Asan Medical Center has developed a speech-based clinical document system with a large-scale language model.
How it works
The system automatically records and analyzes conversations between staff and patients, near real-time, including symptoms identification and disease classification. It also creates medical records of patients on-site that are automatically saved in AMC’s EMR system AMIS 3.0.
It functions in many medical environments, including emergency rooms, wards, and treatment rooms. It is installed on your emergency cart tablet computer, so it can be used during emergency response.
LLM was trained with “tens of thousands of hours” of clinical audio data covering a variety of specialized medical terms. To improve Scribe’s voice recognition capabilities, a dedicated microphone is used to filter background noise and accurately calculate the vocal range of the speaker.
First developed in 2023 and piloted in orthopedic and orthopedic outpatient clinics, AI Scribe lives in 16 AMC departments including oncology, otolaryngology, and psychiatry. The hospital plans to gradually expand the scope of use of scribes.
Why is it important?
The clinical scribe, who was advertised as the first hospital in Korea, said We will improve the accuracy of your medical records and help you develop an accurate treatment plan. It also helps to ensure patient safety by not overlooking details or information, especially during emergency medical situations.
Larger context
The development of AI-based voice writing is part of many projects enabling digital healthcare in one of Korea’s largest hospitals. The basic IT infrastructure running these projects, including digital pathology and automation of robotic processes, has been verified in the globally recognized stage 7 of HIMSS Infrastructure adoption model End of last year.
AMC also applied recently released AI Web portal for international patientsautomatically translates eight major world languages, supporting data management and validation.
record
“A AI-based speech recognition system for treatment allows us to effectively record and store large amounts of speech-based information that are often lost during the treatment process,” said Dr. Young-Hak Kim, director of the AMC Health Innovation Big Data Center.
“An accurate symptom information that reflects the voices of medical staff and patients can be the basis for improving the quality of care and providing personalized care,” he added.
