MEDICAL DIAGNOSIS USING NEURAL NETWORK
S. M. Kamruzzaman, Ahmed Ryadh Hasan†, Abu Bakar Siddiquee and Md. Ehsanul Hoque Mazumder Department of Computer Science and Engineering International Islamic University Chittagong, Chittagong-4203, Bangladesh Email: email@example.com, firstname.lastname@example.org, email@example.com † School of Communication Independent University Bangladesh, Chittagong, Bangladesh, Email: firstname.lastname@example.org ABSTRACT This research is to search for alternatives to the resolution of complex medical diagnosis where human knowledge should be apprehended in a general fashion. Successful application examples show that human diagnostic capabilities are significantly worse than the neural diagnostic system. This paper describes a modified feedforward neural network constructive algorithm (MFNNCA), a new algorithm for medical diagnosis. The new constructive algorithm with backpropagation; offer an approach for the incremental construction of near-minimal neural network architectures for pattern classification. The algorithm starts with minimal number of hidden units in the single hidden layer; additional units are added to the hidden layer one at a time to improve the accuracy of the network and to get an optimal size of a neural network. The MFNNCA was tested on several benchmarking classification problems including the cancer, heart disease and diabetes. Experimental results show that the MFNNCA can produce optimal neural network architecture with good generalization ability. proposed . The most well known constructive algorithms are dynamic node creation (DNC) , feedforward neural network construction (FNNC) algorithm  and the cascade correlation (CC) algorithm. The DNC algorithm constructs single hidden layer ANNs with a sufficient number of nodes in the hidden layer, though such networks suffers difficulty in learning some complex problems. In contrast, the CC algorithm constructs multiple hidden layer ANNs with one node in each layer and is suitable for some complex problems. However, the CC algorithm has many practical problems, such as difficult implementation in VLSI and long propagation delay . This paper describes, a new algorithm, the modified feedforward neural network constructive algorithm (MFNNCA) for medical diagnosis. It begins network design in a constructive fashion by adding nodes one after another based on the performance of the network on training data.
2. NETWORK TOPOLOGY
The size of a feedforward network depends on the number of nodes in the input layer, hidden layer and output layer. The number of nodes in the input layer is defined by the input elements in the input vector; the corresponding output vector defines the number of output nodes in the output layer. 2.1 Automatic Determination of Hidden Units with Constructive Approach Constructive algorithms offer an attractive framework for the incremental construction of nearminimal neural-network architectures. These algorithms start with a small network (usually a single neuron) and dynamically grow the network by adding and training neurons as needed until a satisfactory solution is found. The constructive algorithm proposed in the next section starts with one unit in the hidden layer. Additional units are
Neural networks techniques have recently been applied to many medical diagnosis problems    . One of the network structures that have been widely used is the feedforward network, where network connections are allowed only between the nodes in one layer and those in the next layer. Backpropagation algorithm is the most widely used learning algorithm to train multiplayer feedforward network and applied for applications like character recognition, image processing, pattern classification etc. One of the drawbacks of the traditional backpropagation method is the need to...