Standard

Predicting responders to prone positioning in mechanically ventilated patients with COVID-19 using machine learning. / Dutch ICU Data Sharing Against COVID-19 Collaborators.

In: Annals of intensive care, Vol. 12, No. 1, 99, 12.2022, p. 99.

Research output: Contribution to journalArticleAcademicpeer-review

Harvard

Dutch ICU Data Sharing Against COVID-19 Collaborators 2022, 'Predicting responders to prone positioning in mechanically ventilated patients with COVID-19 using machine learning', Annals of intensive care, vol. 12, no. 1, 99, pp. 99. https://doi.org/10.1186/s13613-022-01070-0

APA

Dutch ICU Data Sharing Against COVID-19 Collaborators (2022). Predicting responders to prone positioning in mechanically ventilated patients with COVID-19 using machine learning. Annals of intensive care, 12(1), 99. [99]. https://doi.org/10.1186/s13613-022-01070-0

Vancouver

Dutch ICU Data Sharing Against COVID-19 Collaborators. Predicting responders to prone positioning in mechanically ventilated patients with COVID-19 using machine learning. Annals of intensive care. 2022 Dec;12(1):99. 99. doi: 10.1186/s13613-022-01070-0

Author

Dutch ICU Data Sharing Against COVID-19 Collaborators. / Predicting responders to prone positioning in mechanically ventilated patients with COVID-19 using machine learning. In: Annals of intensive care. 2022 ; Vol. 12, No. 1. pp. 99.

BibTeX

@article{ea10fdf5574847f19fccc21cfd49dbe1,
title = "Predicting responders to prone positioning in mechanically ventilated patients with COVID-19 using machine learning",
abstract = "BACKGROUND: For mechanically ventilated critically ill COVID-19 patients, prone positioning has quickly become an important treatment strategy, however, prone positioning is labor intensive and comes with potential adverse effects. Therefore, identifying which critically ill intubated COVID-19 patients will benefit may help allocate labor resources.METHODS: From the multi-center Dutch Data Warehouse of COVID-19 ICU patients from 25 hospitals, we selected all 3619 episodes of prone positioning in 1142 invasively mechanically ventilated patients. We excluded episodes longer than 24 h. Berlin ARDS criteria were not formally documented. We used supervised machine learning algorithms Logistic Regression, Random Forest, Naive Bayes, K-Nearest Neighbors, Support Vector Machine and Extreme Gradient Boosting on readily available and clinically relevant features to predict success of prone positioning after 4 h (window of 1 to 7 h) based on various possible outcomes. These outcomes were defined as improvements of at least 10% in PaO 2/FiO 2 ratio, ventilatory ratio, respiratory system compliance, or mechanical power. Separate models were created for each of these outcomes. Re-supination within 4 h after pronation was labeled as failure. We also developed models using a 20 mmHg improvement cut-off for PaO 2/FiO 2 ratio and using a combined outcome parameter. For all models, we evaluated feature importance expressed as contribution to predictive performance based on their relative ranking. RESULTS: The median duration of prone episodes was 17 h (11-20, median and IQR, N = 2632). Despite extensive modeling using a plethora of machine learning techniques and a large number of potentially clinically relevant features, discrimination between responders and non-responders remained poor with an area under the receiver operator characteristic curve of 0.62 for PaO 2/FiO 2 ratio using Logistic Regression, Random Forest and XGBoost. Feature importance was inconsistent between models for different outcomes. Notably, not even being a previous responder to prone positioning, or PEEP-levels before prone positioning, provided any meaningful contribution to predicting a successful next proning episode. CONCLUSIONS: In mechanically ventilated COVID-19 patients, predicting the success of prone positioning using clinically relevant and readily available parameters from electronic health records is currently not feasible. Given the current evidence base, a liberal approach to proning in all patients with severe COVID-19 ARDS is therefore justified and in particular regardless of previous results of proning.",
keywords = "Acute respiratory distress syndrome, COVID-19, Mechanical ventilation",
author = "{Dutch ICU Data Sharing Against COVID-19 Collaborators} and Dam, {Tariq A} and Roggeveen, {Luca F} and {van Diggelen}, Fuda and Fleuren, {Lucas M} and Jagesar, {Ameet R} and Martijn Otten and {de Vries}, {Heder J} and Diederik Gommers and Cremer, {Olaf L} and Bosman, {Rob J} and Sander Rigter and Evert-Jan Wils and Tim Frenzel and Dongelmans, {Dave A} and {de Jong}, Remko and Peters, {Marco A A} and Kamps, {Marlijn J A} and Dharmanand Ramnarain and Ralph Nowitzky and Nooteboom, {Fleur G C A} and {de Ruijter}, Wouter and Urlings-Strop, {Louise C} and Smit, {Ellen G M} and Mehagnoul-Schipper, {D Jannet} and Tom Dormans and {de Jager}, {Cornelis P C} and Hendriks, {Stefaan H A} and Sefanja Achterberg and Evelien Oostdijk and Reidinga, {Auke C} and Barbara Festen-Spanjer and Brunnekreef, {Gert B} and Cornet, {Alexander D} and {van den Tempel}, Walter and Boelens, {Age D} and Peter Koetsier and Judith Lens and Faber, {Harald J} and A Karakus and Robert Entjes and {de Jong}, Paul and Rettig, {Thijs C D} and Sesmu Arbous and Vonk, {Sebastiaan J J} and Tomas Machado and Herter, {Willem E} and {de Grooth}, Harm-Jan and Thoral, {Patrick J} and Girbes, {Armand R J} and Mark Hoogendoorn and Elbers, {Paul W G}",
note = "Funding Information: The authors obliged the learned referee for nice remarks and suggestions. The first author is thankful to the UGC-CSIR, India for a Junior Research Fellowship. The second author is grateful to National Board of Higher Mathematics, Department of Atomic Energy, India for the research grant 02011/11/2020/NBHM(RP)/R &D-II/7830. Publisher Copyright: {\textcopyright} 2022, The Author(s).",
year = "2022",
month = dec,
doi = "10.1186/s13613-022-01070-0",
language = "English",
volume = "12",
pages = "99",
journal = "Annals of intensive care",
issn = "2110-5820",
publisher = "Springer-Verlag GmbH and Co. KG",
number = "1",

