Applying an Improved Stacking Ensemble Model to Predict the Mortality of ICU Patients with Heart Failure

Area of Science:

• Medical Informatics
• Clinical Prediction Models
• Cardiovascular Research

Background:

• Cardiovascular diseases, particularly heart failure (HF), are leading global causes of mortality.
• Intensive Care Units (ICUs) manage high-risk HF patients, necessitating precise mortality prediction for timely interventions.
• Accurate prediction of mortality risk in ICU patients with HF is crucial for resource allocation and patient management.

Purpose of the Study:

• To develop and validate an integrated stacking model for precise prediction of mortality in Intensive Care Unit (ICU) patients with heart failure (HF).
• To identify key clinical features influencing HF patient mortality prediction within the first 24 hours of ICU admission.

Main Methods:

• Utilized data from 6699 HF patients in the MIMIC-III database, focusing on vital signs and tests within the first 24 hours of ICU admission.
• Employed an integrated stacking model with six first-level classifiers (RF, SVC, KNN, LGBM, Bagging, Adaboost) and a second-level classifier for enhanced prediction accuracy.
• Evaluated model performance using accuracy and Area Under the Receiver Operating Characteristic Curve (AUROC).

Main Results:

• The proposed integrated stacking model achieved a high accuracy of 95.25% and an AUROC of 82.55% in predicting HF patient mortality.
• Key clinical features significantly impacting mortality prediction included platelets, glucose, and blood urea nitrogen.
• The model demonstrated outstanding capability and efficiency in predicting mortality risk for HF patients in the ICU.

Conclusions:

• The developed integrated stacking model offers a highly accurate and efficient tool for predicting mortality in ICU patients with heart failure.
• Identifying critical predictive features like platelets, glucose, and BUN enhances clinical understanding and supports optimized healthcare resource utilization.
• This predictive capability facilitates timely and appropriate medical care for high-risk HF patients, potentially improving outcomes.