Artificial Neural Networks and Their Application in Diagnosis of Acute Coronary Complications

1120, either, cat:57

ARTIFICIAL NEURAL NETWORKS AND THEIR APPLICATION IN DIAGNOSIS OF ACUTE CORONARY COMPLICATIONS

J L Patel1, L D Patel2, R K Goyal3, C R Chaudhry4, K S Patel1, B D Antala1,

A M Suthar1

1S. K. Patel College of Pharmaceutical Education and Research, Ganpat Vidyanagar, Mehsana, Gujarat, India, 2C. U. Shah College of Pharmacy and Research, A'bad Surendranagar Highway, Wadhwan City, Surendranagar, Gujarat, India, 3L. M. College of Pharmacy, Gujarat, India, 4U. V. Patel College of Engineering, Ganpat Vidyanagar, Mehsana, Gujarat, India

Artificial Neural Networks (ANNs) are the mathematical algorithms generated by computer that approach the functionality of small neural clusters in a very fundamental manner. The artificial analogue of the biological neuron is referred as a Processing Element (PE). PEs are organized into groups referred as layers. Generally there are three types of layers. The input layer collects information presented, the output layer generates a response to a given input and the layers between input and output layers called hidden layers. PEs in any one layer are joined with all PEs in the layer above. The neural network must first be trained by a sufficient number of input data with output resulted from each input data. Once trained, the neural networks able to recognize similarities when presented with a new input pattern. Trained ANNs can be used in medicine in four basic fields: Modeling, Bioelectrical signal processing, Diagnosing and Prognostics. We have tested the ANNs for prediction of acute coronary complications like myocardial infarction and acute coronary syndromes using the diagnostic and general patient data as inputs. Our study suggests that ANNs can predict such acute coronary complications in suspected patients with high accuracy and the prediction accuracy can be improved by optimizing number of hidden layers and the training data sets. Looking to the potential applications of ANNs, it is emphasized that completely new diagnostic equipments can be designed based on the neural network technology that might be helpful to the physicians for emergency diagnosis.