What question did this study set out to answer?

This study aims to develop an early warning model for predicting sepsis-associated encephalopathy (SAE).

March 26, 2026Critical Care Medicine0 citations

1555: Early Warning Prediction Model Construction and Validation of Sepsis-Associated Encephalopathy

WLWen LiShandong Provincial Hospital ZZZixuan Zhai JCJixuan Cui

Key Points

This study aims to develop an early warning model for predicting sepsis-associated encephalopathy (SAE).
Extracted clinical data from 4727 sepsis patients in the MIMIC-IV database.
Divided patients into SAE and non-SAE groups based on brain damage.
Used univariate and multivariate logistic regression to identify predictive factors.
Evaluated model performance with ROC curves, calibration curves, and decision curve analysis (DCA).
49.74% of sepsis patients developed SAE.
Identified 41 statistically significant indicators between SAE and Non-SAE groups.
Age at admission, ICU stay, and invasive ventilation were independent risk factors for SAE.
Model achieved an AUC of 0.80 for the training set and 0.79 for the validation set, indicating good prediction capability.

Abstract

Introduction: Sepsis-associated encephalopathy (SAE) presents a situation of high incidence, difficult diagnosis, and poor prognosis. Therefore, early identification of SAE is of vital importance. This study aims to construct an early warning and prediction model for SAE, with the expectation of providing a practical assessment tool for the early identification and timely intervention of SAE in clinical settings. Methods: Clinical data of 4727 sepsis patients from the Medical Information Mart for Intensive Care-IV (MIMIC-IV) database from 2008 to 2019 were extracted. Inclusion and exclusion criteria were established, and patients were divided into two groups based on whether they developed sepsis-associated brain damage: the SAE group and the non-sepsis-associated encephalopathy (Non-SAE) group. The data were divided into a training set and a validation set at a ratio of 7:3. Rank sum test and chi-square test were used to screen variables with statistical significance. Univariate and multivariate logistic regression were used to construct the prediction model. The model’s performance was evaluated using ROC curves, calibration curves, and DCA curves. Results: Among the 4727 sepsis patients, 2351 cases (49.74%) developed SAE. Group comparison showed that 41 indicators, including age at admission, ICU stay, and GCS score, were statistically significant between the SAE group and the Non-SAE group (P< 0.05). Univariate and multivariate logistic regression analysis indicated that invasive ventilation OR (95% CI)=2.31 (1.73-3.09), age at admission OR (95% CI)=1.02 (1.01-1.02), ICU stay OR (95% CI)=1.09 (1.06-1.11), respiratory rate OR (95% CI)=1.02 (1.01-1.04), and blood urea nitrogen OR (95% CI)=1.01 (1.01-1.01) were independent risk factors for SAE (P< 0.05), while GCS score OR (95% CI)=0.94 (0.91-0.97) and urine output OR (95% CI)=0.99 (0.99-0.99) were protective factors for SAE (P< 0.05). Model validation showed that the area under the ROC curve (AUC) of the training set was 0.80 (95% CI 0.79-0.82), and the AUC of the validation set was 0.79 (95% CI 0.76-0.81). And both the calibration curve and DCA curve indicated good model fit. Conclusions: The prediction model established in this study can effectively predict the occurrence of SAE. It is rapid to operate and has significant clinical value.

问 AI

Bookmark

View Full Paper