Electrocardiography based System for Characterization of Diabetes

Diabetes is a frequent endocrine disorder that is categorized by the rising of sugar level in blood. Nevertheless, there is no permanent treatment for diabetes, but attention keeps on limiting diseases by controlling the insulin level in different ways. The most frequent illness due to diabetes includes retinopathy, neuropathy, nephropathy, and cardiomyopathy. The current technique accessible for recognition of diabetes is an invasive approach. Regular assault can cause bloodstream and discomfort due to blunt on the fingers. Therefore, this research focuses on the identification of diabetes through ECG signals. The major contribution of the research includes its non-invasive method for the diagnosis of diabetes. The data is acquired form 24 participants. Including 10 diabetic and 14 normal subjects. Empirical Mode Decomposition is used to denoise the ECG signal. The proposed methodology provided an accuracy of 96.8% on the self-collected database via the support vector machine classifier.