Optimization of Community-Consideration Centrality for Breast Cancer Comorbidity Using Various Community Detection Algorithms

Breast cancer is one of the most prevalent diseases worldwide and is often accompanied by various comorbidities. Comorbidity network analysis can be used to identify diseases that play important roles within the network; however, conventional centrality measures do not consider the underlying community structure. This study aims to optimize the Community-Consideration Centrality (CCC) method by comparing several community detection algorithms and evaluating the effect of different \((α)\) parameter values. A comorbidity network was constructed using Disease Ontology Identifier (DOID) data and partitioned using the Leiden, Girvan-Newman, Belief Propagation, and Spinglass algorithms. Community-Consideration Degree, Closeness, and Betweenness Centrality were then calculated. The results show that the Leiden algorithm achieved the highest modularity, indicating the best community structure. The CCC method consistently identified central diseases, including DOID.863, DOID.77, DOID.331, and DOID.150. Furthermore, Jackknife evaluation demonstrated that \((α = 0.9)\) provided the most stable performance. These findings indicate that integrating community detection with CCC is effective for identifying important diseases in breast cancer comorbidity networks.