Neural network system for outpatient management of insulin dependent patients

Describes a decision support system, which is able to predict the insulin regime and dose to be taken by insulin-dependent diabetes mellitus (IDDM) patients for a short time period. The system is implemented using two feedforward neural networks trained with a back-propagation algorithm. The output of the first neural network provides information about the insulin regime that has to be followed by the patient, while the output of the second neural network suggests the type and the dose of insulin to be taken.