Abstract | ||
---|---|---|
The condition of patients admitted to an Intensive Care Unit is complex to the point that it is often very difficult for physicians to accurately determine the most adequate course of action. However, an ICU is a data rich environment where patients are continuously connected to sensors that allow data collection. Datasets containing such data may hide invaluable information regarding the patients' prognosis. Previous work on intensive care data, produced prediction models that were integrated into a decision support system called INTCare. Although presenting interesting results, INTCare uses static models that are expected to become less accurate over time. As an alternative, this paper presents the results of a set of experiments using an ensemble approach to the prediction of the final outcome of ICU patients, given the data collected during the first 24 hours after ICU admission. Results for both the static and dynamic ensembles (where model weights are updated after each prediction) are presented. |
Year | DOI | Venue |
---|---|---|
2007 | 10.1007/978-3-540-77002-2_35 | EPIA Workshops |
Keywords | Field | DocType |
data collection,decision support system,intensive care data,icu patient,adaptive decision support,intensive care unit,icu admission,static model,adequate course,prediction model,data rich environment,decision support | Intensive care unit,Data collection,Data mining,Course of action,Intelligent decision support system,Computer science,Decision support system,Concept drift,Artificial intelligence,Predictive modelling,Intensive care,Machine learning | Conference |
Volume | ISSN | ISBN |
4874 | 0302-9743 | 3-540-77000-3 |
Citations | PageRank | References |
5 | 0.56 | 12 |
Authors | ||
3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Pedro Gago | 1 | 39 | 4.53 |
Álvaro Silva | 2 | 27 | 3.29 |
Manuel Filipe Santos | 3 | 360 | 68.91 |