Pervasive and intelligent decision support in critical health care using ensembles

Abstract

Critical health care is one of the most difficult areas to make decisions. Every day new situations appear and the doctors need to decide very quickly. Moreover, it is difficult to have an exact perception of the patient situation and a precise prediction on the future condition. The introduction of Intelligent Decision Support Systems (IDSS) in this area can help the doctors in the decision making process, giving them an important support based in new knowledge. Previous work has demonstrated that is possible to use data mining models to predict future situations of patients. Even so, two other problems arise: i) how fast; and ii) how accurate? To answer these questions, an ensemble strategy was experimented in the context of INTCare system, a pervasive IDSS to automatically predict the organ failure and the outcome of the patients throughout next 24 hours. This paper presents the results obtained combining real-time data processing with ensemble approach in the intensive care unit of the Centro Hospitalar do Porto, Porto, Portugal.