Citation: Šín, P.; Hokynková, A.;
Marie, N.; Andrea, P.; Krˇc, R.;
Podroužek, J. Machine
Learning-Based Pressure Ulcer
Prediction in Modular Critical Care
Data. Diagnostics 2022, 12, 850.
https://doi.org/10.3390/
diagnostics12040850
Academic Editors: Keun Ho Ryu and
Nipon Theera-Umpon
Received: 3 March 2022
Accepted: 28 March 2022
Published: 30 March 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Article
Machine Learning-Based Pressure Ulcer Prediction in Modular
Critical Care Data
Petr Šín
1
, Alica Hokynková
1,
* , Nováková Marie
2
, Pokorná Andrea
3
, Rostislav Krˇc
4
and Jan Podroužek
4
1
Department of Burns and Plastic Surgery, Faculty Hospital Brno and Faculty of Medicine, Masaryk University,
Jihlavská 20, 625 00 Brno, Czech Republic; p.sin@seznam.cz
2
Department of Physiology, Faculty of Medicine, Masaryk University, Kamenice 5,
625 00 Brno, Czech Republic; majka@med.muni.cz
3
Department of Health Sciences, Faculty of Medicine, Masaryk University, Kamenice 5,
625 00 Brno, Czech Republic; apokorna@med.muni.cz
4
Institute of Computer Aided Engineering and Computer Science, Faculty of Civil Engineering,
Brno University of Technology, Veveˇrí 331/95, 602 00 Brno, Czech Republic; rostislav.krc@gmail.com (R.K.);
podrouzek.j@fce.vutbr.cz (J.P.)
* Correspondence: alicah@post.cz; Tel.: +420-532-233-004
Abstract:
Increasingly available open medical and health datasets encourage data-driven research
with a promise of improving patient care through knowledge discovery and algorithm development.
Among efficient approaches to such high-dimensional problems are a number of machine learning
methods, which are applied in this paper to pressure ulcer prediction in modular critical care data.
An inherent property of many health-related datasets is a high number of irregularly sampled time-
variant and scarcely populated features, often exceeding the number of observations. Although
machine learning methods are known to work well under such circumstances, many choices regarding
model and data processing exist. In particular, this paper address both theoretical and practical
aspects related to the application of six classification models to pressure ulcers, while utilizing one of
the largest available Medical Information Mart for Intensive Care (MIMIC-IV) databases. Random
forest, with an accuracy of 96%, is the best-performing approach among the considered machine
learning algorithms.
Keywords:
pressure ulcer; pressure injury; machine learning; MIMIC database; MIMIC-IV; open
data; artificial neural network; random forest
1. Introduction
Pressure ulcers (PUs), also called pressure injuries (PIs), are classified into the category
of non-healing or complicated healing wounds in most cases [
1
,
2
]. PUs burden not only
the patients (necessity of wound care, pain, limited social interactions and a consequently
worsening psychological status, etc.) but also represent a significant financial load on the
health care services/systems (hospital, home care, caregivers, etc.). Non-healing wounds
often reflect comorbidity or multimorbidity and represent the so-called silent epidemic
affecting a large proportion of the world’s population [3].
The incidence of pressure injuries worldwide and the prevalence of pressure injuries
in healthcare settings ranges from 0% to 72.5% [
4
–
7
]. It is estimated that around 10% of
hospital patients and 5% of community care patients suffer from PUs and that 72% of all
PUs occur in persons older than 65 years [
8
,
9
]. Differences in prevalence and incidence
statistics are influenced by data collection and analysis methodology [
10
,
11
]. In the Czech
Republic, there are two main sources for PUs monitoring. In the national adverse event
reporting, the PUs are reported from all inpatient healthcare providers nationwide. The
Adverse Event Reporting System (AERS) in the Czech Republic monitors the adverse
events’ (AEs) occurrence in clinical practice and the subsequent data transmission to
Diagnostics 2022, 12, 850. https://doi.org/10.3390/diagnostics12040850 https://www.mdpi.com/journal/diagnostics
