Explainable machine learning approach to predict extubation in critically ill ventilated patients: a retrospective study in central Taiwan
Abstract Background Weaning from mechanical ventilation (MV) is an essential issue in critically ill patients, and we used an explainable machine learning (ML) approach to establish an extubation prediction model. Methods We enrolled patients who were admitted to intensive care units during 2015–201...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
BMC
2022-11-01
|
Series: | BMC Anesthesiology |
Subjects: | |
Online Access: | https://doi.org/10.1186/s12871-022-01888-y |
_version_ | 1811217071032238080 |
---|---|
author | Kai-Chih Pai Shao-An Su Ming-Cheng Chan Chieh-Liang Wu Wen-Cheng Chao |
author_facet | Kai-Chih Pai Shao-An Su Ming-Cheng Chan Chieh-Liang Wu Wen-Cheng Chao |
author_sort | Kai-Chih Pai |
collection | DOAJ |
description | Abstract Background Weaning from mechanical ventilation (MV) is an essential issue in critically ill patients, and we used an explainable machine learning (ML) approach to establish an extubation prediction model. Methods We enrolled patients who were admitted to intensive care units during 2015–2019 at Taichung Veterans General Hospital, a referral hospital in central Taiwan. We used five ML models, including extreme gradient boosting (XGBoost), categorical boosting (CatBoost), light gradient boosting machine (LightGBM), random forest (RF) and logistic regression (LR), to establish the extubation prediction model, and the feature window as well as prediction window was 48 h and 24 h, respectively. We further employed feature importance, Shapley additive explanations (SHAP) plot, partial dependence plot (PDP) and local interpretable model-agnostic explanations (LIME) for interpretation of the model at the domain, feature, and individual levels. Results We enrolled 5,940 patients and found the accuracy was comparable among XGBoost, LightGBM, CatBoost and RF, with the area under the receiver operating characteristic curve using XGBoost to predict extubation was 0.921. The calibration and decision curve analysis showed well applicability of models. We also used the SHAP summary plot and PDP plot to demonstrate discriminative points of six key features in predicting extubation. Moreover, we employed LIME and SHAP force plots to show predicted probabilities of extubation and the rationale of the prediction at the individual level. Conclusions We developed an extubation prediction model with high accuracy and visualised explanations aligned with clinical workflow, and the model may serve as an autonomous screen tool for timely weaning. |
first_indexed | 2024-04-12T06:49:02Z |
format | Article |
id | doaj.art-4d428658243b49f6821433c6e921926e |
institution | Directory Open Access Journal |
issn | 1471-2253 |
language | English |
last_indexed | 2024-04-12T06:49:02Z |
publishDate | 2022-11-01 |
publisher | BMC |
record_format | Article |
series | BMC Anesthesiology |
spelling | doaj.art-4d428658243b49f6821433c6e921926e2022-12-22T03:43:26ZengBMCBMC Anesthesiology1471-22532022-11-0122111110.1186/s12871-022-01888-yExplainable machine learning approach to predict extubation in critically ill ventilated patients: a retrospective study in central TaiwanKai-Chih Pai0Shao-An Su1Ming-Cheng Chan2Chieh-Liang Wu3Wen-Cheng Chao4College of Engineering, Tunghai UniversityArtificial Intelligence Center, Tunghai UniversityDivision of Critical Care and Respiratory Therapy, Department of Internal Medicine, Taichung Veterans General HospitalDepartment of post-Baccalaureate Medicine, College of Medicine, National Chung Hsing UniversityDepartment of post-Baccalaureate Medicine, College of Medicine, National Chung Hsing UniversityAbstract Background Weaning from mechanical ventilation (MV) is an essential issue in critically ill patients, and we used an explainable machine learning (ML) approach to establish an extubation prediction model. Methods We enrolled patients who were admitted to intensive care units during 2015–2019 at Taichung Veterans General Hospital, a referral hospital in central Taiwan. We used five ML models, including extreme gradient boosting (XGBoost), categorical boosting (CatBoost), light gradient boosting machine (LightGBM), random forest (RF) and logistic regression (LR), to establish the extubation prediction model, and the feature window as well as prediction window was 48 h and 24 h, respectively. We further employed feature importance, Shapley additive explanations (SHAP) plot, partial dependence plot (PDP) and local interpretable model-agnostic explanations (LIME) for interpretation of the model at the domain, feature, and individual levels. Results We enrolled 5,940 patients and found the accuracy was comparable among XGBoost, LightGBM, CatBoost and RF, with the area under the receiver operating characteristic curve using XGBoost to predict extubation was 0.921. The calibration and decision curve analysis showed well applicability of models. We also used the SHAP summary plot and PDP plot to demonstrate discriminative points of six key features in predicting extubation. Moreover, we employed LIME and SHAP force plots to show predicted probabilities of extubation and the rationale of the prediction at the individual level. Conclusions We developed an extubation prediction model with high accuracy and visualised explanations aligned with clinical workflow, and the model may serve as an autonomous screen tool for timely weaning.https://doi.org/10.1186/s12871-022-01888-yMechanical ventilationExtubationMachine learningExplanationCritically ill patients |
spellingShingle | Kai-Chih Pai Shao-An Su Ming-Cheng Chan Chieh-Liang Wu Wen-Cheng Chao Explainable machine learning approach to predict extubation in critically ill ventilated patients: a retrospective study in central Taiwan BMC Anesthesiology Mechanical ventilation Extubation Machine learning Explanation Critically ill patients |
title | Explainable machine learning approach to predict extubation in critically ill ventilated patients: a retrospective study in central Taiwan |
title_full | Explainable machine learning approach to predict extubation in critically ill ventilated patients: a retrospective study in central Taiwan |
title_fullStr | Explainable machine learning approach to predict extubation in critically ill ventilated patients: a retrospective study in central Taiwan |
title_full_unstemmed | Explainable machine learning approach to predict extubation in critically ill ventilated patients: a retrospective study in central Taiwan |
title_short | Explainable machine learning approach to predict extubation in critically ill ventilated patients: a retrospective study in central Taiwan |
title_sort | explainable machine learning approach to predict extubation in critically ill ventilated patients a retrospective study in central taiwan |
topic | Mechanical ventilation Extubation Machine learning Explanation Critically ill patients |
url | https://doi.org/10.1186/s12871-022-01888-y |
work_keys_str_mv | AT kaichihpai explainablemachinelearningapproachtopredictextubationincriticallyillventilatedpatientsaretrospectivestudyincentraltaiwan AT shaoansu explainablemachinelearningapproachtopredictextubationincriticallyillventilatedpatientsaretrospectivestudyincentraltaiwan AT mingchengchan explainablemachinelearningapproachtopredictextubationincriticallyillventilatedpatientsaretrospectivestudyincentraltaiwan AT chiehliangwu explainablemachinelearningapproachtopredictextubationincriticallyillventilatedpatientsaretrospectivestudyincentraltaiwan AT wenchengchao explainablemachinelearningapproachtopredictextubationincriticallyillventilatedpatientsaretrospectivestudyincentraltaiwan |