Fuzzy Clustering: A New Clustering Method in Heterogeneous Medical Records Searching

Clustering of heterogeneous medical records plays an extremely important role in understanding pathology, identifying correlations between medical records, and adjuvant treatment of medical records. In view of the instability of the existing medical record clustering algorithm in the processing of heterogeneous medical record data, this paper proposes a medical record clustering algorithm based on fuzzy matrix for integrated structure and unstructured data. Firstly, the algorithm de-correlates the initial data based on the Spearman correlation coefficient to avoid the data correlation error of subsequent analysis. Second, this paper introduces the posterior probability theory for stability weighting, comprehensive structure and unstructured data. Finally, according to fuzzy transitive closure principle, the medical records are clustered from the perspective of relationship transformation. Compared with the existing partial clustering algorithm, the algorithm proposed in this paper improves the clustering accuracy. In addition, it also solves the dynamic and hierarchical problems of medical record clustering to some extent.