Development Of Web-Based Software For Acute Coronary Syndrome And A Medical Data Mining Application

Abstract

Aim: Medical data mining is based on data mining methods and related intelligent methods (e.g., granular computing, neural networks and soft computing) used in medicine. In this research, it was aimed to develop a web-based software and to implement medical data mining on the records of the patients with acute coronary syndrome. Materials and Methods: The data in this study included retrospective observations recorded in the database from the webbased software developed for Cardiology Department, Turgut Özal Medical Center, Inonu University. PHP (Personal Home Page) programming language and MySQL Database Management System were employed for the development of the web-based software system. Laplace Support Vector Machines (LSVM) was constructed to predict absence or presence of diabetes mellitus in patients with acute coronary syndrome. Results: A web based software performing data entry, query, delete, update, etc. was developed. As a result of medical data mining application, the accuracy and area under ROC curve with 95% CI were obtained as; 0.9804 (0.9716 - 0.987) and 0.9332 (0.9096 - 0.9567), respectively. Conclusion: The developed web-based software created a very important infrastructure for implementing medical data mining applications. It was determined that the LSVM model produced very good predictive results to estimate absence or presence of diabetes mellitus in patients with acute coronary syndrome.