Retrospective Classification of ARDS in ICU Time-series data using Random Forest with a focus on Data Pre-processing

IFAC-PapersOnLine, vol. 58, pp. 129–134

Abstract

Acute Respiratory Distress Syndrome (ARDS) is a severe lung injury associated with high mortality. Epidemiological studies have shown that ARDS is often diagnosed too late or not at all. Artificial intelligence (AI) can help clinicians identify ARDS and initiate appropriate therapy earlier. Various data must be collected and processed for the training of such AI methods. It is particularly important to consider the data basis and describe the pre-processing steps of the data, as this has a major influence on the results of an AI model. A random forest algorithm is proposed to automatically assess a patient's condition for compatibility with an ARDS using time-series data (like vital signs, laboratory values and other parameters). We emphasize the data preparation and its influence on the results. The model achieved moderate to excellent results depending on the preparation and dataset.

Authors 5

1. Simon Fonck corresponding Aachen Informatik 11 – Embedded Software

   Forschungszentrum Jülich · RWTH Aachen University

   Affiliation as printed

   Center for Advanced Simulation and Analytics (CASA), Forschungszentrum Jülich, 52428 Jülich, Germany

   Informatik 11 - Embedded Software, RWTH Aachen University, 52074 Aachen, Germany

2. Sebastian Johannes Fritsch Aachen Department of Intensive Care Medicine

   Forschungszentrum Jülich · Universitätsklinikum Aachen · Jülich Supercomputing Centre · RWTH Aachen University

   Affiliation as printed

   Center for Advanced Simulation and Analytics (CASA), Forschungszentrum Jülich, 52428 Jülich, Germany

   Department of Intensive Care Medicine, University Hospital RWTH Aachen, 52074 Aachen, Germany

   Jülich Supercomputing Centre, Forschungszentrum Jülich, 52428 Jülich, Germany

3. Alexander Baron Aachen Informatik 11 – Embedded Software

   RWTH Aachen University

   Affiliation as printed

   Informatik 11 - Embedded Software, RWTH Aachen University, 52074 Aachen, Germany

4. Stefan Kowalewski Aachen Informatik 11 – Embedded Software

   RWTH Aachen University

   Affiliation as printed

   Informatik 11 - Embedded Software, RWTH Aachen University, 52074 Aachen, Germany

5. André Stollenwerk Aachen Informatik 11 – Embedded Software

   Forschungszentrum Jülich · RWTH Aachen University

   Affiliation as printed

   Center for Advanced Simulation and Analytics (CASA), Forschungszentrum Jülich, 52428 Jülich, Germany

   Informatik 11 - Embedded Software, RWTH Aachen University, 52074 Aachen, Germany