iScribes was created by Jared Pelo, MD because of a major technology lag in voice recognition.  This lag is especially apparent when using medical terminology with voice recognition.  Wired puts it this way, “the task of providing accurate transcriptions of long blocks of actual human conversation remains beyond the abilities of even today’s most advanced software.”

Many current options are making advancements BUT they allow for an error rate of anywhere from around 10 to 20% AND that’s processing audio as a straight dictation.  This is very time consuming for a physician who may see over 30 patients in one day.  

Things get even trickier for physicians.  Medical records for patients should not and must not have errors.  That’s where iScribes comes in.  We use technology to obtain voice recordings but we rely on humans to decipher audio, type, and enter information into the EMR.  Until technology catches up, way up, humans are still necessary to decipher language, especially between two important people: the physician and the patient.

You can read Wired’s entire article, “Why Our Crazy-Smart AI Still Sucks at Transcribing Speech” here.  Look out for more blog posts in our iScribes series – Humans Do It Better.

 

Source: https://www.wired.com/2016/04/long-form-voice-transcription/

Comments are closed.