Hiring in life sciences? Share your open positions with our professional community. Read more Close

Advertisement

[Development and evaluation of a machine learning-based model for predicting weaning outcomes from mechanical ventilation in patients with congestive heart failure].

Created on 18 Nov 2026

Authors

Jiapan An, Hongmei Shen, Xinqi Xu, Tingyu Yang, Bin Li, Zhimin Dou

Published in

Zhonghua wei zhong bing ji jiu yi xue. Volume 38. Issue 10. Pages 916-925.

Abstract

To develop a machine learning-based predictive model for weaning outcomes from mechanical ventilation in patients with congestive heart failure (CHF), evaluate its performance, and to identify key influencing factors.
A retrospective cohort study was conducted. Clinical data of CHF patients who were first admitted to the intensive care unit (ICU) and received invasive mechanical ventilation between 2008 and 2022 were extracted from the Medical Information Mart for Intensive Care-IV v3.1 (MIMIC-IV v3.1) database. Patients were divided into a successful weaning group and a failed weaning group according to weaning outcomes. Data on demographics, vital signs, ventilator parameters, and laboratory indicators were extracted using structured query language (SQL). The dataset was randomly split into a training set and a validation set at a ratio of 7 : 3. Variables with statistical significance in univariate Logistic regression analysis were selected and then included in multivariate Logistic regression analysis to further screen the final predictors for weaning failure. Based on the selected variables, six machine learning models, including decision tree (DT), random forest (RF), extreme gradient boosting (XGBoost), light gradient boosting machine (LightGBM), support vector machine (SVM), and artificial neural network (ANN), were constructed. Model performance was assessed using the area under the receiver operator characteristic curve (AUC), accuracy, precision, sensitivity, and F1-score. SHapley Additive exPlanations (SHAP) values were used to visually interpret the optimal model and quantify the contribution of each feature to the prediction.
A total of 2 496 CHF patients who were first admitted to the ICU and received invasive mechanical ventilation were enrolled finally, including 1 367 in the successful weaning group and 1 129 in the failed weaning group, with an overall weaning failure rate of 45.23%. Through univariate and multivariate Logistic regression analyses, eight independent predictors were identified. Tracheostomy [odds ratio (OR)=3.194, 95% confidence interval (95%CI) was 2.169-4.704, P<0.001], prolonged mechanical ventilation duration (OR=1.002, 95%CI was 1.001-1.003, P<0.001), elevated MB isoenzyme of creatine kinase (CK-MB; OR=1.009, 95%CI was 1.004-1.013, P<0.001), elevated cardiac troponin T (cTnT; OR=1.158, 95%CI was 1.033-1.298, P=0.012), and advanced age (OR=1.010, 95%CI was 1.001-1.019, P=0.027) increased the risk of weaning failure, being considered as risk factors for failure weaning, while higher oxygenation index (PaO2/FiO2; OR=0.999, 95%CI was 0.998-1.000, P=0.014), increased 24-hour urine output (OR=0.824, 95%CI was 0.760-0.894, P<0.001), and higher Glasgow Coma Scale (GCS) score (OR=0.939, 95%CI was 0.914-0.966, P<0.001) reduced the risk of weaning failure, being considered as protective factors for successful weaning. Among the six machine learning models, the XGBoost model demonstrated the best comprehensive performance in the validation set (AUC=0.775, accuracy was 69.8%, precision was 67.8%, sensitivity was 63.0%, F1-score was 0.653). SHAP analysis revealed that mechanical ventilation duration and 24-hour urine output before weaning were the most critical factors influencing the model prediction, followed by tracheostomy, cTnT, CK-MB, GCS score, PaO2/FiO2, and age.
The XGBoost-based prediction model for weaning failure in CHF patients, incorporating eight readily available clinical indicators, including mechanical ventilation duration, 24-hour urine output before weaning, tracheostomy, cTnT, CK-MB, GCS score, PaO2/FiO2, and age, can effectively identify high-risk individuals among CHF patients on mechanical ventilation.

PMID:
43045522
Bibliographic data and abstract were imported from PubMed on 18 Nov 2026.

Read full publication at:
Please sign in to see all details.

Advertisement

Stats

  • Community rating n/a 0 votes
  • Reviewers' rating n/a 0 votes
  • Your rating

1-terrible, 9-excellent. How would you rate this publication? Sign in in to submit your rating.

  • Recommendations n/a n/a positive of 0 vote(s)
  • Views 4
  • Comments 0

Recommended by

  • No recommendations yet.

Post a comment

You need to be signed in to post comments. You can sign in here.

Comments

There are no comments yet.

Advertisement