Using machine learning models to predict oxygen saturation following ventilator support adjustment in critically ill children: A single center pilot study.
Journal
PloS one
ISSN: 1932-6203
Titre abrégé: PLoS One
Pays: United States
ID NLM: 101285081
Informations de publication
Date de publication:
2019
2019
Historique:
received:
24
05
2018
accepted:
04
02
2019
entrez:
21
2
2019
pubmed:
21
2
2019
medline:
4
12
2019
Statut:
epublish
Résumé
In an intensive care units, experts in mechanical ventilation are not continuously at patient's bedside to adjust ventilation settings and to analyze the impact of these adjustments on gas exchange. The development of clinical decision support systems analyzing patients' data in real time offers an opportunity to fill this gap. The objective of this study was to determine whether a machine learning predictive model could be trained on a set of clinical data and used to predict transcutaneous hemoglobin oxygen saturation 5 min (5min SpO2) after a ventilator setting change. Data of mechanically ventilated children admitted between May 2015 and April 2017 were included and extracted from a high-resolution research database. More than 776,727 data rows were obtained from 610 patients, discretized into 3 class labels (< 84%, 85% to 91% and c92% to 100%). Due to data imbalance, four different data balancing processes were applied. Then, two machine learning models (artificial neural network and Bootstrap aggregation of complex decision trees) were trained and tested on these four different balanced datasets. The best model predicted SpO2 with area under the curves < 0.75. This single center pilot study using machine learning predictive model resulted in an algorithm with poor accuracy. The comparison of machine learning models showed that bagged complex trees was a promising approach. However, there is a need to improve these models before incorporating them into a clinical decision support systems. One potentially solution for improving predictive model, would be to increase the amount of data available to limit over-fitting that is potentially one of the cause for poor classification performances for 2 of the three class labels.
Sections du résumé
BACKGROUND
In an intensive care units, experts in mechanical ventilation are not continuously at patient's bedside to adjust ventilation settings and to analyze the impact of these adjustments on gas exchange. The development of clinical decision support systems analyzing patients' data in real time offers an opportunity to fill this gap.
OBJECTIVE
The objective of this study was to determine whether a machine learning predictive model could be trained on a set of clinical data and used to predict transcutaneous hemoglobin oxygen saturation 5 min (5min SpO2) after a ventilator setting change.
DATA SOURCES
Data of mechanically ventilated children admitted between May 2015 and April 2017 were included and extracted from a high-resolution research database. More than 776,727 data rows were obtained from 610 patients, discretized into 3 class labels (< 84%, 85% to 91% and c92% to 100%).
PERFORMANCE METRICS OF PREDICTIVE MODELS
Due to data imbalance, four different data balancing processes were applied. Then, two machine learning models (artificial neural network and Bootstrap aggregation of complex decision trees) were trained and tested on these four different balanced datasets. The best model predicted SpO2 with area under the curves < 0.75.
CONCLUSION
This single center pilot study using machine learning predictive model resulted in an algorithm with poor accuracy. The comparison of machine learning models showed that bagged complex trees was a promising approach. However, there is a need to improve these models before incorporating them into a clinical decision support systems. One potentially solution for improving predictive model, would be to increase the amount of data available to limit over-fitting that is potentially one of the cause for poor classification performances for 2 of the three class labels.
Identifiants
pubmed: 30785881
doi: 10.1371/journal.pone.0198921
pii: PONE-D-18-15649
pmc: PMC6382156
doi:
Substances chimiques
Oxygen
S88TT14065
Types de publication
Journal Article
Research Support, Non-U.S. Gov't
Langues
eng
Sous-ensembles de citation
IM
Pagination
e0198921Déclaration de conflit d'intérêts
The authors have declared that no competing interests exist.
Références
Crit Care. 2015 Feb 24;19:48
pubmed: 25887887
Crit Care. 2012 May 16;16(3):R85
pubmed: 22591622
Pediatr Crit Care Med. 2015 Jun;16(5):428-39
pubmed: 25647235
Crit Care Med. 2018 May;46(5):e375-e379
pubmed: 29406422
Am J Respir Crit Care Med. 2016 Jan 1;193(1):4-5
pubmed: 26720783
Ann Intensive Care. 2011 Dec 21;1(1):51
pubmed: 22189095
Pediatrics. 2011 Oct;128(4):740-52
pubmed: 21930554
Proc IEEE Inst Electr Electron Eng. 2016 Feb;104(2):444-466
pubmed: 27765959
Pediatr Crit Care Med. 2012 Mar;13(2):152-7
pubmed: 21760567
J Clin Monit Comput. 2012 Jun;26(3):145-51
pubmed: 22407178
Heart Lung. 2010 Mar-Apr;39(2):147-52
pubmed: 20207275
Sci Data. 2016 May 24;3:160035
pubmed: 27219127
Crit Care Res Pract. 2013;2013:943281
pubmed: 23533735
Crit Care Med. 2005 May;33(5):995-1000
pubmed: 15891327
Intensive Care Med. 2001 Apr;27(4):655-9
pubmed: 11403067
JAMA. 2016 Oct 18;316(15):1583-1589
pubmed: 27706466
Pediatr Crit Care Med. 2018 Apr;19(4):e189-e198
pubmed: 29406373
Respir Care. 2003 Apr;48(4):386-96; discussion 397-8
pubmed: 12667266