Pancreatic ductal adenocarcinoma (PDAC) is a highly aggressive cancer characterized by late-stage diagnosis and poor prognosis. The challenge of tumor resection in PDAC lies in accurately determining resectability, as factors such as tumor size, location, and involvement of surrounding structures can significantly impact surgical outcomes and patient survival rates. This can be addressed using a multimodal dataset combined with Artificial Intelligence, enabling more accurate predictions of tumor resectability for comprehensive analysis. The presented study presents a hybrid model integrating Explainable Boosting Machines (EBM) and TabNet to predict tumor resectability in patients with PDAC. Utilizing a real-time dataset of 70 patients, the model combines clinical, radiomic, and imaging features to enhance predictive accuracy. The EBM was employed for feature selection, identifying key predictors of resectability, which were subsequently input into the TabNet model. The hybrid approach achieved an impressive classification accuracy of 98.77%, with a precision of 97.54% and a recall of 98.8%. These results demonstrate the model’s robust performance in clinical decision-making, providing valuable insights for the management of PDAC.