Authors:
Marta Vilas-Boas
1
;
Manuel Filipe Santos
1
;
Filipe Portela
1
;
Álvaro Silva
2
and
Fernando Rua
2
Affiliations:
1
Universidade do Minho, Portugal
;
2
Hospital Geral de Santo António, Portugal
Keyword(s):
INTCare, Intelligent Decision Support Systems, Clinical Data Mining, Real-Time prediction, Hourly prediction, Intensive Care Medicine.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Biomedical Engineering
;
Business Analytics
;
Data Engineering
;
Data Mining
;
Databases and Information Systems Integration
;
Datamining
;
Enterprise Information Systems
;
Health Information Systems
;
Sensor Networks
;
Signal Processing
;
Soft Computing
Abstract:
The use of Data Mining techniques makes possible to extract knowledge from high volumes of data. Currently, there is a trend to use Data Mining models in the perspective of intensive care to support physicians’ decision process. Previous results used offline data for the predicting organ failure and outcome for the next day. This paper presents the INTCare system and the recently generated Data Mining models. Advances in INTCare led to a new goal, prediction of organ failure and outcome for the next hour with data collected in real-time in the Intensive Care Unit of Hospital Geral de Santo António, Porto, Portugal. This experiment used Artificial Neural Networks, Decisions Trees, Logistic Regression and Ensemble Methods and we have achieved very interesting results, having proven that it is possible to use real-time data from the Intensive Care Unit to make highly accurate predictions for the next hour. This is a great advance in terms of intensive care, since predicting organ failur
e and outcome on an hourly basis will allow intensivists to have a faster and pro-active attitude in order to avoid or reverse organ failure.
(More)