浙江医学2025，Vol.47Issue(3)：257-262,后插2,7.DOI:10.12056/j.issn.1006-2785.2025.47.3.2024-1502
基于LASSO回归的重型颅脑损伤患者并发CRKP感染列线图模型构建与评价
Construction and evaluation of a nomogram model for CRKP infection in patients with severe traumatic brain injury based on LASSO regression
沈翔 1何兰兰 2杜雷涛 3朱岗 3董德胜 4张夏兰 2盛文国2
作者信息
- 1. 313000 湖州学院附属南太湖医院护理部
- 2. 313000 湖州学院附属南太湖医院神经医学科
- 3. 陆军第72集团军医院重症医学科
- 4. 陆军第72集团军医院急诊科
- 折叠
摘要
Abstract
Objective To construct and evaluate a nomogram model for carbapenem resistant Klebsiella pneumonia(CRKP)infection in patients with severe traumatic brain injury(sTBI)based on least absolute shrinkage and selection operator(LASSO)regression.Methods A total of 159 sTBI patients with Klebsiella pneumonia infection were collected in South Taihu Hospital Affiliated to Huzhou College from June 2019 to December 2023 and Hospital of the 72nd Group Army from January 2016 to December 2023.According to drug resistance,40 cases were divided into the CRKP group and 119 cases were divided into carbapenem-sensitive Klebsiella pneumonia group.The characteristics of CRKP infection was analyzed,LASSO regression was used to screen the optimal feature variables,and multivariate logistic regression analysis was used to analyze independent influencing factors for CRKP infection in sTBI patients to construct a nomogram model.ROC curve,calibration curve and decision curve were drawn to evaluate the differentiation,calibration and clinical net benefit of the nomogram model,respectively.Results The detection rate of CRKP infection was 25.16％,and the isolated specimens were mainly sputum and urine,accounting for 80.85％and 8.51％,respectively.Based on LASSO regression,six non-zero coefficient variables were screened,including age,multiple bacterial infections,traumatic cerebral infarction,Glasgow Coma Scale(GCS),controlling nutritional status(CONUT)score,and intensive care unit(ICU)stay duration.Multivariate logistic regression analysis showed that age,multiple bacterial infections,traumatic cerebral infarction,GCS,CONUT score and ICU stay duration were independent risk factors for sTBI complicated with CRKP infection(all P＜0.05).Based on the above risk factors,the nomogram model of CRKP infection was established,and results showed that the AUC predicted by the model for CRKP infection was 0.906(95％CI:0.849-0.962).The calibration curve showed good consistency between the predicted probability of the model and the actual probability(P=0.673).The decision curve showed that the model had a high net benefit when the risk thresholds for CRKP occurrence were 0.05-0.89 and 0.92-0.94.Conclusion The nomogram model construction based on LASSO regression has good predictive efficacy,and can be used as an assessment tool for screening CRKP infection in sTBI patients.关键词
重型颅脑损伤/耐碳青霉烯类肺炎克雷伯菌感染/最小绝对值收缩和选择算子回归/列线图模型Key words
Severe traumatic brain injury/Carbapenem-resistant Klebsiella pneumoniae infection/Least absolute shrinkage and selection operator regression/Nomogram model引用本文复制引用
沈翔,何兰兰,杜雷涛,朱岗,董德胜,张夏兰,盛文国..基于LASSO回归的重型颅脑损伤患者并发CRKP感染列线图模型构建与评价[J].浙江医学,2025,47(3):257-262,后插2,7.
