The Ministry of Science and ICT launches the AI-specialized foundation model project. Led by Lunit, 23 domestic institutions including SK Biopharm, KAIST, and Seoul National University join to build an AI-based medical science research and development ecosystem./Courtesy of microsoft

Medical artificial intelligence (AI) corporation Lunit said on the 31st that it was finally selected as the lead institution in the medical science field for the national strategic project "AI specialized foundation model project," promoted by the Ministry of Science and ICT.

This project is a government initiative to build a large-scale AI ecosystem, with a total budget of 18.21073 billion won. Lunit, as the lead institution and overall technology coordinator for this project, has pushed it forward by forming a consortium from the initial stage.

The consortium is composed of 23 entities in total: seven corporations, including Trillion Labs, AIGEN Sciences, SK Biopharm, Kakao Healthcare, Standigm, Rebellions, and D-Circle; six research teams from KAIST and Seoul National University; and nine medical institutions, including National Health Insurance Service Ilsan Hospital, Yongin Severance Hospital, and Kyung Hee University Medical Center, establishing an industry-academia-research-hospital cooperation system. With this selection, the 23 institutions will carry out the project together.

The project is titled "Development of a multiscale medical science specialized foundation model for end-to-end medical science innovation from molecules to populations." A foundation model refers to a basic AI model that learns vast amounts of data and can be adapted to various tasks. Simply put, it means developing AI that links diverse medical information and integrates its use.

Until now, knowledge systems at each stage in the medical science field have been disconnected from one another. Although there were AI models specialized by stage, there was no integrated agent system that could use them comprehensively. Because of that, each line of research faced limitations in assembling a chain of evidence that connects to clinical success.

The goal of this project is to develop a foundation model that internalizes evidence facts by linking various omics information analyzing biological macromolecules—such as genomics, proteomics, and metabolomics—with drugs, medical science papers and guidelines, and clinical knowledge. Based on this, it will build six application agent systems as autonomous AI. An agent is a software system that uses AI to perform tasks on behalf of a user.

Lunit explained, "We will receive government support for GPU (graphics processing unit) infrastructure resources worth about 17.4 billion won and conduct research and development in two phases through September 2026," adding, "We aim to secure a level of technology that can be commercialized immediately upon project completion."

Lunit plans to develop a medical science specialized foundation model with 32 billion parameters and to release this model, which integrates end-to-end knowledge data from molecules to the clinic, as a commercial open source. It also said it would build an autonomously operating agent system based on this and expand it into various application services.

After development is completed, it plans to conduct demonstration studies at 13 institutions in Korea (11 hospitals and 2 pharmaceutical institutions), expand distribution to tertiary general hospitals and pharmaceutical companies nationwide, and ultimately pursue the global spread of Korean medical AI.

Seo Beom-seok, CEO of Lunit, said, "As the lead institution of this project, Lunit has led the preparation and will continue to lead the consortium to produce innovative results," adding, "Based on the end-to-end AI technology and global network accumulated through this project, we will continue to expand our business beyond Korea to overseas, while realizing medical innovation that the public can feel."

SK Biopharm will be responsible within the consortium for AI-based drug development and development of a Digital Twin model. A Digital Twin is a technology that implements the real world in a computer virtual world. The company said it can run simulations of new drugs using a human Digital Twin.

This project will be led by SK Biopharm's AI/DT (artificial intelligence/digital transformation) Center. The AI/DT Center is a CEO-direct organization that oversees AI acceleration and digital transformation, including a research and development (R&D) platform for drug development, data-driven decision-making tools, and the introduction of AI work support systems.

Lee Dong-hoon, president of SK Biopharm, said, "Focusing on the AI/DT Center, we will work with domestic AI-leading institutions such as Lunit to accelerate AI-based drug development and Digital Healthcare innovation, and further strengthen our technological competitiveness."

The project will run for about 10 months from Nov. to September 2026, and the corporations and institutions participating in the finally selected consortium will each be provided with 256 of NVIDIA's latest GPUs.

※ This article has been translated by AI. Share your feedback here.