The final, formatted version of the article will be published soon.
ORIGINAL RESEARCH article
Front. Cardiovasc. Med.
Sec. Intensive Care Cardiovascular Medicine
Volume 12 - 2025 |
doi: 10.3389/fcvm.2025.1488097
In-hospital mortality prediction of critically ill patients with pulmonary embolism complicated by heart failure based on a machine learning method
Provisionally accepted- Department of Cardiology, Fourth Hospital of Hebei Medical University, Shijiazhuang, Hebei Province, China
Background: Pulmonary embolism(PE) patients have been reported a high proportion when complicated by heart failure(HF). However, few studies have observed the in-hospital mortality and developed predictive tools of in-hospital mortality for this group in intensive care unit(ICU).This study aimed to construct and validate a machine learning(ML) model to predict in-hospital mortality for PE patients complicated by HF in ICU.Methods:We included PE patients complicated by HF from the Medical Information Mart for Intensive Care Database -IV (MIMIC-IV). Six ML models were processed after feature selection by logistic regression and least absolute shrinkage and selection operator (LASSO) regression analysis. Further, patients from eICU-CRD were extracted for external validation. The area under curves(AUC), calibration curves, decision curve analysis(DCA),net reclassification improvement(NRI), and integrated discrimination improvement(IDI) were all involved to evaluate the efficiency of each model. SHapley Additive exPlanations(SHAP) was performed for the interpretability of the optimal model. Results:A total of 506 PE patients complicated by HF were included as derivation cohort, while 182 patients for external validation. We developed six ML models by the 12 selected features. After internal validation, the extreme gradient boosting(XGBoost) model performed best with an AUC of 0.821, a superior calibration degree, and a wider risk threshold(from 10 to 70%) for obtaining clinical benefit, which also outperformed traditional mortality risk evaluation systems when evaluated by NRI and IDI. The XGBoost model was still reliable after external validation. SHAP was performed to explain the model. Ultimately, an online application was developed for further clinical use.This study developed a potential tool for detecting in-hospital mortality risk to guide clinical decision making for PE patients complicated by HF in the ICU. The SHAP method also helped clinicians to better understand the model.
Keywords: machine learning, Intensive Care Unit, Pulmonary Embolism, Heart Failure, Prediction model, In-hospital mortality
Received: 27 Sep 2024; Accepted: 14 Jan 2025.
Copyright: © 2025 Liu, Ma, Yao, Liu, Guo, Guan, Gao, Wang and Li. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
* Correspondence:
Jing Liu, Department of Cardiology, Fourth Hospital of Hebei Medical University, Shijiazhuang, Hebei Province, China
Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.