Seoul St. Mary's Hospital Launches Development of Multimodal AI Model for Optimal Liver Cancer Treatment

Seoul St. Mary's Hospital, Catholic University of Korea, announced on April 21 that Professor Han Jiwon and her research team in the Division of Gastroenterology are launching a full-scale project to develop a multimodal artificial intelligence (AI) model designed to predict prognosis and recommend optimal treatment strategies for patients with hepatocellular carcinoma (HCC). The team has been selected for the Young Researcher (Type B) category of the 2026 Individual Basic Research Program funded by the Ministry of Science and ICT and the National Research Foundation of Korea, and will receive up to KRW 600 million in national research funding over five years.

Professor Jaewon Han, Division of Gastroenterology, Seoul St. Mary's Hospital

Liver cancer is a serious disease with a poor prognosis, ranking second only to lung cancer in cancer-related mortality in Korea (based on 2024 Statistics Korea data). In recent years, the introduction of immunotherapy—specifically, immune checkpoint inhibitor combination therapy—as the first-line standard for advanced liver cancer has expanded treatment options and improved outcomes. However, the objective response rate (ORR) remains at only about 30%. This means that 7 out of 10 patients do not experience a tumor response. Despite this, there is still no reliable clinical model or biomarker that can predict therapeutic efficacy in advance, making this a major challenge in clinical practice.

The research team will develop a multimodal AI prognosis prediction model that integrates and analyzes diverse heterogeneous data such as imaging, pathology, and clinical information, under the core concept of “from big data to deep data.” Using multicenter clinical cohorts and biobank resources from the Catholic Medical Center network, the team will apply deep learning and machine learning methods to extract high-resolution, patient-specific clinical information that is difficult to capture with conventional big data approaches. This will enable risk stratification and prediction of treatment response before therapy, facilitate optimal treatment planning for each patient, and is expected to help reduce unnecessary side effects.

Professor Han Jiwon, who is leading this study, is a specialist in liver diseases such as liver cancer and transplantation, as well as a physician-scientist with a Ph.D. from KAIST Graduate School of Medical Science and Engineering. She has continued convergence research that bridges clinical medicine, immunology, and artificial intelligence, having been selected for national programs such as the Ministry of Health and Welfare’s Global Physician-Scientist Training Project and the K-Medi Convergence Talent Training Program (for young physician-scientists).

Professor Han Ji-won's Liver Cancer AI Research Image. Seoul St. Mary's Hospital

In 2025, Professor Han plans to release a machine learning-based decision support model in JAMA Network Open, based on a nationwide Korean liver cancer cohort of 3,915 patients and an external validation cohort of 614 patients from Seoul St. Mary's Hospital. The model demonstrated that the risk of death was reduced by approximately 54% compared to actual clinical decision-making, accumulating internationally validated results from prior research. Additionally, in 2024, she published a machine learning-based prognostic model for advanced hepatocellular carcinoma in Clinical Cancer Research, and last year she developed and filed a patent for an AI technology that predicts immunotherapy response using PD-L1 CPS based solely on non-invasive blood test data. In January of this year, she published a study in PLOS Medicine that compared treatment recommendations from large language models (LLMs) such as ChatGPT, Gemini, and Claude with real-world physician decisions and survival outcomes in a nationwide registry of 13,614 patients, thereby clarifying both the clinical potential and limitations of LLMs using nationwide data.

Professor Han stated, “The ultimate goal of this research is to leverage AI to comprehensively analyze each patient’s high-resolution clinical information and present the most suitable treatment strategy for every individual. In five years, we aim to complete an AI-based prognosis prediction platform that can be used in real-world clinical settings to provide direct benefits to liver cancer patients.”

The team aims to publish multiple SCI(E)-level papers and file domestic and international patents within the five-year research period, and plans to expand the developed AI methodology to other gastrointestinal cancers beyond liver cancer.

This content was produced with the assistance of AI translation services.