The University Hospital in Seoul recently presented a system of supporting clinical decisions in the treatment of precision medicine.
Why does it matter
It is said that Snuh Polaris (Precision Oncology and Sporadic Common Disease) is the first of this kind in South Korea that supports personalized treatment powered by AI and based on huge clinical and genomic data.
Powered with a huge language model (LLM), extracting, integrates and performed previously dispersed data on pathology, diagnostics, genoms, surgery and information about HES treatment.
It also includes a fresh generation cancer panel in the hospital to quickly ensure a comparison and analysis of huge genomic data, informing the diagnosis of cancer at the time of care. TAs a result, there is also cross -validation by clinical and genomic experts.
Snuh Polaris, developed by a team covering 30 departments and 87 multidisciplinary employees, It is initially used in cases of cancer, and later it will be extended to scarce diseases and chronic diseases.
Greater trend
Snoh has recently rejected his older IT systems and He moved to the private environment and the SDN -based environment to simplify and minimize the physical configuration of the network. He also adopted advanced encryption, multi -component authentication, network and final security, real -time monitoring and recovery systems after failure, and also acquired the National Information Security Certificate from the Korean government to fully protect patient data, operating continuity and regulatory regulations. These funds were recognized in the last validation for stage 6 of the world recognized all over the world Infrastructure acceptance model.
Thanks to the solid IT infrastructure, the hospital is prepared to implement more huge data sets. Lately, Snoh and his two partner hospitals in Bundang and Boramae have They consolidated their common databases to enable research and online cooperation.
Meanwhile, Snoh introduced in April what could be The first medical LLM in South Korea. Although it has not yet been formally used to support clinical trials, the study showed that the model exceeded the Korean medical licensing study and can process 50,000 words at the same time.