A Re-parameterized Network-Based Lightweight Multimodal Fusion Network for Early Lung Cancer Risk Prediction

Early detection of lung cancer using low-dose computed tomography (LDCT) is critical for improving patient outcomes. In this study, we propose a RepViT-based lightweight multimodal fusion network (Rep-LMFNet) framework, which integrates LDCT imaging and structured clinical information for risk prediction. The framework incorporates adaptive multi-scale module (ASMM), dynamic channel recalibration module (DCRM), adaptive feature enhancement module (AFEM), and hierarchical multimodal fusion module (HMFM) to improve feature representation and multimodal interaction. Experimental results under stratified five-fold cross-validation demonstrate that the proposed framework achieves an area under the receiver operating characteristic curve (AUC) of 0.877 (95% CI: 0.874-0.880), accuracy of 0.922 ± 0.002, sensitivity of 0.847 ± 0.010, specificity of 0.924 ± 0.002, and F1-score of 0.385 ± 0.010. Comparative experiments reveal that our model exhibits competitive advantages over several representative lightweight and hybrid architectures, while maintaining low computational cost with 11.9 M parameters and 1.9G floating-point operations per second (FLOPs). Multimodal learning consistently outperformed the single-modal configuration, indicating the effectiveness of integrating CT imaging with clinical features for early lung cancer risk assessment. The proposed framework provides an effective and computationally efficient solution for computer-aided early lung cancer screening.