Background/Aim: Bladder cancer (BCa) is associated with high recurrence rates, emphasizing the importance of early and accurate detection. This study aimed to develop a lightweight and fast deep learning model, Light-Bladder-Net (LBN), for non-invasive BCa detection using conventional urine data. Materials and Methods: We improved LBN’s generalization by applying data transformations, adding uniform noise, and employing feature selection methods (mRMR, PCA, SVD, t-SNE) to extract key vectors from its fully connected layer. These vectors were integrated into the original dataset, and multiple machine learning models were trained to enhance classification accuracy. Lastly, weighted voting was used to assign importance across these models. Results: Our approach achieved an accuracy of 0.83, a sensitivity of 0.85, a specificity of 0.80, and a precision of 0.81, indicating robust performance in detecting BCa from urine data. Conclusion: This non-invasive diagnostic method offers rapid, cost-effective predictions. A free online tool is available for clinicians and patients to conveniently detect BCa using standard urine samples at http://merlin.nchu.edu.tw/LBN/
Protein interactions form large-scale networks known as protein–protein interaction networks (PPINs) or protein complex networks (PCNs). Extracting meaningful structural frameworks from these molecular relationships through mathematical modeling enables a deeper understanding of biological processes. Although static protein network models have provided valuable insights into the organization of PPINs, they are limited in their ability to capture the dynamic and cooperative nature of protein complexes. This review begins by introducing fundamental concepts in graph and hypergraph theory, with an emphasis on centrality measures. We then discuss the evolution of PPIN analysis from static representations to dynamic graph- and hypergraph-based frameworks. Specifically, we review dynamic PPINs and the challenges associated with their interpolation, dynamic centrality measures, and network models capable of representing multi-node relationships that have been applied to PPINs. Finally, we highlight recent advances in machine learning and deep learning approaches that integrate interaction data with functional annotations, sequence information, and cellular context to predict novel interactions and reconstruct transient protein complexes. Taken together, dynamic PPIN modeling combined with experimental validation provides an integrated framework for understanding coordinated protein functions in cellular processes and across biological systems as well as supporting drug development.