Padronização e clusterização da informação de saúde: uma proposta para descoberta de conhecimento no auxílio à colaboração médica

Abstract

The use of Information and Communication Technology in health care institutions has increased considerably in recent years. Spurred by technological evolution, and the need to provide organized and complete information of patient on clinical data, this research presents subsidies to organization and standardization of medical information, and presents an analysis of information contained in a medical database through the process of knowledge discovery by data mining process. Initially, based on data from an Electronic Patient Record, there was the standardization of relevant information to health information systems, so it can be presented neatly and completely. This standardization was performed using tools of data representation in computers, such as eXtensible Markup Language (XML) and Document Type Definition (DTD). Subsequently, we analyze data using k-means clustering algorithm, aiming to discover knowledge to help physicians in collaborative work. In the process of data mining, we used the Waikato Environment for Knowledge Analysis (WEKA) tool. This tool has open source and it was developed in Java, being possible to use it for free on any other system. Through cluster analysis, specific knowledge of diagnostics and drugs were generated, as well as some peculiarities of the groups of patients. The relationship between these groups should be association rules, in order to use an electronic medical records system that leverages collaboration among physicians.