Random forest analysis: a new approach for classification of Beta Thalassemia

  • Massimiliano Sacco
  • Mariangela Sciandra Dipartimento di Scienze Economiche Statistiche e Aziendali
  • Aurelio Maggio
Keywords: Random forest, Unsupervised classification, Clustering, Thalassemia


In recent years, Thalassemia care providers started classifying patients as transfusion-dependent-Thalassemia (TDT) or non-transfusion-dependent-Thalassemia (NTDT) owing to the established role of transfusion therapy in defining the clinical complication profile, although this classification was also based on expert opinion and is limited by reliance on patients’current transfusion  status. Starting from a vast set of variables indicating severity phenotype, through the use of both classification and clustering techniques we want to explore the presence of
two (TDT vs NTDT) or more clusters, in order to approaching to a new definition for the classification of Beta-Thalassemia in Thalassemia Syndromes (TS).