Predicting Plateau Pressure in Intensive Medicine for Ventilated Patients - New Contributions in Information Systems and Technologies - page 166

Predicting Plateau Pressure in Intensive Medicine
for Ventilated Patients

Sérgio Oliveira 1 , Filipe Portela 1 , Manuel Filipe Santos 1 , José Machado 1 ,
António Abelha 1 , Álvaro Silva 2 , and Fernando Rua 2

1 Algoritmi Centre, University of Minho, Portugal
2 Intensive Care Unit, Centro Hospitalar do Porto, Porto, Portugal
[e-mail], {cfp,mfs}@dsi.uminho.pt,
{jmac,abelha}@di.uminho.pt,
[e-mail], [e-mail]

Abstract. Barotrauma is identified as one of the leading diseases in Ventilated
Patients. This type of problem is most common in the Intensive Care Units. In
order to prevent this problem the use of Data Mining (DM) can be useful for
predicting their occurrence. The main goal is to predict the occurence of
Barotrauma in order to support the health professionals taking necessary
precautions. In a first step intensivists identified the Plateau Pressure values as a
possible cause of Barotrauma. Through this study DM models (classification)
where induced for predicting the Plateau Pressure class (>=30 cm O ) in a real
environment and using real data. The present study explored and assessed the
possibility of predicting the Plateau pressure class with high accuracies. The
dataset used only contained data provided by the ventilators. The best models
are able to predict the Plateau Pressure with an accuracy ranging from 95.52%
to 98.71%.

Keywords: Barotrauma, Plateau Pressure, Intensive Medicine, Data Mining,
INTCare, Mechanical Ventilation.

1 Introduction

With the advancement of technology, organizations have a set of data acquisition
mechanisms able to collect information from various business processes. In this point
organizations are recognized as a source of implicit knowledge. The process of
extracting information from knowledge bases can provide better working practices or
even redesign work processes [1]. Health institutions, in particular hospitals, possess
multidisciplinary databases that can be an important knowledge source for predicting
or diagnosing potential complications in patients. The development and application of
Data Mining (DM) models has been experimented and proved to be very useful [2].
Using this new knowledge, physicians are able to identify more effective treatments
and best practices. Consequently, patients can receive healthcare according to their
condition. DM provides not only the methodology, but also the technology to
transform the collected data into useful knowledge for the decision [3].