Meta-classifier for Type 2 Diabetes Mellitus comorbidities in Colombia

This article presents a general meta classifier model for Type 2 Diabetes Mellitus (T2DM) comorbidities which is based on business intelligence and data mining techniques. The proposed meta classifier has two phases: i) the model predicts whether a patient can develop a comorbidity and ii)the model predicts which kind of comorbidity could be: micro or macro vascular. Experiments were carried out with a cohort of 14162 T2DM patients from 2009 to 2012. 3459 of them were comorbidity patients. Obtained results show an accuracy of 87% in the first phase of the meta-classifier and an accuracy of 68% in the second phase.