This article is more than 1 year old

Healthcare org with over 100 clinics uses OpenAI's GPT-4 to write medical records

The doctor, and their steno-bot, will see you now. Then see another patient quickly because they don't have to stop and scrawl notes

US healthcare chain Carbon Health has introduced an AI tool to generate medical records automatically, based on conversations between physicians and their patients. 

If a patient consents to having their meeting recorded and transcribed, the audio recording is passed to Amazon's AWS Transcribe Medical cloud service, which converts the speech to text. The transcript – along with data from the patient's medical records, including recent test results – is passed to an ML model that produces notes summarizing important information gathered in the consultation.

The screenshot of an example medical chart below shows what type of text the software, nicknamed Carby, generates. The hypothetical patient's information and vital measurements are included, as well as a summaries of medical records and diagnoses.


The promise ... what Carbon touts to industry

Carbon Health CEO Eren Bali said the software is directly integrated into the firm's electronic health records (EHR) system, and is powered by OpenAI's latest language model, GPT-4.

Carbon Health said the tool produces consultation summaries in four minutes, compared to the 16 consumed by a flesh and blood doctor working alone. Clinics can therefore see more patients

"The use of scribes and transcription services is standard in the healthcare industry, and a majority of patients provide consent to have their visit recorded by their provider," a spokesperson told The Register on Monday.

"Once the note is ready in our EHR, we notify the provider to review, edit, and validate the medical decision making summary as needed (approval of the visit summary is always completed by the provider). We are still refining the feature with more provider feedback."

Generative AI models aren't perfect, and often produce errors. Physicians therefore need to verify the AI-generated text. Carbon Health claims 88 percent of the verbiage can be accepted without edits.

Carbon Health said the model is already supporting over 130 clinics, where over 600 staff have access to the tool. A clinic testing the tool in San Francisco reportedly saw a 30 percent increase in the number of patients it could treat. 

Rival providers and startups working on the same problem – but which do not have immediate access to patients – are racing to sell their software to other clinics and hospitals. 

Abridge, an upstart that has built its own custom system to transcribe and summarize doctor-patient conversations, is partnering with the University of Kansas Health System in trials, for example. Ambience Healthcare, backed by OpenAI's Startup Fund, has built AutoScribe – a similar GPT-4-powered product that is being used by primary care clinics like Pine Park Health in California. ®

More about


Send us news

Other stories you might like