0
0This research presents an interpretable machine learning framework designed to predict drug combination synergy in breast cancer treatment. Researchers developed a random forest model that utilizes simulated protein activities, generated through Boolean modeling of signaling pathways, as its primary input features. Unlike "black box" models, this approach allows for local interpretability, meaning the specific contribution of individual proteins can be analyzed to explain why a particular drug pair is effective or ineffective. The study demonstrates the model's utility by identifying molecular mechanisms behind drug resistance and sensitivity, such as how AKT hyperactivity diminishes the synergy of specific inhibitors. Ultimately, the framework bridges the gap between computational predictions and biological understanding, offering a more transparent method for designing complex therapeutic strategies.
References:
Taoma, K., Ruengjitchatchawalya, M., Kusonmano, K. et al. Interpretable prediction of drug synergy for breast cancer by random forest with features from Boolean modeling of signaling pathways. Sci Rep 15, 17735 (2025). doi.org
