Thyroid disease diagnosis using genetic algorithm and fuzzy neural network

Abstract EN

The use of expert systems Techniques were still have a great importance in the diagnosis of diseases, thyroid disease diagnosis is one of the important classification problems because thyroid hormones help regulation of the body’s metabolism.

In this paper, hybrid intelligent model for diagnosis is proposed which is able to assist the physicians to take effective decisions.

The behavior of two models has been described in classification of thyroid disease and diagnosis its state (normal, hyperthyroidism, and hypothyroidism).

Firstly, training of fuzzy neural network have been used trial and error approach, then combining it with Genetic algorithm to obtain the best parameters using the proposed representation for the chromosomes and fitness function.

Training the two models was on the same dataset that are available on UCI dataset.

The experimental results show high classification accuracy for both systems , but the best results have been gotten when combining with genetic algorithm obtaining smaller structure with less number of both hidden layer and neurons for each one , also much less error for training and testing .

That is mean the proposed model can be used to facilitate thyroid disease diagnosis in more accuracy.