A graph neural network-based model with out-of-distribution robustness for enhancing antiretroviral therapy outcome prediction for HIV-1

Giulia Di Teodoro, Federico Siciliano, Valerio Guarrasi, Anne-Mieke Vandamme, Valeria Ghisetti, Anders Sönnerborg, Maurizio Zazzi, Fabrizio Silvestri, Laura Palagi

Abstract: Highlights•Introduce a joint fusion model of an FC and GNN for HIV-1 therapy prediction.•Utilize Stanford scores in GNN to enhance model robustness.•Address the clinical need for predicting therapies with novel drugs with limited or no data.•Show our model consistently outperforms baseline models across diverse drugs.•Demonstrate robustness to out-of-distribution drugs, crucial for real world use.