Data mining in healthcare information systems: Case studies in Northern Lebanon

The growing demands on service quality and cost containment challenge the healthcare organizations around the world. They are particularly crucial with respect to common illness or medical conditions. One strategy that can be used to address these issues is the utilization of healthcare information systems (HIS) for decision support and knowledge management. Indeed, healthcare facilities have at their disposal vast amounts of data which are often organized into useful models. The most critical defy is to generate relevant information from this data. The extracted knowledge should support system to act in an appropriate manner. In this paper, we offer four case studies from the Lebanese healthcare domain (acute appendicitis, premature birth, Osteoporosis, and coronary heart disease) in which prediction systems were developed for decision support using data mining. Our contribution is the application, of our novel reduction technique called Dynamic Rough Sets Attribute Reduction (DRSAR) with multi classifier Random Forest (RF), in HIS. The efficiency of our approach has been carefully tackled within these case studies. Definitely, our algorithms are extendible to deal with mobile/online solutions to support patients as well for clinical diagnosis. As a first step, we also developed web-based interfaces to support patients in calculating risk level for each medical case. These interfaces could be certainly customized in the future for smartphones deployment.