Abstract
Our study aims to address the challenges in drug development for glioblastoma, a highly aggressive brain cancer with poor prognosis. We propose a computational framework that utilizes machine learning-based propensity score matching to estimate counterfactual treatment effects and predict synergistic effects of drug combinations. Through our in-silico analysis, we identified promising drug candidates and drug combinations that warrant further investigation. To validate these computational findings, we conducted in-vitro experiments on two GBM cell lines, U87 and T98G. The experimental results demonstrated that some of the identified drugs and drug combinations indeed exhibit strong suppressive effects on GBM cell growth. Our end-to-end pipeline showcases the feasibility of integrating computational models with biological experiments to expedite drug repurposing and discovery efforts. By bridging the gap between in-silico analysis and in-vitro validation, we demonstrate the potential of this approach to accelerate the development of novel and effective treatments for glioblastoma.