Diagnosis of acute appendicitis in children using Artificial neural network

Acute appendicitis is one of the most common causes of emergency surgery especially in children. Proper and on-time diagnosis may decrease the unwanted complications. In despite of diagnostic methods, a significant number of patients yet and up with negative laparotomies. The aim of this study was to assess the role of artificial neural networks in diagnosis of acute appendicitis in children with acute abdomen.
Data from 206 patients presenting with acute abdomen referred to ALI ASGHAR pediatric Hospital in Tehran during April 2005 to March 2015 were used in this research. Two train functions, Levenberg-Marquardt and Scaled Conjugate Gradient were used for the feed-forward back propagation neural network.
Results showed that the feed-forward back propagation algorithm with topology of 12-10-2, Levenberg-Marquardt training algorithm and similar functions for all of the layer (Hyperbolic tangent sigmoid) was the best order to diagnosis acute appendicitis in children. The sensitivity, specificity, and accuracy of the artificial neural network were 100 %, 100 %, and 100 % respectively. These results indicated a high potential of neural network as strong tool in diagnosis acute appendicitis in children.
Conclusion and we have used a neural network methods targeted at aiding medical specialist in their diagnosis of acute appendicitis disease. Artificial neural networks could be an effective tool for accurately diagnosing acute appendicitis. Such systems may reduce unnecessary appendectomies, diagnostic costs and time.