Yoo Seung-chan, a professor at Yonsei University’s Department of Biomedical Systems Informatics, explains key features of Severance Hospital’s next-generation electronic medical records system, Y-NOT, on Dec. 18. Photo by Asia Today
Dec. 28 (Asia Today) — A research team at Yonsei University has built a generative AI-based medical record system that is now being used at Severance Hospital, aiming to cut doctors’ documentation time and allow more focus on patient care.
The system, known as Y-NOT and implemented through the hospital’s “Y-NOT” record platform, uses a large language model to draft admission and discharge notes for clinicians to review, according to Professor Yoo Seung-chan of Yonsei University’s Department of Biomedical Systems and Information Science.
“Why should we spend more time sitting in front of computers than seeing patients?” Yoo said, describing a question often raised by emergency care staff that helped drive the project.
Yoo said he began full-scale development last year as administrative burdens on medical staff intensified during tensions between doctors and the government. He said the team judged AI technology had matured enough to meaningfully reduce record-keeping workload and started development.
The project began in July last year and was deployed in clinical settings by November, Yoo said, with model development and hospital rollout carried out in parallel. He said the team focused first on achievable clinical usefulness rather than pushing only for maximum model performance.
Some medical staff initially expressed concerns, including the risk of errors in records, questions over responsibility for mistakes and worries that the system could infringe on physicians’ authority. Yoo said two surveys conducted after implementation showed those concerns eased, with especially strong satisfaction among older staff.
He said the team framed the system as supporting, not replacing, clinicians. Doctors continue to diagnose and make decisions, he said, while the AI drafts and organizes documentation for verification. Yoo added that some staff said the system made care easier because it reduced the need to manually search through past electronic records.
The “Y-NOT” system is now used beyond the emergency department, including operating rooms and inpatient wards, for broader record management, Yoo said.
The time required to create emergency room medical records fell by more than half, according to the report, dropping from 69.5 seconds to 32 seconds. Staff have said the reduced documentation burden gives them time to make eye contact with patients, Yoo said.
He said evaluations indicate record completeness and standardization have improved across care teams, including nurses, and that the time saved helps emergency physicians move quickly to the next patient or offer additional guidance to patients leaving the hospital.
Yoo said the longer-term goal extends beyond a documentation tool to an intelligent agent system designed to support safe care aligned with global standards. That direction is tied to Severance Hospital’s “Doctor Answer 3.0” project, he said, and future plans include exploring ways for patients to communicate with an AI system based on their own medical records.
Yoo said AI could help address rising medical demand tied to population aging and a decline in essential medical staff. He said it could support guideline-based care for clinicians and help patients maintain a sense of continuous connection to the hospital after discharge.
— Reported by Asia Today; translated by UPI
© Asia Today. Unauthorized reproduction or redistribution prohibited.