Identificação de pacientes com potencial para desenvolver o pé diabético baseada em técnicas de reconhecimento de padrões e ações de auto cuidado

Abstract

The diabetic foot is one of the most serious complications of diabetes mellitus. Generally, from a foot ulcer, amputation of a lower extremity or all may occur. In order to avoid diabetic foot, this work proposes an automatic non invasive system based on computational intelligence algorithms to identify patients with diabetes who have a high risk of developing diabetic foot. To design the method, information was collected regarding the social, habits and self-care of diabetic patients. We used the K-means algorithm to divide the data into two and three groups. After this, classifiers based on centroids of the groups, neural networks, decision tree and k-neighbor nearest were applied to classify the diabetic patient as being of high or low risk of developing diabetic foot. Both real and simulated data were used to evaluate the method. Accuracy of 100% for simulated data and 68% for real data, considering the classification of experts as the gold standard were achieved. The method requires a simple computational processing and can be useful for Basic Health Screening for diabetic patients to help the healthcare team to reduce the number of cases of diabetic foot.