This abstract discusses using CT image-based radiomics model as a non-invasive solution to predict EGFR mutation status in NSCLC. The study collected CT images from multiple centers and open-source databases to investigate the performance of the model. The model achieved promising results with an AUC of 0.83 on cross-validation and an AUC of 0.76 on the test set. The authors conclude that AI-powered medical image analysis has the potential to serve as predictive biomarkers for guiding targeted therapies in the future.