An Artificial Intelligence Based Glucometer for Diabetic Patients using Urinal Analysis

ISSN 2277-5072 | © 2013 Bonfring Abstract--There are nearly about 50.8 million people in the world who have diabetes and maintaining the blood sugar level of diabetic patients is very important. The diabetic patient regularly monitor their glucose level by pricking the finger for taking blood samples 4-5 times a day and controls their sugar level by taking appropriate dosage of medicine. An artificial intelligence based image processing application has been developed, which non-invasively measures the glucose concentration present in the urine sample of a person and hence the equivalent blood glucose level of that person is inferred. Blood sugar level of a person from his/her urine sample has been monitored by noting the colour change of the test sample, when it is reacted with Benedict’s reagent. The colour change of the sample is identified with the help of the camera and displays the result in the form of hue (predominant colour) value. This measurement has become possible by training the neural network using the hue value as the input vector and the glucose value as the test vector. A linear relationship has been obtained successfully with an accuracy of about 96.93%.