Cluster analysis in assessing the interdependencies of clinical and morphological parameters of patients with mixed infections (pulmonary tuberculosis, chronic viral hepatitis, HIV infection) and heroin addiction

Cluster analysis was performed for different indexes in patients with mixed infections (pulmonary tuberculosis, HCV, HIV) and heroin addiction. Morphological peculiarities of liver biopsy specimens were compared with non-invasive clinical and laboratory parameters. Non-hierarchical Partitioning Around Medoids algorithm and fuzzy clustering method were applied. Associations were found for morphological peculiarities and biochemical parameters. Among all indexes under consideration such biochemical parameters as ALT, AST, α₁-globulin and α₂-globulin were the most important ones.