An ensemble and modular neural network approach to the diagnosis of acute appendicitis
Crawford, William Jeffrey
Acute Appendicitis is a disease of the appendix by which the appendix becomes inflamed and may become perforated. By looking for particular signs and symptoms and performing diagnostic tests, experienced clinicians diagnose cases of acute appendicitis with an accuracy rate between 75-80%. Artificial neural networks perform quite well with complex tasks such as pattern networks have been applied to many areas of the medical field for analysis of various diseases and conditions. Application of artificial neural networks to the diagnosis of acute appendicitis is a fairly new area, and not much analysis has been performed with some of the neural models. This thesis is concerned with applying some neural models such as ensembles of networks and modular neural networks in the hopes of obtaining similar results to those of trained physicians, and to gain insights into applying multi-network systems towards other medical related problems.