}

RIS

TY - JOUR

T1 - Predicting responders to prone positioning in mechanically ventilated patients with COVID-19 using machine learning

AU - Dutch ICU Data Sharing Against COVID-19 Collaborators

AU - Dam, Tariq A

AU - Roggeveen, Luca F

AU - van Diggelen, Fuda

AU - Fleuren, Lucas M

AU - Jagesar, Ameet R

AU - Otten, Martijn

AU - de Vries, Heder J

AU - Gommers, Diederik

AU - Cremer, Olaf L

AU - Bosman, Rob J

AU - Rigter, Sander

AU - Wils, Evert-Jan

AU - Frenzel, Tim

AU - Dongelmans, Dave A

AU - de Jong, Remko

AU - Peters, Marco A A

AU - Kamps, Marlijn J A

AU - Ramnarain, Dharmanand

AU - Nowitzky, Ralph

AU - Nooteboom, Fleur G C A

AU - de Ruijter, Wouter

AU - Urlings-Strop, Louise C

AU - Smit, Ellen G M

AU - Mehagnoul-Schipper, D Jannet

AU - Dormans, Tom

AU - de Jager, Cornelis P C

AU - Hendriks, Stefaan H A

AU - Achterberg, Sefanja

AU - Oostdijk, Evelien

AU - Reidinga, Auke C

AU - Festen-Spanjer, Barbara

AU - Brunnekreef, Gert B

AU - Cornet, Alexander D

AU - van den Tempel, Walter

AU - Boelens, Age D

AU - Koetsier, Peter

AU - Lens, Judith

AU - Faber, Harald J

AU - Karakus, A

AU - Entjes, Robert

AU - de Jong, Paul

AU - Rettig, Thijs C D

AU - Arbous, Sesmu

AU - Vonk, Sebastiaan J J

AU - Machado, Tomas

AU - Herter, Willem E

AU - de Grooth, Harm-Jan

AU - Thoral, Patrick J

AU - Girbes, Armand R J

AU - Hoogendoorn, Mark

AU - Elbers, Paul W G

N1 - Funding Information: The authors obliged the learned referee for nice remarks and suggestions. The first author is thankful to the UGC-CSIR, India for a Junior Research Fellowship. The second author is grateful to National Board of Higher Mathematics, Department of Atomic Energy, India for the research grant 02011/11/2020/NBHM(RP)/R &D-II/7830. Publisher Copyright: © 2022, The Author(s).

