Risco para desenvolver o pé diabético utilizando redes neurais artificiais: uma tecnologia para o cuidado de enfermagem

Abstract

Introduction: Diabetes mellitus is a metabolic disease characterized by hyperglycemia and, in general, it appears in association to other diseases such as arterial hypertension. It is a chronic disease and the number of bearers is considerably increasing. If not adequately treated and accompanied, many complications might arise, among which the diabetic foot is highlighted, being one of the most fearsome complications of the disease, which affects the inferior members. The delay in treating and accompanying the diabetic foot might lead to serious complications and even to amputation. This research is justified since, in the daily work of the nurse, in primary health care, there are several demands this professional must face and, most times, he is unable to accompany, directly, the history of the DM bearer and possible complications he might come to develop. The objective of this study consisted in the “elaboration of a risk classification of a diabetes mellitus bearer developing the diabetic foot as a technology for nursing care”. Objective: Classify the risk for the development of the diabetic foot in diabetes mellitus bearers for nursing care, using the artificial neural network (ANN) and identify the main risk factors for the development of the diabetic foot using the Artificial Neural Network (ANN). Methodological design: Quantitative study, performed in a Diabetic Association of the municipality of Juiz de Fora, MG, Brazil. The studied population were 250 patients, bearers of diabetes mellitus. The data-sampling instrument used was an adapted questionnaire, in accordance to the guidelines proposed by the International Working Group on the Diabetic Foot (2001) and the Ministério da Saúde (2006). The data sampling occurred in the period from September to December of 2013. The data were codified and normalized in a certain numeric scale (-1 to 1). For the data analysis the Matlab software was used, in which an Artificial Neural Network (ANN) was created to classify the risk of developing the diabetic foot, based on the variables of the questionnaire answered by the participants. After the classification, the main variables were analyzed regarding the discrimination in the separation of the two groups. Results and discussion: A statistical analysis of the data was performed in which were found percentages of the responses: 55.6% were women; 49.6% had 61 years of age or over; 52.8% were married or lived with a partner. Regarding education, 34.4% did not study or presented less than three years of schooling; 58.4% had monthly income of 500.00 to 1000.00 reais. Regarding the time of diagnosis of the disease, 34.8% presented from 11 to 20 years of diagnosis; 72.4% had the latest blood glucose value altered; 72% were above weight of obese. Among 95,2% of the participants, the diabetes was of type 2; 59.6% were retired; 78% controlled alimentation and 64% did not practice physical activity; 59% never walked barefoot and 45% use medication on their feet when they perceive alterations; 74.4% were hypertensive and 17.6% have or have already had wound on their feet. The ANN created found two risk groups and for the definition of the groups, the main risk characteristics in developing the diabetic foot were used, finding 40.8% of the participants in the high risk group of developing diabetic foot and 59.2% in the low risk group. The most discriminant variables in the formation of the risk groups, selected by the ANN were: sensitivity loss, tingling sensation, numbness, burning, shock in the legs and/or feet, pain, stabbing pain in the legs and/or feet, and hot and reddish feet. The least discriminant variables were those that represent, respectively, the characteristics: diabetes type, alimentation control, smoking habit, arterial hypertension, peripheral vascular disease, habit of washing the feet, internal aspect of the shoe and presence of amputation.