Development and Validation of an ICU-Venous Thromboembolism Prediction Model Using Machine Learning Approaches: A...

Area of Science:

• Medical Informatics
• Artificial Intelligence in Medicine
• Clinical Decision Support

Background:

• Venous thromboembolism (VTE) poses a significant risk to intensive care unit (ICU) patients.
• Early identification of high-risk individuals is crucial for effective VTE prevention strategies.

Purpose of the Study:

• To develop and validate machine learning (ML) models for predicting VTE risk in ICU patients.
• To identify key clinical variables associated with VTE development in this population.

Main Methods:

• Utilized clinical data from 1494 ICU patients across three tertiary hospitals.
• Employed the Boruta algorithm for feature selection and five ML algorithms (RF, XGBoost, SVM, GBDT, LR) for model development.
• Optimized hyperparameters and evaluated model performance using AUC, specificity, and F1 score.

Main Results:

• The incidence of VTE in the study cohort was 26.04%.
• Nineteen crucial features were identified for VTE risk prediction.
• The Random Forest (RF) model demonstrated superior performance with an AUC of 0.788.

Conclusions:

• Successfully developed and validated ML-based models for VTE risk prediction in ICU patients.
• The developed models can assist clinicians in early identification of high-risk patients.
• Facilitates timely implementation of preventive measures to reduce VTE burden in ICUs.