<p>Lymph node metastasis (LNM) is a commonly used indicator in clinical diagnoses. However, existing diagnostic methods often struggle to achieve efficient and comprehensive analysis due to computational capacity limitations, especially when dealing with LNM dataset, where some models encounter bottlenecks such as low computational efficiency and poor scalability. Based on this, we selected the real LNM dataset from Fujian Provincial Maternity and Children's Hospital as a case, using a novel extended belief rule base (EBRB) expert system based on clustering search framework for LNM diagnosis. The results show that: (1) The introduction of the clustering search framework is capable of effectively identifying the positive data within the samples and enhancing the precision and efficiency of clustering. (2) The EBRB model can fully utilize high-performance computing resources, maintaining high accuracy while also retaining strong interpretability. (3) Compared with other existing mainstream models, the novel EBRB expert system based on the clustering search framework has excellent predictive performance and computational efficiency, providing support for clinical auxiliary diagnosis. Consequently, this study integrates rule-based reasoning with clustering algorithms to fully leverage the potential of supercomputing resources in handling complex medical data, providing a new and scalable path for clinical decision support in lymph node metastasis. The experimental results show that compared with the original model, the proposed method has improved by 20.36 in Sensitivity, by 17.76 in Specificity, by 19.18 in G-means, by 34.34 in F1-Score, and by 17.68 in AUC. Compared with the existing mainstream models, it has achieved significant improvements, demonstrating the reliability and high performance of this expert system.</p>