Characterization of People with Type II Diabetes Using Electrical Bioimpedance

Abstract

Diabetes is a disease that causes the death of a person every seven seconds around the world and is also expensive. In 2014, it invested 600 billion dollars to be treated worldwide, that is why the need arises develop technological projects that allow the analysis of specific patterns in people suffering from this disease, in order to detect the pathology in a non-invasive manner and reduce costs, for which an electrical bioelectrical impedance analyzer was developed for the analysis of diabetes. The integrated AD5933 was used as a bioimpedance signal acquisition device and a beaglebone black development platform to process said data. Bioelectrical impedance data were taken from 5 healthy people and 3 people with the pathology. The data were processed by mathematical methods such as linearization with least squares and correlation, which allowed us to find parameters to differentiate between the signals of people with diabetes and people without diabetes. It was determined that people with diabetes have a curve that relates their bioimpedance in a range of frequencies from 10 kHz to 80 kHz, a curve that presented a high correlation to a power function of the form aXb. It was observed that the people who presented values of coefficients (a) greater than 38000 and exponents (b) less than −0.659 were people with diabetes, this in turn allowed to find the equation of a line that separates the two populations W0 + 1W * a + W2 * b = 0.