Scholarlink

<p>Medical data clustering is an important step in healthcare analytics, as it allows identifying meaningful patterns, disease relationships, and patient subpopulations, which are vital to the correct diagnosis, prognosis, and patient treatment. Nevertheless, the nature of medical data, which is characterized by high dimensionality, noise, redundancy, and complex nonlinear relationships, poses and major challenge to clustering algorithms such as K-Means and hierarchical algorithms. Such traditional approaches are typically afflicted with problems, including sensitivity to the starting point, local optima, and loss of clustering power with some heterogeneous medical data. To solve these difficulties, this paper introduces a new hybrid metaheuristic clustering method, named GSBCO (Gravitational Search Bacterial Colony Optimization), a combination of the merits of the Bacterial Colony Optimization (BCO) and the Gravitational Search Algorithm (GSA). The suggested GSBCO approach is useful in the balancing of exploration and exploitation of clustering search. The BCO component mimics the foraging behavior of bacteria to search the space of potential solutions in a wide manner and to produce variation between candidate solutions, whereas the GSA component can be used to model gravitational interaction between agents in order to deepen the search in promising regions, increasing exploitation and accuracy of convergence. Such synergy between BCO and GSA contributes to the prevention of premature convergence, enhanced cluster compactness and separation, and worldwide optimal solutions. The GSBCO model was tested on a variety of benchmark UCI medical datasets to test its performance. The findings show that the proposed approach has always shown higher performance than the conventional and the existing metaheuristic-driven clustering techniques (K-Means, Hierarchical Clustering, Particle Swarm Optimization (PSO), Bacterial Foraging Optimization (BFO), BCO, and Opposition-Based BCO (OBL-BCO). The results of performance evaluation, such as clustering accuracy, silhouette coefficient, Fowlkes-Mallows score, Beta index, and Rand index, establish the effectiveness and consistency of the GSBCO method. The suggested GSBCO model has a high potential of future development in medical data analysis, disease trends prediction, and predictive analytics of healthcare. GSBCO facilitates better clinical decision-making, patient stratification, and personalized healthcare management by allowing better and more accurate, and reliable clustering results.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

GSBCO: an efficient hybrid medical data clustering approach based on bacterial colony optimization and gravitational search algorithm

  • 期刊论文

  • 发表日期：

    2026年6月2日

查看全文
  • V. S. Prakash,
  • V. Vinothina,
  • Jasmine Gnanadurai

摘要

Medical data clustering is an important step in healthcare analytics, as it allows identifying meaningful patterns, disease relationships, and patient subpopulations, which are vital to the correct diagnosis, prognosis, and patient treatment. Nevertheless, the nature of medical data, which is characterized by high dimensionality, noise, redundancy, and complex nonlinear relationships, poses and major challenge to clustering algorithms such as K-Means and hierarchical algorithms. Such traditional approaches are typically afflicted with problems, including sensitivity to the starting point, local optima, and loss of clustering power with some heterogeneous medical data. To solve these difficulties, this paper introduces a new hybrid metaheuristic clustering method, named GSBCO (Gravitational Search Bacterial Colony Optimization), a combination of the merits of the Bacterial Colony Optimization (BCO) and the Gravitational Search Algorithm (GSA). The suggested GSBCO approach is useful in the balancing of exploration and exploitation of clustering search. The BCO component mimics the foraging behavior of bacteria to search the space of potential solutions in a wide manner and to produce variation between candidate solutions, whereas the GSA component can be used to model gravitational interaction between agents in order to deepen the search in promising regions, increasing exploitation and accuracy of convergence. Such synergy between BCO and GSA contributes to the prevention of premature convergence, enhanced cluster compactness and separation, and worldwide optimal solutions. The GSBCO model was tested on a variety of benchmark UCI medical datasets to test its performance. The findings show that the proposed approach has always shown higher performance than the conventional and the existing metaheuristic-driven clustering techniques (K-Means, Hierarchical Clustering, Particle Swarm Optimization (PSO), Bacterial Foraging Optimization (BFO), BCO, and Opposition-Based BCO (OBL-BCO). The results of performance evaluation, such as clustering accuracy, silhouette coefficient, Fowlkes-Mallows score, Beta index, and Rand index, establish the effectiveness and consistency of the GSBCO method. The suggested GSBCO model has a high potential of future development in medical data analysis, disease trends prediction, and predictive analytics of healthcare. GSBCO facilitates better clinical decision-making, patient stratification, and personalized healthcare management by allowing better and more accurate, and reliable clustering results.