The National Cancer Center has begun developing technology that uses artificial intelligence (AI) to predict the actual therapeutic effects of new drug candidates. The aim is to improve the success rate and shorten the development timeline of new drugs by predicting drug responses in animal experiments and patient environments based on cell experiment results.
The National Cancer Center announced Tuesday that a research team led by Dr. Shin Dong-gwan of the Bioinformatics Research Branch has begun developing a "biological world model" that predicts drug responses by linking results from different experimental environments, including cell lines, organoids, and animal models.
In drug development, candidate substances that show outstanding efficacy in cell experiments often fail to deliver the expected results in animal experiments or actual patients. As a result, enormous costs and time are invested, but only a small fraction of candidates ultimately receive approval.
The research team plans to develop technology in which AI learns from cell experiment results and then predicts what responses will occur in organoids and animal models that more closely resemble actual biological environments. The AI effectively serves as a "virtual clinical trial site" for new drug candidates.
The research has been selected as a flagship bio-sector project of the Ministry of Science and ICT's "AI+Science and Technology Innovation Technology Development Project." The total project budget is 3 billion won. The broader program will run for four years from 2026 to 2029 with a total budget of 22.5 billion won, aimed at innovating research methods through AI and strengthening R&D competitiveness in strategic national fields.
One of the biggest challenges in current drug development is the "scale gap" that arises from differences between laboratory environments and the actual human body. The complex characteristics of the human body, including the tumor microenvironment, immune responses, and cell-to-cell interactions, cannot be fully reproduced in laboratories.
The research team plans to define cell lines, organoids, and animal models as independent biological worlds and develop AI technology that can transfer drug response information obtained in one world to another for prediction. This is expected to enable predictions of how cancer cells change when a specific drug is administered, and which genes are activated or suppressed.
The research is particularly meaningful in that it goes beyond simply determining whether a drug is effective and uses generative AI to predict the changes that drugs cause within cells. If successful, the research is expected to identify the likelihood of failure for new drug candidates at an early stage, reducing development costs and accelerating the screening of promising therapeutics.
It can also be applied to precision medicine research that predicts which drugs are effective for which patients, helping to establish customized treatment strategies for cancer patients.
The research is jointly conducted by Dr. Shin Dong-gwan as principal investigator, along with a team led by Professor Eum Su-bin of the School of Artificial Intelligence at Kookmin University and a team led by Professor Kim Yoon-hee of the Department of Cancer Biomedical Science at the National Cancer Center Graduate School of Cancer Science and Policy.
"Predicting whether a drug that worked in the laboratory will also be effective in actual patients is one of the most difficult challenges in drug development," Shin said. "Through this research, we will secure core technology that can increase the likelihood of success in new drug development and improve the efficiency of cancer drug development."
