Intelligent Sensor for Fault Detection in Glucose Measuring System

Technological advances have allowed us to have increasingly portable and efficient glucose measuring systems in the blood, but it has been shown that problems of loss of information are present in the measuring sensors, these failures usually are associated with activities performed by the patient such as exercise, climbing stairs, among others. These failures where the sensor position is affected may cause wrong measurement and in this case the loss of information may occur, as described above we propose a work in which we use the patient sensor to take continuous time samples of the glucose level in the blood of the patient, Those samples are the inputs for the neural network (this neural network will emulate the intelligent sensor) which takes those values and relate them with the past values of glucose and with the insulin dosing. Once the failure of the sensor is detected the prediction mode of the neural network is activated in order to minimize or reduce the loss of information because the neural network will take the place of the sensor temporally.