Related Experiment Video
Updated: Sep 14, 2025
14:08
Automated Midline Shift and Intracranial Pressure Estimation based on Brain CT Images
Published on: April 13, 2013
42.8K
Development and validation of machine learning-based risk prediction models for ICU-acquired weakness: a prospective
Yimei Zhang1, Yu Wang1, Jingran Yang1
1Department of Nursing, The First Affiliated Hospital of Kunming Medical University, No. 295, Xichang Road, Kunming, 650032, China.
European Journal of Medical Research
|July 24, 2025
Summary
Machine learning effectively predicts intensive care unit-acquired weakness (ICUAW) risk. This approach aids early intervention and standardized management, potentially reducing ICUAW incidence in critically ill patients.
Area of Science:
- Critical Care Medicine
- Machine Learning in Healthcare
- Predictive Analytics
Background:
- Intensive care unit-acquired weakness (ICUAW) is a common complication affecting critically ill patients, leading to significant long-term morbidity.
- Traditional risk prediction models struggle with the complexity and heterogeneity of critical illness, limiting early identification and prevention of ICUAW.
- Machine learning (ML) offers advanced capabilities for integrating diverse data to improve predictive accuracy and personalize risk assessment for ICUAW.
Purpose of the Study:
- To develop and validate machine learning models for predicting the risk of intensive care unit-acquired weakness (ICUAW).
- To leverage ML algorithms to identify high-risk patients for early intervention and improved outcomes.
- To explore the potential of ML in uncovering novel risk factors and mechanisms associated with ICUAW.
Main Methods:
- Four distinct machine learning algorithms were utilized for model development.
- Patient data, including clinical, laboratory, and physiological parameters, were assessed using bedside ultrasound within 24 hours and on day 7 of ICU admission.
- Model performance was rigorously evaluated using metrics such as Area Under the Receiver Operating Characteristic Curve (AUC), accuracy, sensitivity, specificity, and F1 score.
Main Results:
- The study enrolled 749 patients, with 51% developing ICUAW.
- The eXtreme Gradient Boosting model demonstrated superior performance, achieving an AUC of 0.978, with high accuracy (0.924), sensitivity (0.911), and specificity (0.941).
- Decision Curve Analysis further supported the clinical utility and robustness of the developed ML prediction models.
Conclusions:
- Machine learning models can effectively identify patients at high risk for developing intensive care unit-acquired weakness (ICUAW).
- The developed ML prediction model offers a standardized approach for managing ICUAW.
- Implementing this ML-based prediction strategy holds promise for reducing the incidence of ICUAW and improving patient prognosis.
