Abstract:
Diabetes mellitus is a chronic disease which occurs when the pancreas does
not produce sufficient insulin, or when thebody cannot effectively use the
insulin it produces. It is an important and relatively common medical
condition and isa risk factor for many other medical conditions like
stroke,renal failure, blindness, kidney disease and coronary artery disease.
Physicians have to elicit a comprehensive medical history and thorough
physical examination before diabetes mellituscan be suspected. In this study
a lot of data has been collected on the diseases diagnosis
The use of neural networks for diabetic Diagnosis has also attracted the
interest of the medical informatics community because of their ability to
model the nonlinear nature ,the goal of this study is to design a novel
approach for diagnosing diabetes using Artificial Neural Networks. The
study aims also identify the best ANNs type which is more suitable for
diagnoses diabetes .so three ANNs was designed which are feed forwardback propagation ,Recurrent and Elman network to designed which one has
the best performance .
Using MATLAB BPF was designed and trained in the BB NN , RNN, and
elman network ,first one hidden layer between input and output layer with
changed about (3,4,5,6,7,8,9) neuron in hidden layer . type of activation
function which chooses firstly is log sigmoid and tan sigmoid function.
The result obtained that amount the three neural network.
1- theelman network with 5 neuron in the hidden layer with TAN
sigmoid activation function has performance 0.00083289 2- For the
BPNN with 7 neuron in the hidden layer with TAN sigmoid activation
function has performance 0.00011178 that is best performance network in
this study. 3- For the RNN with 8 neuron in the hidden layer with LOG
sigmoid activation function has performance 0.00019097
