AUTHOR=Jin Jun , Yu Lei , Zhou Qingshan , Zeng Mian 

TITLE=Improved prediction of sepsis-associated encephalopathy in intensive care unit sepsis patients with an innovative nomogram tool

JOURNAL=Frontiers in Neurology

VOLUME=Volume 15 - 2024

YEAR=2024

URL=https://www.frontiersin.org/journals/neurology/articles/10.3389/fneur.2024.1344004

DOI=10.3389/fneur.2024.1344004

ISSN=1664-2295

ABSTRACT=Sepsis-associated encephalopathy (SAE) is a complication arising from sepsis-induced systemic inflammation, often observed in ICU patients, with many survivors facing long-term cognitive issues. This study aimed to develop a predictive nomogram for identifying SAE risk factors in ICU sepsis patients, using data from the MIMIC-IV database. SAE was defined by a Glasgow Coma Scale score ≤15 or delirium presence. Patients were split into training and validation groups, with LASSO regression used for feature optimization and multivariable logistic regression to identify independent risk factors and construct the prediction model. The model's efficacy was evaluated through metrics like AUC, calibration plots, Hosmer-Lemeshow test, DCA, NRI, and IDI. Out of 4,476 sepsis patients analyzed, 62.1% developed SAE, with the SAE group showing higher in-hospital mortality (9.5%) compared to non-SAE patients (3.7%). Key variables such as age, gender, BMI, mean arterial pressure, temperature, platelet count, sodium level, and midazolam use were instrumental in the nomogram's development, which outperformed traditional methods like the SOFA score in predicting SAE. The study highlights significant risk factors for SAE in sepsis patients and presents a reliable model for early detection, offering a crucial clinical tool for managing SAE risk.