[AI Pick] SK Biopharm builds AI foundation model for new drug development

SK Biopharmaceuticals [326030] is speeding up AI conversion (AX) by building an artificial intelligence (AI) foundation model specialized in new drug development.

SK Biopharmaceuticals announced on the 6th that it has completed the verification work of the ‘AI Specialized Foundation Model Project’ hosted by the Ministry of Science and ICT.

A consortium involving major companies and medical institutions, including SK Biopharmaceuticals, Lunit [328130], Trillion Labs, Eisen Science, and Kakao Healthcare, developed an AI foundation model specialized in medical science.

Based on this model, these organizations created a respecialization model in three areas: new drug development (compounds and proteins), medical science (omics), and clinical trials. Omics is a discipline that studies the totality of biological information.

SK Biopharmaceuticals said that during the project execution process, it built 7.14 million cases and 4.4 billion tokens worth of new drug development-specific learning data, reflecting pharmaceutical research and development (R&D) expertise and research site needs.

These data used for model learning and performance verification included 66 new drug development tasks, including drug properties/toxicity, drug/target interactions, drug reactivity, and prediction of clinical trial results.

SK Biopharmaceuticals cited verification of predictive performance in various tasks through a specialized model for new drug development as a major achievement of the project.

The company explained that it used the new drug development project as a benchmark to evaluate model performance overall and conducted additional verification with pharmaceutical company's undisclosed drug data to confirm usability and competitiveness.

An official from SK Biopharmaceuticals said, "We also explored the possibility of using AI in the clinical development field, such as pharmaceutical R&D work, patient prognosis analysis, and selection and design of clinical trial subjects," and said that an AI work support system was also established.

SK Biopharmaceuticals plans to begin the AI ​​transition of translational research that connects non-clinical research results to clinical results by combining its experience in developing specialized foundation models with AI computing infrastructure based on NVIDIA's graphics processing unit (GPU) B200.

In addition, we plan to advance research capabilities across diseases, targets, and compounds, and expand the use of AI throughout the entire process of new drug development, from discovery of new drug candidates to clinical trials.

Lee Dong-hoon, CEO of SK Biopharmaceuticals, emphasized, “The competitiveness of AI in new drug development depends on how meaningful answers are provided at the research site,” and added, “Through AI, we will increase the speed and likelihood of success in new drug development.”