A Partitional Clustering Approach for the Identification and Analysis of Coexisting Bacteria in Groups of Bacterial Vaginosis Patients
Abstract
Bacterial vaginosis is a condition where there is a large ecosystem of microorganisms and an unclear pathogenesis, making it a disease complex in the dynamic of coexistence of bacteria in groups of patients. The main objective of this study is to provide a partitioning clustering model that allows further analysis of coexisting bacteria in a grouped way in BV-positive patients. K-Means variants (Lloyd, Forgy, Hartigan & Wong, and MacQueen) with three distance measures were applied to a BV dataset from an urban population in southeastern Mexico, which consists of 201 patient records with 15 attributes. In the clustering results obtained, it is possible to identify different notable groups of patients. The most prevalent coexisting bacteria between patients with BV were Atopobium + Gardnerella vaginalis with 31.37%, Atopobium + Megasphaera with 15.68% in the cluster that assigned all BV-positive patients. Whereas, the model that achieved to group BV-positive elements into different clusters, the coexisting bacteria were Atopobium + Gardnerella vaginalis with 56.25% and Atopobium + Megasphaera with 68.75% for group C1. The second group bacterial coexistence was Atopobium + Gardnerella vaginalis with 37.14%. Finally, we provided evidence that, using the partitioning algorithm, it was possible to create a clustering model that helps analyze the complex dynamics among bacteria in groups of patients with BV.
Keywords
Clustering, bacterial vaginosis, coexisting bacteria