PS 05-43 A HYBRID INTELLIGENT MODEL BASED ON MODULAR NEURAL NETWORK AND FUZZY LOGIC FOR HYPERTENSION RISK DIAGNOSIS

    loading  Checking for direct PDF access through Ovid

Abstract

Objective:

The main goal is to design a hybrid model using modular neural networks, and as response integrator we use fuzzy systems to provide the risk diagnosis of hypertension, so we can prevent a futures disease in people based on systolic pressure, diastolic pressure and patients pulse of ages between 15 to 95 years

Design and Method:

Records of the blood pressure are obtained by the ABPM for 100 people, and these data have been obtained from students of the master and doctorate in computer science from Tijuana Institute of Technology. The modular neural network is trained with each database, in others words, the first module was trained with the records of systolic pressure, the second with the diastolic pressure and the third module with the pulse, the network is modeling the data for learning the blood pressure behavior.

Results:

The modular neural network was trained with patient data to observe the trends and find optimal results. We can note from the results that the training methods, which were the Levenberg-Marquardt (LM) and Gradient descent with momentum and adaptive learning rate backpropagation (GDX), with different layers and neurons. There is significant improvement in the training of LM method. The accuracy of model is more than 99%.

Conclusions:

This paper has presented a hybrid intelligent system for providing risk diagnosis in patients with hypertension, this type of system can be helpful for reducing the complexity of the problem to be solved. We used a modular neural network with a fuzzy response integrator for providing an accurate result.

Related Topics

    loading  Loading Related Articles