Despite the initial success of EGFR-targeted therapies in non-small cell lung cancer (NSCLC), the emergence of drug resistance remains a significant clinical challenge. While several approved anticancer drugs exist, the development of resistance to current EGFR inhibitors necessitates the identification of novel repurposable drugs and rapid strategies to screen drugs that could address resistance. Therefore, this study aimed to develop a machine learning-aided and structure-guided rapid screening framework to identify repurposable inhibitors from an anticancer drug library with the potential activity against EGFR in NSCLC. We developed a Random Forest model (cross-validated R2 = 0.8919 \(\pm\) 0.0128) to predict EGFR inhibitory activity and validated it through molecular docking, molecular dynamics simulations, binding free energy calculations, computational bioactivity profiles, predicted cytotoxicity against NSCLC cell lines, and literature-based validation of inhibitory potential. Among screened compounds, Idarubicin and Larotrectinib emerged as putative candidates with predicted IC50 values of 226.55 nM and 479.06 nM, respectively. Molecular docking revealed higher binding affinities for both Idarubicin (− 9.98 kcal/mol) and Larotrectinib (− 9.42 kcal/mol) compared to the reference drug Erlotinib (-8.91 kcal/mol). Subsequent molecular dynamics simulations revealed highly stable conformations for both compounds (RMSD: Idarubicin 1.49 \(\pm\) 0.24 Å, Larotrectinib 1.34 \(\pm\) 0.29 Å), with consistent binding modes throughout the 100 ns trajectory, where MET793 was a common and most stable hydrogen bond with reference drug Erlotinib. Additionally, they demonstrated favorable predicted cytotoxicity against NSCLC cell lines. In conclusion, our integrated bioinformatics analysis identifies idarubicin and larotrectinib as putative candidates for drug repurposing targeting EGFR in NSCLC, providing a rational foundation for future experimental validation and further preclinical and clinical investigations.
