Establishment of a prediction model for in-hospital mortality risk in patients with sepsis-induced coagulopathy based on LASSO regression
10.3969/j.issn.1673-4130.2024.15.015
- VernacularTitle:基于LASSO回归的脓毒症凝血病患者院内死亡风险预测模型建立
- Author:
Xueyan FAN
1
;
Zuyu ZHANG
;
Heng ZHAO
;
Fei ZHOU
;
Chenming DONG
Author Information
1. 兰州大学第二临床医学院,甘肃兰州 730000
- Keywords:
sepsis-induced coagulopathy;
prediction model;
LASSO regression
- From:
International Journal of Laboratory Medicine
2024;45(15):1874-1882
- CountryChina
- Language:Chinese
-
Abstract:
Objective To establish a prediction model for in-hospital mortality risk in patients with sepsis-induced coagulopathy based on LASSO regression.Methods Patients with sepsis-induced coagulopathy ad-mitted to intensive care unit(ICU)at Beth Israel Deaconess Medical Center during 2008 to 2019 were selected from the Medical Information Market for Intensive Care(MIMIC)-Ⅳ database(version 2.1)for retrospective study.The study subjects were randomly divided into modeling group and verification group,and the feature variables were screened by LASSO regression.The feature variables were analyzed by multivariate Logistic re-gression to determine independent risk factors,and the nomogram prediction model was established at the same time.The model performance was evaluated by drawing calibration curve and receiver operating charac-teristic(ROC)curve,as well as decision curve analysis.Results A total of 4 994 patients with sepsis-induced coagulopathy admitted to ICU for the first time were enrolled in this study.They were randomly divided into a model group(n=3 495)and a validation group(n=1 499)at a ratio of 7:3.Multivariate Logistic regres-sion analysis showed that age,mean respiratory rate,mean corpuscular hemoglobin concentration,red blood cell count,platelet count,prothrombin time,anion gap,acute physiological score Ⅲ and acute kidney injury were independent risk factors for in-hospital mortality of patients with sepsis-induced coagulopathy.Based on the above independent risk factors,a nomographic prediction model was constructed.The area under the ROC curve and 95％confidence interval of the nomogram in the modeling group and validation group were 0.864(0.849-0.880)and 0.877(0.852-0.901),respectively.The sensitivity was 0.795 and 0.763,and the speci-ficity was 0.779 and 0.843,respectively.The calibration curve suggested that the predicted probability was ba-sically consistent with the actual probability,and the decision curve analysis showed that it had good clinical net benefits within a wide range of threshold.Conclusion The nomogram model based on MIMIC-Ⅳ database has good predictive value for predicting the in-hospital mortality of patients with sepsis-induced coagulopathy and can be used to guide clinical work.