PY - 2022/12

Y1 - 2022/12

N2 - BACKGROUND: For mechanically ventilated critically ill COVID-19 patients, prone positioning has quickly become an important treatment strategy, however, prone positioning is labor intensive and comes with potential adverse effects. Therefore, identifying which critically ill intubated COVID-19 patients will benefit may help allocate labor resources.METHODS: From the multi-center Dutch Data Warehouse of COVID-19 ICU patients from 25 hospitals, we selected all 3619 episodes of prone positioning in 1142 invasively mechanically ventilated patients. We excluded episodes longer than 24 h. Berlin ARDS criteria were not formally documented. We used supervised machine learning algorithms Logistic Regression, Random Forest, Naive Bayes, K-Nearest Neighbors, Support Vector Machine and Extreme Gradient Boosting on readily available and clinically relevant features to predict success of prone positioning after 4 h (window of 1 to 7 h) based on various possible outcomes. These outcomes were defined as improvements of at least 10% in PaO 2/FiO 2 ratio, ventilatory ratio, respiratory system compliance, or mechanical power. Separate models were created for each of these outcomes. Re-supination within 4 h after pronation was labeled as failure. We also developed models using a 20 mmHg improvement cut-off for PaO 2/FiO 2 ratio and using a combined outcome parameter. For all models, we evaluated feature importance expressed as contribution to predictive performance based on their relative ranking. RESULTS: The median duration of prone episodes was 17 h (11-20, median and IQR, N = 2632). Despite extensive modeling using a plethora of machine learning techniques and a large number of potentially clinically relevant features, discrimination between responders and non-responders remained poor with an area under the receiver operator characteristic curve of 0.62 for PaO 2/FiO 2 ratio using Logistic Regression, Random Forest and XGBoost. Feature importance was inconsistent between models for different outcomes. Notably, not even being a previous responder to prone positioning, or PEEP-levels before prone positioning, provided any meaningful contribution to predicting a successful next proning episode. CONCLUSIONS: In mechanically ventilated COVID-19 patients, predicting the success of prone positioning using clinically relevant and readily available parameters from electronic health records is currently not feasible. Given the current evidence base, a liberal approach to proning in all patients with severe COVID-19 ARDS is therefore justified and in particular regardless of previous results of proning.

AB - BACKGROUND: For mechanically ventilated critically ill COVID-19 patients, prone positioning has quickly become an important treatment strategy, however, prone positioning is labor intensive and comes with potential adverse effects. Therefore, identifying which critically ill intubated COVID-19 patients will benefit may help allocate labor resources.METHODS: From the multi-center Dutch Data Warehouse of COVID-19 ICU patients from 25 hospitals, we selected all 3619 episodes of prone positioning in 1142 invasively mechanically ventilated patients. We excluded episodes longer than 24 h. Berlin ARDS criteria were not formally documented. We used supervised machine learning algorithms Logistic Regression, Random Forest, Naive Bayes, K-Nearest Neighbors, Support Vector Machine and Extreme Gradient Boosting on readily available and clinically relevant features to predict success of prone positioning after 4 h (window of 1 to 7 h) based on various possible outcomes. These outcomes were defined as improvements of at least 10% in PaO 2/FiO 2 ratio, ventilatory ratio, respiratory system compliance, or mechanical power. Separate models were created for each of these outcomes. Re-supination within 4 h after pronation was labeled as failure. We also developed models using a 20 mmHg improvement cut-off for PaO 2/FiO 2 ratio and using a combined outcome parameter. For all models, we evaluated feature importance expressed as contribution to predictive performance based on their relative ranking. RESULTS: The median duration of prone episodes was 17 h (11-20, median and IQR, N = 2632). Despite extensive modeling using a plethora of machine learning techniques and a large number of potentially clinically relevant features, discrimination between responders and non-responders remained poor with an area under the receiver operator characteristic curve of 0.62 for PaO 2/FiO 2 ratio using Logistic Regression, Random Forest and XGBoost. Feature importance was inconsistent between models for different outcomes. Notably, not even being a previous responder to prone positioning, or PEEP-levels before prone positioning, provided any meaningful contribution to predicting a successful next proning episode. CONCLUSIONS: In mechanically ventilated COVID-19 patients, predicting the success of prone positioning using clinically relevant and readily available parameters from electronic health records is currently not feasible. Given the current evidence base, a liberal approach to proning in all patients with severe COVID-19 ARDS is therefore justified and in particular regardless of previous results of proning.

KW - Acute respiratory distress syndrome

KW - COVID-19

KW - Mechanical ventilation

UR - http://www.scopus.com/inward/record.url?scp=85140261014&partnerID=8YFLogxK

U2 - 10.1186/s13613-022-01070-0

DO - 10.1186/s13613-022-01070-0

M3 - Article

C2 - 36264358

VL - 12

SP - 99

JO - Annals of intensive care

JF - Annals of intensive care

SN - 2110-5820

IS - 1

M1 - 99

ER -

ID: 26511400