- Researchers at Google said recently speech recognition technology can be practical for transcribing medical conversations.
- The researchers stated their two disparate research models achieved a word error rate of 20.1% and 18.3%. "[O]ur research shows that it is possible to build an [automatic speech recognition] model which can handle multiple speaker conversations covering everything from weather to complex medical diagnosis," they said.
- The company plans on working with physicians and researchers at Stanford University to gather more information on how speech technology can assist in physician note-taking.
It's no secret that physicians dread the EHR.
Evidence is piling up that EHR places more administrative burden on physicians and keeps them from interacting with a patient. Some have hired scribes to help with tasks such as note-taking and charting.
Speech technology has been in the healthcare field for quite some time, but a dominant player hasn't emerged where physicians are rushing to use its technology. Users have to "train" the technology to understand their voice, and even then a person will have to correct mistakes the computer made. To be fair, physicians everyday speech can be difficult as it includes medical language.
Google's researchers state they have cracked the code to identify multiple speakers and understand medical terminology. While more research is needed, the implications of a useable dictation technology for physicians would be welcomed by the community. If it works, then physicians could have one less administrative task to worry about and more time to focus on what should be in front of their faces: the patient.