process, the application will learn how you speak, and adapt to your voice patterns.  This will prove to be very beneficial in the long run.
  2. Follow the recommendations for the settings for your EHR vendor.  The Dragon representative will have recommendations on how settings should be configured based on the EHR you are using.
  3. When words are not typed correctly, correct them using the built in features of Dragon to Train it on how you speak those words.  This will save you time and energy as you become a more advanced user of the speech recognition software.
  4. Have reasonable expectations.  Dragon is a tool that improves over time.  When you first begin using speech recognition software, it only has a basic understanding of your vocabulary and how you speak.  It will take time for the application to improve, which will occur naturally as you train it when words are not recognized correctly.

These are a few items that will hopefully help you be more successful when using speech recognition software, such as Dragon Medical.  I have also found that it is beneficial to have a follow-up training session with Dragon after the user has been using it for a few weeks/months.  At this point, the user understands some basic functionality, and is usually interested in how to do more complex functions such as Macros.

Dictating with Dragon

We recently brought our first client live on the Community Health Record using a Stark-donated implementation of TouchWorks.

At one client, who is donating TouchWorks under the Stark exception, we’re using Citrix to present the EHR to the users. Because of the resource requirements for Dragon (2GB+ RAM), we’ve installed it locally on the client PC’s instead of on the Citrix server and are just using Citrix for the EHR.

Using Dragon directly in the EHR works ok, but the accuracy and consistency is better when using the dictation box. Unfortunately when using the EHR in Citrix and Dragon locally on the PC, the ‘Transfer’ button in the Dictation Box is not able to determine where in the application to put the dictated text. The Phillips SpeechMike can be configured to solve this problem of getting the dictated text from the Dictation Box into the TouchWorks application with minimal effort.

Configure the SpeechMike’s fast forward button so that the following commands are issued: – Ctrl-A, Ctrl-C, Tab, Tab, Enter

This selects the text from the dictation box, copies it to the clipboard, and then closes the dictation box.

So the entire work flow would be something like this:

  • Press the ‘Insert Overwrites’ button on the microphone to bring up the dictation box
  • Press the record button to dictate
  • Press the fast forward button
  • Click in the appropriate field in the EHR
  • Using the microphone say “paste that”

Of course you can always dictate directly into TouchWorks, but the commands don’t seem to work as consistently as they do in the Dictation Box.

Thanks to Melissa Heath for the SpeechMike configuration.