AI accelerates search for cancer therapeutics in Korea

Researchers from SK Telecom and SK Biopharmaceuticals discuss the results of their artificial intelligence-powered drug discovery research. Courtesy of SK Telecom

By Lee Kyung-min

• Published Jul 15, 2026 1:05 PM KST

Artificial intelligence (AI) has helped SK Telecom and SK Biopharmaceuticals identify promising early compounds for a targeted cancer therapy in just five months, a breakthrough the companies said could significantly accelerate the earliest stage of drug development.

SK Telecom said Wednesday that the joint research successfully identified two early hit compounds capable of binding to ROR1, a protein found on the surface of cancer cells that has emerged as a promising target for treating a range of blood cancers and solid tumors.

The companies said AI generated and screened a large pool of binder candidates before laboratory testing confirmed that two showed potential as early-stage hit compounds.

Binders are molecules designed to attach to specific biological targets, such as proteins found on cancer cells.

Finding new binders typically requires evaluating multiple factors, including how strongly they bind to a target and whether their molecular structures remain stable.

SK Biopharmaceuticals developed the strategy for discovering new binders based on its drug development expertise, while SK Telecom applied its AI technologies to generate candidate molecules and analyze their likelihood of binding to ROR1.

The companies said conventional AI approaches often struggle because early drug discovery lacks sufficient training data.

To overcome that limitation, SK Telecom applied machine learning that combines protein fragments in diverse ways and used reinforcement learning to reward structurally stable molecular designs.

The company also used its graphics processing unit computing resources to process large numbers of candidate binders simultaneously.

Its AI models rapidly predicted how each candidate could interact with ROR1, allowing researchers to narrow the field before laboratory validation.

The companies said the research was completed in about five months, reducing a process that typically takes one to two years by more than 60 percent.

This article was published with the assistance of generative AI and edited by The Korea Times.