Medical Speech Recognition
By Richard P. Jovanovich
May 5, 2015
There are a small group of manufacturers of speech recognition specifically for the medical market so therefore not a lot of options to consider. The best fit for a provider is somewhat dependant upon the environment they will work in. Some of the issues that should be considered include: mobility requirements, background noise in area, network level, internet speed, workstation computer specifications, other applications running on provider computer, IT staff knowledge level, user computer comfort level, do they have a transcription staff, and a number of other considerations that should go into making a solution choice.
Mobility is a priority for a number of physicians as they may practice at the hospital, at home, at their office, etc. For these physicians you should prefer an application that has a user license that can be utilized at any location. You also would want the physicians profile to be accessible from any of these locations as otherwise you cannot take advantage of any of the learning that goes on or any new words that have been added to your personal vocabulary. Also, you would not have access to all of the templates that you use.
In terms of noise level in the surroundings where you are using the systems some microphones supply noise suppression technology and some medical speech recognition companies provide within their applications further noise suppression. This would primarily come into play in some Emergency Room settings or possibly an outpatient clinic.
The internal network would have match or exceed the specifications of the system which is somewhat dependent on the amount of network traffic you would have at your organization. This would have to be tested at peak times during the day to provide a worst case scenario. If the network does not meet these specifications then it opens up to a number of possible problems which all would affect a physician’s productivity thus timeliness of chart completion and billings.
Another area that needs attention are the workstation specifications as some voice recognition situations are so called “memory hogs”. Not only do you have to be concerned with the medical speech recognition application but any application running on this workstation including your EMR.
The Information Technology staff supporting the above computers/networks will also have a part in the success or failure of any voice recognition solution for the reasons listed above. They must make sure that the application will also function with the EMR. With some medical speech recognition solutions they are cloud based or hosted so these concerns are basically eliminated as all that is required is an internet connection so thus simplifies matters.
Another possible differentiator between Medical Speech Recognition Solutions is that of ease of use for the provider. As some providers are not very comfortable with the use of applications on a computer the less keyboard touches the better. In these cases a physician would find it helpful to utilize macros and templates in order to save time and reduce frustration. The macros would enable you to create short phrases that expand automatically into paragraphs or pages of a report. Report templates can be created in a system which allow you to simply fill in the blanks or fields in certain reports.
Some healthcare facilities have their own transcription staff that transcribe reports for the physicians thus eliminating them from having to do this work. In this situation you want to utilize what is called backend voice recognition which improves the transcriptionists’ productivity and has no effect on the physician. In this configuration it is more like traditional dictation to the physician where they dictate and the dictation shows up on the transcriptionist’s computer where the transcriptionist can hear what is being said and compare it to what the speech engine has interpreted it as on the computer screen where the words are displayed automatically. Any corrections would be made by the transcriptionist. Ideally for a system with the most flexibility you would prefer a speech recognition system that can do what is called front end where the physician does all the corrections and/or backend (above example) where the transcriptionist does all the corrections.
In summary, the above discussed capabilities would give a provider or facility a pretty good idea of what to look for in a solid medical speech recognition solution. One last thing which is very important is the training you would receive from any vendor you would be dealing with as this is critical when relating to the success or failure of any medical speech recognition project.