As hospitals increasingly adopt artificial intelligence (AI), a recommendation emerged to build "AI governance" that covers data management, legal regulations, and accountability systems, rather than focusing solely on the performance of AI models.
Law firm Barun (managing attorney Lee Dong-hoon) said on the 9th that it co-hosted the "1st AI Innovation Leaders Forum — a great transformation and leap in hospital AI: from data and governance to clinical solutions" with Galeb & Company on the 8th at El Tower in Seocho-gu, Seoul. About 100 people, including management, planning, and information technology (IT) officials from general hospital-level or higher medical institutions, attended the event.
Yang Gwang-mo, a professor at Sungkyunkwan University School of Medicine (head of the AI Research Center at Samsung Medical Center), cited data disconnection among institutions, tasks, and evaluations as a reason hospital AI fails to take root in actual operations even after performance verification. Yang said, "After AI makes a judgment, who will handle it, where, and how must be designed in advance," emphasizing the need for an operational system that achieves consolidation of data and AI judgments, task execution, and performance evaluation.
Attorney Ahn Ju-hyun of Barun explained that hospital AI can be subject not only to the Basic Act on Artificial Intelligence but also to various laws simultaneously, including regulations related to digital medical products and the Personal Information Protection Act. Ahn noted, "The object of regulation is not the model itself, but the combination of data processing and clinical use," and said that not only at the time of adopting AI but also during operation and changes, matters related to personal information, medical devices, and the Medical Service Act must be continuously checked.
Bae Gi-won, head of the Innovation Center at Galeb & Company, recommended first identifying the AI used inside hospitals and then differentiating approval and verification procedures according to risk levels. The explanation was that reviewing all AI at the same level could increase the use of so-called "shadow AI," which operates outside internal controls.
Cases of applying AI in clinical settings were also introduced. Aination presented a stage-by-stage AI detection solution for pressure ulcers co-developed with Severance Hospital, and ZUNGWON EN-SYS separately presented technology that detects lesions in real time in endoscopic images and assists diagnosis.
In the subsequent panel discussion, participants said hospitals need a system that clarifies approval, verification, and monitoring procedures and responsible persons according to the risk level of AI used, and that records clinicians' final decision-making authority and AI change and incident response processes.
Barun managing attorney Lee Dong-hoon said, "To successfully introduce medical AI, a comprehensive response is needed not only to AI regulations but also to various laws and systems such as digital medical products and medical devices, personal information protection and the use of medical data, and the Medical Service Act."