基于序贯器官功能衰竭评分的LASSO-Logistic诊断模型在脓毒症中的效能分析
Analysis of the efficacy of LASSO-Logistic diagnostic model based on sequential organ failure assessment score in sepsis
张立琳 1章金鹏 2金律 2王雷 2蔡榕松 2杨亚东3
作者信息
- 1. 长江大学医学部 湖北 荆州 434023
- 2. 长江大学附属黄冈市中心医院 湖北 黄冈 438000
- 3. 长江大学医学部 湖北 荆州 434023;长江大学附属黄冈市中心医院 湖北 黄冈 438000
- 折叠
摘要
目的:使用LASSO-Logistic分析影响脓毒症发生的相关因素,构建基于序贯器官功能衰竭评分(SOFA)的脓毒症诊断模型,并进行深度验证.方法:本研究回顾性收集了长江大学附属黄冈市中心医院重症医学科在 2020年 7月至 2022年 10月间收治的危重症患者在入科 24 h内包括实验室检测指标在内的临床数据,采用SOFA评分评估患者器官功能障碍.依据患者入住ICU后的诊断分为脓毒症组和非脓毒症组,采用最小绝对收缩和选择算子(LASSO)回归模型初步筛选变量,使用多因素Logistic回归建立脓毒症诊断模型,计算曲线下面积(AUC),应用K折交叉验证深度评估模型效能.结果:共纳入患者235例,其中130例诊断为脓毒症,非脓毒症患者105例.LASSO-Logistic分析显示,抗凝血酶Ⅲ活性物质(ATⅢ)、纤维蛋白原(FIB)、超敏C反应蛋白(hs-CRP)和SOFA评分被评定为独立危险因子,将其构建联合模型,模型的AUC值为0.928(95%CI:0.896～0.960),优于各单项指标(P<0.05),100 次 10 折交叉验证提示模型具有较好的泛化能力.结论:基于 SOFA 评分构建LASSO-Logistic诊断模型,对脓毒症诊断表现出较好预测效能和泛化能力,可提高对脓毒症患者的早期诊断.
Abstract
Objective:To analyze the relevant factors affecting the occurrence of sepsis using LASSO-Logistics,to construct the efficacy of a sequential organ failure assessment(SOFA)score-based diagnostic model for sepsis,and to conduct in-depth validation.Methods:This study retrospectively collected clinical data of critically ill patients,including laboratory test indicators,ad-mitted to the Department of Critical Care Medicine(CCM)of Huanggang Central Hospital affiliated to Yangtze University between July 2020 and October 2022 within 24 hours of admission,and used the SOFA score to assess patients'organ dysfunction.Patients were divided into sepsis and non-sepsis groups based on their diagnosis on admission to the intensive care unit,and variables were initially screened using a LASSO regression model,and a sepsis diagnostic model was developed using multi-factorial logistic regression to calculate the area under the curve(AUC).Results:A total of 235 pa-tients were included,and 130 were diagnosed with sepsis and 105 without.LASSO-logistic analysis showed that antithrombin Ⅲ active substance(ATⅢ),fibrinogen(FIB),hypersensitive C-reactive protein(hs-CRP),and SOFA were assessed as independent risk factors,and a joint diagnostic model was constructed.The joint model with an AUC value of 0.928(95%CI:0.896-0.960)was better than each indicator(P<0.05),with better generalization ability by the ten-fold cross-validated sugges-tive model.Conclusion:The LASSO-logistic diagnostic model was constructed based on the SOFA score,which showed better predictive efficacy and generalization ability for sepsis diagnosis and could improve the early diagnosis of septic patients.
关键词
脓毒症/SOFA评分/LASSO回归/预后分析:K折交叉验证/深度验证Key words
Sepsis/SOFA Score/LASSO Regression/Prognostic Analysis/K Fold Cross Validation/Deep Validation引用本文复制引用
基金项目
湖北省自然科学基金资助项目(2021CFB530)
出版年
2024
NETL
NSTL
万方数据