Abstract

LóPEZ FERNANDEZ, Raúl et al. Statistical Model for Prediction of Diabetic Foot Disease in Type 2 Diabetic Patients. Medisur [online]. 2016, vol.14, n.1, pp. 42-52. ISSN 1727-897X.

Background: the need to predict and study diabetic foot problems is a critical issue and represents a major medical challenge. The reduction of its incidence can lead to positive results for improving the quality of life of patients and the impact on the socio-economic sphere, due to the high prevalence of diabetes in the working population. Objective: to design a statistical model for prediction of diabetic foot disease in type 2 diabetic patients. Methods: a descriptive study was conducted in patients attending the Diabetes Clinic in Cienfuegos from 2010 to 2013. Significant risk factors for diabetic foot disease were analyzed as variables. To design the model, binary logistic regression analysis and Chi-squared automatic interaction detection decision tree were used. Results: two models that behaved similarly based on the comparison criteria considered (percentage of correct classification, sensitivity and specificity) were developed. Validation was established through the receiver operating characteristic curve. The model using Chi-squared automatic interaction detection showed the best predictive results. Conclusions: Chi-squared automatic interaction detection decision trees have an adequate predictive capacity, which can be used in the Diabetes Clinic of Cienfuegos municipality.

Keywords : design; models, statistical; diabetic foot; forecasting; diabetes mellitus, type 2; quality of life.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )