昆明医科大学学报2025，Vol.46Issue(8)：33-42,10.DOI:10.12259/j.issn.2095-610X.S20250805
基于Lasso-Logistic回归模型拟合NRS 2002评分、CONUT评分和RDW-SD、ALB评估肺结核患者静脉血栓栓塞症风险的价值
The Value of Fitting NRS 2002,CONUT,RDW-SD and ALB in Assessing the Risk of Venous Thromboembolism in Patients with the Pulmonary Tuberculosis Based on Lasso-Logistic Regression Model
摘要
Abstract
Objective To evaluate the relationship between nutritional parameters and the risk of venous thromboembolicism(VTE)in patients with tuberculosis so as to identify the risk factors and predictors of thrombosis and assist in the early identification of high-risk factors for VTE in patients with the pulmonary tuberculosis.Methods A total of 323 patients diagnosed with the pulmonary tuberculosis and hospitalized in Kunming Third People's Hospital from August 2021 to August 2023 were collected.According to the VTE risk assessment of non-operative patients,they were divided into the high-risk group and the low-risk group respectively with 116 and 207 in each group.The nutritional indicators with statistically significant differences between the two groups were screened by Lasso regression.Multivariate Logistic regression was used to screen the independent risk factors for high VTE risk in pulmonary tuberculosis patients,and a nomogram prediction model was constructed.The prediction model was evaluated by receiver operating characteristic curve(ROC),calibration curve,decision curve,and influence curve.Results Patients in the high-risk group were significantly older than those in the low-risk group(59 vs.41,P<0.001),hypertension,gender,and Type 2 diabetes did not differ significantly(P values were 0.084,0.724 and 0.488,respectively).9 variables were selected from the inter-group comparison and Lasso regression,including ALB,HCT,NRS2002 scores,HBDH,RDW-SD,RDW-CV,TG,CONUT scores,and NEFA.Multivariate Logistic regression analysis showed that ALB,NRS2002 scores,RDW-SD,and CONUT scores were independent influencing factors for the high risk of VTE scores in patients with tuberculosis(P<0.005).Area under the ROC curve showed that the AUC(0.892)for high-risk VTE scores in patients with the pulmonary tuberculosis was greater than that of ALB(0.803),NRS2002 score(0.735),RDW-SD(0.685),and CONUT score(0.774).Fitting prediction model:Logit(P):Y=0.433×NRS-0.136×ALB+0.411×CONUT score+0.072×RDW-SD-1.770,P=1/(1+e-Y)(Y:prediction index,P:prediction probability).Calibration curve showed that the model prediction tended to be consistent with the actual results(U:>0.05),and the decision curve and influence curve showed that the model can bring clinical benefits.Conclusion ALB,NRS2002 scores,RDW-SD,and CONUT scores are independent influencing factors for the high risk of VTE scores in patients with tuberculosis.They can guide the clinical practice,improve these indicators as soon as possible,reduce VTE scores,and reduce the thrombosis risk.At the same time,the prediction model performs well in the verification cohort,with its discrimination ability,calibration accuracy and clinical utility(decision curve analysis)all reaching a satisfactory level.关键词
肺结核/静脉血栓栓塞/营养风险筛查2002/控制营养评分/红细胞分布宽度标准差/血清白蛋白/营养评价Key words
Tuberculosis pulmonary/Venous thromboembolism/NRS 2002/CONUT/RDW-SD/Serum albumin/Nutrition assessment分类
医药卫生引用本文复制引用
施春晶,李文明,罗云,张华杰,韩留鑫,赵雁红,李龙芬,沈凌筠..基于Lasso-Logistic回归模型拟合NRS 2002评分、CONUT评分和RDW-SD、ALB评估肺结核患者静脉血栓栓塞症风险的价值[J].昆明医科大学学报,2025,46(8):33-42,10.基金项目
国家自然科学基金(82460001) （82460001）
云南省科技厅地方高校联合专项基金(202401BA070001-063,202401BA-070001-082) （202401BA070001-063,202401BA-070001-082）
昆明市卫健委科研课题项目(2023-03-08-012,2023-03-02-019) （2023-03-08-012,2023-03-02-019）
昆明市卫生科技人才医学技术中心建设项目[2024-SW(技术)-11] （技术）
