US Clinics Utilize GPT-4 for Patient Medical Cards

American network Carbon Health Clinic I presented AI-tool for the automatic creation of medical records based on conversations between doctors and their patients.

If the patient agrees that his meeting is recorded and deciphered, the audio recording is transferred to the cloud service of Amazon AWS Transcribe Medical, which transforms speech into text. The transcript along with the data from the patient’s medical cards, including recent test results, is transmitted to the machine learning model, which creates notes that generalize the important information collected during the consultation.

Screenshot from the site Carbon Health

The representative of Carbon Health said that a tool called Carby is directly integrated into the system of electronic medical cards (EHR) and works on the GPT-4 language model. Carby collects information about the patient and vital measurements, and also summarizes medical records and diagnoses.

Carbon Health said that the tool is a resume of consultations in 4 minutes. With manual notes of a doctor, this process takes 16 minutes. Thus, clinics can take more patients. After creating a note in an electronic medical record, doctors need to check the generated text. Carbon Health claims that 88% of the formulations can be accepted without amendments.

According to Carbon Health, the model already supports more than 130 clinics, where more than 600 employees have access to the tool. According to reports, the tool helped one of the network clinics increase the number of patients taken by 30%.

/Reports, release notes, official announcements.