Seoul National University Hospital (SNUH) has introduced a new clinical decision support system aimed at enhancing precision medicine treatment.

Importance of the Initiative

The SNUH POLARIS (Precision Oncology and Rare-Common Disease Supporter) is recognized as the first system of its kind in South Korea, designed to facilitate personalized treatment through artificial intelligence and extensive clinical and genomic data.

This system utilizes a large language model (LLM) to extract, integrate, and refine previously fragmented data related to pathology, diagnostics, genomes, surgery, and treatment from the hospital’s Health Information System (HIS).

Additionally, it integrates the hospital’s next-generation sequencing cancer panel, enabling rapid real-time comparisons and analyses of large genomic datasets, which aids in cancer diagnosis at the point of care. The insights generated are also validated by clinical and genomic experts.

The development of SNUH POLARIS involved collaboration among 30 departments and 87 multidisciplinary staff. Initially focused on cancer cases, the system is expected to expand its applications to include rare and chronic diseases.

Broader Context

Recently, SNUH has transitioned from its legacy IT systems to a private cloud and software-defined networking (SDN) environment, aiming to streamline and reduce physical network configurations. The hospital has implemented advanced security measures, including:

• Advanced encryption
• Multi-factor authentication
• Network and endpoint security
• Real-time monitoring
• Disaster recovery systems

These efforts have led to the hospital receiving national information security certification from the Korean government and achieving Stage 6 validation in the globally recognized HIMSS Infrastructure Adoption Model.

With a strong IT infrastructure established, SNUH is poised to embark on more big data initiatives. Recently, SNUH and its affiliate hospitals in Bundang and Boramae have consolidated their databases to facilitate network-wide studies and collaborations.

In April, SNUH also introduced what may be the first medical LLM in South Korea. Although it has not yet been formally utilized in clinical research, preliminary studies indicate that the model has outperformed candidates in the Korean Medical Licensing Examination and can process texts of up to 50,000 words simultaneously.