Last Monday (2 December), Amazon launched a new medical transcription service called Transcribe Medical. According to the tech giant, this machine learning-driven automatic speech recognition (ASR) tool picks up conversations during practitioner/patient consultations as well as medical and pharmacological terms used in dictated notes or telemedicine and turned them to text for the purpose of clinical documentations or further analysis via natural language processing (NLP).

Rationale behind introducing voice recognition tool

The rationale behind Transcribe Medical is the fact that physicians are spending too much time performing detailed data entry into the electronic health record (EHR) systems. As Amazon highlighted in a separate article, on average, clinicians have to spend an additional six hours on top of their daily medical responsibilities on manual note taking. This deprives them from attentive patient engagement as consultations often have to take place in a hasty manner. In the long run, this also contributes to burnout and decrease job satisfaction.

On the other hand, engaging in third party transcription services can be expensive and time-consuming. Likewise, it may not be ideal to have a human scribe sitting in for consultations as some patients may find it uncomfortable or even disruptive. Existing transcription software lacks the knowledge to accurately comprehend complex medical terms and may not necessarily fits well into physicians’ workflow.

As such, Transcribe Medical is trying to bring about a change. Since it’s able to be integrated into any voice-enabled application or device which comes with a microphone, and one does not have to vocalize explicit punctuation commands, hopefully, clinicians can deploy it at their comfort level.

EHR and health information supplier, Cerner had already established a cloud collaboration with Amazon Web Services (AWS) this summer and it will also be one of the first Transcribe Medical customers. Amazon noted to all its customers to encrypt all protected health information when using the service as the company will not be using any of these data. At the moment, Amazon is charging $0.0004 per second for the use of Transcribe Medical.

Prospective challenges

It’s certain that ASR and other voice recognition tools may result in lesser paper work and save up time for physicians to pay closer attention to patients’ needs. However, continuous validation is still required especially in an environment where ambient noise is unavoidable. Hospitals and clinics host not only patients and staff but also equipment and all of which contribute a certain amount of sounds which may affect the performance and even accuracy of ASR.

Likewise, a diversity of medical staff and patients mean the ASR needs to be resilient towards all kinds of accents or accuracy and error may still pose new challenges to physicians who are using or dependent on the tool.

A workshop solely dedicated to NLP and other artificial intelligence (AI) driven voice recognition tool will be held during this year’s AIMed 19. Do join us to find out more about how these innovations impact medicine and healthcare.

Session Focus: Workshop 5 – Natural Language Processing in healthcare

When: Friday, December 13th 2019 (08:00 – 09:30)

A comprehensive session to learn about NLP; Natural Language Understanding (NLU), and other speech recognition tools employed in medicine and healthcare.

Attendees will gain the following knowledge:

Uncover the potential of voice/ conversation AI in medicine and healthcare.

Understand the challenges in implementing and deploying NLP tools in clinical settings.

Insights into building NLP and NLU tools.

Review existing NLP and NLU applications and closely examine the kind of clinical challenges they can overcome.

Moderator:

Piyush Mathur. Staff Anesthesiologist/ Intensivist, Quality Improvement Officer, Cleveland Clinic.

Speakers:

Louis Ehwerhemuepha. Data Scientist, Children’s Hospital of Orange County.

Jai Nahar. Pediatric Cardiologist, Children’s National Health System and Associate Professor of Pediatrics at The George Washington University School of Medicine and Health Sciences.

Author Bio

Hazel Tang A science writer with data background and an interest in the current affair, culture, and arts; a no-med from an (almost) all-med family. Follow on Twitter.