Artificial neural network approach to diabetic management

Applicability of different types of Artificial Neural Networks (ANNs) in the field of diabetic management is discussed for patient classification and insulin treatment adviser. Different solutions are introduced to eliminate the static characteristic of feedforward ANNs in the latter case where the purpose is to model the highly nonlinear dynamic human Blood Glucose control system. The use of a multilayer feedforward ANN trained with error back-propagation as part of an insulin dosage adviser is described in detail. Due to the sparse, inconsistent and incomplete characteristic of data in out-patient treatment, an interpolation technique is required to produce appropriate input for the network. The evaluation and validation of the system is in progress.

[1]  F. Ham,et al.  Neural Network Based Real-time Detection Of Glucose Using A Non-chemical Optical Sensor Approach , 1990, [1990] Proceedings of the Twelfth Annual International Conference of the IEEE Engineering in Medicine and Biology Society.

[2]  E R Carson,et al.  Computer-assisted diabetic management: a complex approach. , 1990, Computer methods and programs in biomedicine.

[3]  E. Carson,et al.  A consultation system for insulin therapy. , 1990, Computer methods and programs in biomedicine.