Prediction of Sudden Cardiac Arrest Due to Diabetes Mellitus Using Fuzzy Based Classification Approach

Abstract

In the past few decades, Diabetes mellitus has been considered as a chronic disease and one of the foremost serious health confronts all over the world. Based on International Diabetic Federation, approximately over 250 million diabetic patients all over the world, and expected to rise to 350 million in 2022. Moreover, 3.8 million deaths are due to diabetic complications. Around, 80% of death is due to diabetic mellitus (Type II), which can be prevented by prior detection of people with this risk. However, now machine learning approaches are utilized for diagnosis of diabetics more accurately. In this investigation, an effectual Fuzzy based classification using variable updation and normalization is anticipated as an intelligent representation of Fuzzy diagnosis (classification) decision. An efficient effort is made to identify cardiac death at initial stage arising from the severity of diabetic mellitus; where feature prediction prior to heart rate variability analysis is performed. The features of diabetic mellitus were examined from diabetic’s database to identify the cause of sudden cardiac arrest. Various derived/performance measures of accuracy with Fuzzy classifier confirm strongly the cause of sudden cardiac arrest in prior stage. For the purpose of clinical applications, incorrect detection of heart rate (BPM) is considered as significant and assessed here. Simulation was carried out in MATLAB environment, real time dataset demonstrates that Fuzzy classifier offer a promising solution for cardiac arrest prediction due to diabetics mellitus. A rule set has been generated using Mamdani membership function with prediction accuracy of 95%, sensitivity of 94% and specificity of 95%. Here, extracted rules are effectual and outcomes are relevant to diabetic and cardiac medical studies.