<p>Accurately detecting distant metastatic lung cancer (DMLC) poses significant challenges in early diagnosis and treatment planning, necessitating advanced computational methods. Traditional approaches often fail in classification accuracy, feature selection, and optimization. This paper presents a hybrid AI-driven architecture integrating ensemble deep learning with an Adaptive Inertia Weight-based Dragonfly Algorithm (AIW-DA) to enhance DMLC diagnosis. AIW-DA optimizes feature selection by reducing dimensionality while preserving essential diagnostic information. The selected features are then processed using an ensemble deep learning classifier, combining convolutional neural networks (CNNs), transformers, and gradient-boosted decision trees (GBDT) for improved accuracy and generalizability. The proposed hybrid method comprehensively assesses metastatic lung cancer spread by integrating histopathological images with multimodal biomedical imaging such as computed tomography (CT) and positron emission tomography (PET). Experimental results on benchmark datasets demonstrate that AIW-DA, coupled with ensemble deep learning, outperforms conventional optimization and classification techniques in accuracy, sensitivity, and specificity. This novel approach enhances diagnostic precision while reducing computational load, providing radiologists and oncologists with a valuable tool for clinical decision-making and personalized treatment planning in early DMLC detection. The experimental test on benchmark datasets shows that the AIW-DA with ensemble deep learning can obtain classification accuracy of 93.2%, feature selection efficiency of 98.8%, the classification performance of 97.4% and the decision support efficiency of 93.7%, reducing the computational complexity by 12.8% compared to conventional methods. The results showed the superiority of the proposed framework in terms of diagnostic accuracy, sensitivity, specificity, and computational efficiency for distant metastatic lung cancer detection.</p>