Learning to Love Speech Recognition Technology

Perspectives

Published/Updated Date: March 11, 2020

Recently, a Cancer Registrar knocked on the door to the Medical Records Manager’s office, and said: “I can’t stand it anymore!  Can we please let the transcriptionists know that histology is not spelled, ‘hystology’?”

The Medical Records Manager agreed to address the problem, indicating that a machine (not a person) would be given “re-training”.

With all that technology promises, it certainly has its limitations.

Here are some examples of machine transcription bloopers and what should have been translated/transcribed:

  • “Consideration for biopsy of the abdominal thyroid nodule could be considered.” / “Consideration for biopsy of the ABNORMAL thyroid nodule could be considered.”
  • “Oral internal fixation of proximal humerus” (ouch!). / “OPEN internal fixation of proximal humerus”.
  • “Patient with prostate cancer. Discussed wrist, benefits and alternatives.” / “Patient with prostate cancer. Discussed RISKS, benefits and alternatives.”
  • “Not really instituted right now.” / “Not really INTERESTED right now.”
  • “Patient currently works to full time jobs leads at 7 in the morning for his day job but operating company.” / “Patient current works TWO full time jobs; LEAVES at 7 in the morning for his day job AT BOTTLING company.”  (maybe?)
  • “Will consider RT to pelvic notes” / “Will consider RT to pelvic NODES”.
  • “Near obstructing cervical chills tumor”. / “Near obstructing cervical chills tumor”.  (Still pondering this one).
  • “Age 3 adenocarcinoma”. / “STAGE 3 adenocarcinoma”.
  • “Ulcerative sleep apnea”. / “OBSTRUCTIVE sleep apnea”.
  • “It peers increased in size”. / “It APPEARS (to be) increased in size”.
  • “Focal extra-prosthetic extension is identified in the left prostatic base”. / “Focal extra-PROSTATIC extension is identified in the left prostatic base”.
  • “Achilles Heal”. / “Achilles HEEL”.
  • “Referred by Dr. Smith, to houma report will be sent”. / “Referred by Dr. Smith, to WHOM A report will be sent”.
  • “Splincter Tone Normal”. / “SPHINCTER Tone Normal”.
  • “Launch will waiting for BPH”. / “WATCHFUL WAITING for BPH”.
  • “Does note that the smell orthotic boot maker nauseous.” / “Does note that the smell OR THOUGHT OF FOOD MAKES HER nauseous.”
  • “Patient was waited for the source of her confusion”. / “Patient was EVALUATED for the source of her confusion”.
  • “Ending pain sing masses seen in posterior aspect of the uterus”. / “ENHANCING masses seen in posterior aspect of the uterus”.

Probably not the machine’s fault:

  • “Consistent with carcinoma and appears indeterminate.” 
  • “Sessile polypoid mass in bladder”. 
  • “History of C-Section” (Male patient).
  • “Brother deceased x 2”.

Have you had any of your own speech recognition failures?  Example – creating a text message via speech recognition – how often does your phone transcribe a message with zero errors?  It doesn’t happen often for me resulting in many humorous messages to others!

On my wish list: A T-shirt that says, “Made Using Speech Wreck Ignition Technology”.

Over time speech recognition will become more and more perfected.  In the meantime, try to find the humor in technology – laughter is the best medicine!

Image by Tumisu from Pixabay

Related

Breaking the Cycle of “Keeping Up” in Registry Management

Breaking the Cycle of “Keeping Up” in Registry Management

For many healthcare organizations, registry operations become centered on keeping pace with today's work instead of improving how the work gets done. Production continues, and reporting requirements are met, but there is rarely enough time to evaluate workflows....

The Hidden Cost of Inefficient Cancer Registry Workflows

The Hidden Cost of Inefficient Cancer Registry Workflows

Healthcare organizations depend on cancer registry data to evaluate outcomes, monitor quality, support research, meet accreditation requirements, and guide strategic decision-making. The value of that data depends on its accuracy, completeness, and timeliness, making...

Supporting Stroke Programs with Efficient Registry Operations

Supporting Stroke Programs with Efficient Registry Operations

Every minute counts during a stroke. Fast recognition, coordinated treatment, and adherence to evidence-based care are critical to improving patient outcomes. While clinicians focus on delivering life-saving care at the moment, improving outcomes for future patients...

Stay in Touch

Get our latest articles, registry insights, and company news straight to your inbox.

This field is for validation purposes and should be left unchanged.