Is a virtual system for diabetes patients
The customer’s company produces medical sensors and tracking tools. They created an innovative sensor for people with diabetes helping to monitor the level of insulin and count the quantity of hormone for injection in order to gain the required level. Noveo team was charged with the preparation of a software concept for this sensor and its testing on a virtual patient that also needed to be developed during the first work stage.
As it was the first product of this kind, the project was fulfilled in a R&D mode. The research was started from a neural network study and development. This is a self-learning system that analyses input data and can adjust its behavior and calculations in case of new models and components. The learning process is based on the back prolongation method when inserted data are treated firstly in one way and then if an error occurs the data treatment goes in inverse way. As a result the neural network doublechecks the final calculations to ensure their correctness and remembers this case as a new possible model.
The testing of the system was started with basic input parameters, then their number was increased that allowed monitor the patient data during the day and make calculations of the required insulin injection for the several next hours.