Authors:
Femida Gwadry-Sridhar
1
;
Benoit Lewden
2
;
Selam Mequanint
2
and
Michael Bauer
1
Affiliations:
1
University of Western Ontario, Canada
;
2
Lawson Health Research Institute, Canada
Keyword(s):
Sepsis, Decision support, Decision trees.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Biomedical Engineering
;
Business Analytics
;
Cardiovascular Technologies
;
Computing and Telecommunications in Cardiology
;
Data Engineering
;
Data Mining
;
Databases and Information Systems Integration
;
Datamining
;
Enterprise Information Systems
;
Health Engineering and Technology Applications
;
Health Information Systems
;
Information Systems Analysis and Specification
;
Knowledge Management
;
Medical and Nursing Informatics
;
Ontologies and the Semantic Web
;
Sensor Networks
;
Signal Processing
;
Society, e-Business and e-Government
;
Soft Computing
;
Support for Clinical Decision-Making
;
Web Information Systems and Technologies
Abstract:
Sepsis is a significant cause of mortality and morbidity and is often associated with increased hospital resource utilization, prolonged intensive care unit (ICU) and hospital stay. The economic burden associated with sepsis is severe. With advances in medicine, there are now aggressive goal oriented treatments that can be used to help these patients. If we were able to predict which patients may be at risk for sepsis we could start treatment early and potentially reduce the risk of mortality and morbidity. Analytic methods currently used in clinical research to determine the risk of a patient developing sepsis may be further enhanced by using multi-modal analytic methods that together could be used to provide greater precision. Researchers commonly use univariate and multivariate regressions to develop predictive models. We hypothesized that such models could be enhanced by using multi-modal analytic methods that together could be used to provide greater precision. In this paper, we
analyze data about patients with and without sepsis using a decision tree approach. A comparison with a regression approach shows strong similarity among variables identified, though not an exact match. We compare the variables identified by the different approaches and draw conclusions about the respective predictive capabilities.
(More)