How to make speech recognition a documentation benefit, not a bother
What role can speech recognition play in developing quality documentation? On one hand, proper use of the technology can be a boon to productivity and efficiency, saving doctors’ time and saving facilities money. On the other hand, misuse of the technology can have the opposite effect, wasting physicians’ time and causing rifts between hospital leadership and medical staff.
Speech recognition is a proven technology that has been used in the healthcare industry for decades. Most solutions are trained to recognize the specific vocal patterns of clinicians and use that recognition to analyze recorded clinical dictation. The analysis results in a written document that, once edited, can be added to the medical record. This process can save healthcare providers money because it decreases the amount of time a transcriptionist needs to complete a record.
Sounds great, right? The problem comes when transcriptionists are taken out of the equation. Many of today’s electronic medical records include a speech recognition component. We know of hospitals that have severed relationships with transcription services in favor of using EMR speech recognition, then required attending physicians to edit their own transcription.
Physicians certainly have the ability to edit their own transcription, but is this the best use of their time? In our experience, doctors would rather be spending time with patients than doing clerical work.
We recommend facilities utilize speech recognition with the help of a transcription service. It still saves time and money, it results in accurate documentation, and it limits the time a doctor needs to spend reviewing his or her text. There are three phases of such as service:
Phase 1 – Full transcription plus speech recognition training: During this phase, transcriptionists review and transcribe recorded dictation, while the speech recognition system is trained to recognize each physician’s speech patterns. It “learns” by analyzing the voice of the physician compared with the completed and accurate transcription.
Phase 2 – Editing transcription plus speech recognition training: Taking a step closer to full speech recognition, physician recordings are analyzed by the system and a draft document is produced. The draft document is reviewed by the transcriptionist, who compares the recording of the physician with the system-generated document. The transcriptionist then edits the speech-to-text document and produces a complete transcription, making sure the final document is grammatically and functionally correct and based on the facility’s standardized templates. The completed transcription is sent back to the voice recognition system for the second level of training.
Phase 3 – Front end speech recognition: In this final phase, the speech recognition system is fully trained and produces its most accurate document. The system analyzes physician recordings, then produces an editable document. The document only needs minimal editing by the transcriptionist, who sends the document to the physician. The physician then reviews, signs and approves the document.
It should be noted that current speech recognition technology won’t be able to translate voice into text with 100 percent accuracy, hence the need for an edit by a transcriptionist. Also, some voices can’t be recognized by these systems with a high level of accuracy. We estimate that that only two out of three physicians will be able to progress to Phase 3.
Speech recognition can be a tremendous time saver. And, used in the right way, it can save organizations’ money without wasting physicians’ time.
Photo credit: Denise Krebs (cc)