Development and Validation of an Early Risk Prediction Model for Sepsis-Induced Coagulopathy Based on Machine...

Area of Science:

• Critical Care Medicine
• Machine Learning in Healthcare
• Coagulation Disorders

Background:

• Sepsis-induced coagulation dysfunction (SIC) is a severe complication in intensive care units (ICUs).
• Early prediction of SIC risk is crucial for timely intervention and improved patient outcomes.
• Current diagnostic criteria for SIC may not be met at the time of ICU admission.

Purpose of the Study:

• To develop and validate machine learning models for the early prediction of SIC risk.
• To identify key clinical factors associated with SIC development upon ICU admission.
• To assess the predictive performance of different machine learning models for SIC.

Main Methods:

• A retrospective cohort study involving septic ICU patients.
• Development and validation of four machine learning models, including a random forest model.
• Feature selection using LASSO and logistic regression, with 10-fold cross-validation and external validation.

Main Results:

• 57.9% of patients developed SIC.
• Key predictors identified include prothrombin time, diastolic blood pressure, activated partial thromboplastin time, D-dimer, white blood cell count, renal insufficiency, and partial pressure of carbon dioxide.
• The random forest model achieved an AUC of 0.76 (internal) and 0.79 (external).

Conclusions:

• The random forest model demonstrates moderate predictive capability for SIC risk upon ICU admission.
• Early identification of high-risk patients is feasible using machine learning.
• Further prospective validation is recommended prior to clinical implementation.