The University of Edinburgh -
Division of Informatics
Forrest Hill & 80 South Bridge

MSc Thesis #93130

Title:The Use of Neural Networks in the Prediction of Acute Myocardial Infarction and Complications
Date: 1993
Abstract:Acute myocardial infarction, or heart attack, is now responsible for more deaths in Britain than any other medical condition. However a method of accurately diagnosing the condition in the first few hours from the onset of symptoms would allow much better treatment, improving both the patient's prognosis and the efficiency of the coronary care unit. Neural network methods have already been successfully used to diagnose acute myocardial infarction based on symptoms available at presentation. in this project a large database of over 5000 patients was used with two types of network (the multi-layer perceptron and radial basis function network) to confirm the results of this previous work and extend the technique to the prediction of major complications, most notably death. While the results from diagnosing acute myocardial infarction were not as good as the results from earlier studies, the extension to complications gave promising results in predicting death.

[Search These Pages] [DAI Home Page] [Comment]