Sydney lab uses AI to design cancer drug combos and predictive tests

Researchers at the University of NSW, under the direction of Fatemeh Vafaee, are leveraging artificial intelligence to overhaul traditional drug development, which historically succeeds only about ten percent of the time, costs roughly $2 billion and spans fifteen years. Their AI-driven platform evaluates potential drug combinations, reducing the experimental burden from millions of possibilities to a shortlist for lab testing.

So far, twenty-four promising combos have passed pre-clinical trials at the Peter MacCallum Cancer Centre in Melbourne. In collaboration with a commercial partner, the team also created a blood test that can exclude recurrent breast cancer, a condition that impacts roughly fifteen percent of survivors. Vafaee’s work aims to apply transfer learning to apply insights from common cancers to rarer diseases, arguing that large-scale data access is essential for truly personalized medicine. The university will feature these advances in its upcoming Social Impact of Science report, which proposes new metrics for research commercialization and societal benefit.