medical billing transcription supplier information

Next generation of voice recognition products comes with gentle learning curves and integration to clinical workflow.

Discard any preconceived notions or past experience with voice recognition (VR) and what it can do for your practice. Once a "not ready for prime time" oddity, voice recognition technology has become a productive citizen.

Early VR products in healthcare had several shortcomings. First and foremost, they were not integrated into the clinical workflow. Medical documentation generally follows a predictable flow from front office sign-in to chart review, patient encounter and documentation. Early VR systems simply sat on a stand-alone computer where a physician had to walk over and dictate, apart from workflow. There was no interface to existing systems, which created inefficiencies and slowed physicians down.

Voice recognition software works by hearing the spoken word and converting it to text by making an "educated guess" at what the user is saying based on an individual voice model and associated vocabulary. Earlier voice engines (and many still today) used a medium such as Microsoft Word or WordPerfect to display and store the text. The second problem arose with these forms and the inability to organize them in a database format. Dictated and completed notes were stored in Windows Explorer and provided no organized access, useful search ability, means for electronic distribution such as faxing or e-mail, or anything more than transcriptionists normally supplied.

Third was the problem of editing limitations. Voice recognition requires an edit process to improve. As dictated text is edited for accuracy, any changes update and improve the user's voice model. Most VR products digitally record and save the user's voice and synchronize it with the created text as they dictate. This is done to create an audible reference. If the user said "the lumbar spine" and the system interpreted it as "the lumber is pine," there would be a recording associated to the incorrect text that would allow the user to understand what was actually said and make the correction.

In earlier products, editing for recognition accuracy had to be completed immediately after dictation; otherwise this recorded audio file was lost. In addition, dictated text and audio files could not be moved across a network so the physician had to complete the editing tasks himself, instead of delegating it to a staff member.

Finally, the learning curve was unbearable. Early models required two or three hours to "enroll" or create a voice model and there was virtually no training available. Many were discrete speech, which meant a user could not dictate at normal conversational speeds, but had to pause between each word. Training consisted of a user's guide or a video, and users had to self-train and self-implement.

Early products were designed for the masses and generally sold off-the-shelf; they were not designed for the specialized healthcare utilization. Even those designed for healthcare only offered a preloaded medical vocabulary of terms.

Today's Difference

Today some VR systems can interface and share data with legacy billing and electronic medical records (EMR) systems. This is significant for several reasons. First, when starting any note, the physician typically identifies the patient or chart number, visit date and type, and maybe a carbon copy reference. Some systems today can import this information from the billing system and shorten dictation time by eliminating dictation of redundant data.

Second, some current systems can export dictated or captured information such as charge codes back to the billing system and can export completed dictation into an existing EMR system. When interfaced with the electronic patient scheduler, they can also manage the outstanding dictation queue, thereby eliminating the possibility of missed dictation.

Import of demographic data elements, including referring physician information, can be downloaded from the practice management system. Dictations can then be automatically faxed or e-mailed to a referring physician or colleague after dictation is complete.

Some systems today can provide useful and timesaving dictation templates. These forms or note templates should be in some database format to maximize their benefit. The application also should be able to provide voice macros whereby the user can give a short verbal command and have repetitive statements or preformatted text inserted into the note automatically. Macros can also contain information that has been downloaded from the billing system.

# posted by Medical Information : 1:06 AM