首页|期刊导航|遵义医科大学学报|基于四项常规实验室指标的主动脉夹层与肺栓塞鉴别诊断模型构建

基于四项常规实验室指标的主动脉夹层与肺栓塞鉴别诊断模型构建OA

Construction of a differential diagnosis model for aortic dissection and pulmo-nary embolism based on four routine laboratory indicators

中文摘要英文摘要

目的 构建基于常规实验室指标的主动脉夹层(AD)与肺栓塞(PE)鉴别诊断模型,评估其诊断效能并明确核心影响特征,为临床早期精准鉴别提供辅助工具.方法 收集AD与PE患者的临床基线资料,采用Boruta法筛选最优特征变量,基于逻辑回归算法构建鉴别诊断模型;通过训练集与测试集验证模型效能,结合SHAP方法分析特征贡献度以提升模型可解释性.结果 最终纳入淋巴细胞比值(LYM-R)、前白蛋白(PA)、纤维蛋白原(Fib)、甘油三酯(TG)4项核心特征构建模型,最佳截断值为0.65.训练集AUC为0.79,各维度诊断指标良好,模型泛化稳定性佳,且SHAP分析证实淋巴细胞比值为鉴别核心特征.结论 模型依托常规实验室指标构建,内部验证显示区分度与校准度尚可,结合SHAP分析可解释性强,可作为急诊AD与PE快速初筛的辅助工具.

Objective To construct a differential diagnosis model for aortic dissection(AD)and pulmonary em-bolism(PE)based on the logistic regression algorithm,evaluate its diagnostic efficacy,and identify core influ-ential features,so as to provide an auxiliary tool for early accurate clinical differentiation.Methods Clinical baseline data of patients with AD and PE were collected.The Boruta method was used to screen optimal feature variables,and a differential diagnosis model was built based on the logistic regression algorithm.The model's ef-ficacy was verified using training and test sets,and the SHAP method was integrated to analyze feature contribu-tions for improving model interpretability.Results Finally,four core features lymphocyte ratio(LYM-R),preal-bumin(PA),fibrinogen(Fib),and triglycerides(TG)were included to construct the model,with an optimal cut-off value of 0.65.The AUC of the training set was 0.79,with favorable diagnostic indicators in all dimen-sions and good generalization stability of the model.SHAP analysis confirmed that LYM-R was the core feature for differentiation.Conclusion The model was developed based on routine laboratory indicators.Internal valida-tion shows acceptable discrimination and calibration,and it features strong interpretability combined with SHAP analysis,thus serving as an auxiliary tool for the rapid initial screening of AD and PE in emergency settings.

吴东阳;李佩轩;罗滔;程莉钧;彭镨;穆铃;刘英

西南医科大学附属医院急诊医学部,四川泸州 646000西南医科大学附属医院急诊医学部,四川泸州 646000西南医科大学附属医院急诊医学部,四川泸州 646000西南医科大学附属医院急诊医学部,四川泸州 646000西南医科大学附属医院急诊医学部,四川泸州 646000叙永县人民医院急诊医学科,四川泸州 646400西南医科大学附属医院急诊医学部,四川泸州 646000

医药卫生

主动脉夹层肺栓塞鉴别诊断逻辑回归预测模型

aortic dissectionpulmonary embolismdifferential diagnosislogistic regressionprediction models

《遵义医科大学学报》 2026 (9)

1002-1010,9

四川省科技厅重点研发计划项目(NO:2023YFS0241)叙永县人民医院-西南医科大学科技战略合作项目(NO:2024XYXNYD22).

评论