摘要
目的 :研究无创参数SpO_2/FiO_2(S/F)和PaO_2/FiO_2(P/F)之间的关系,探索使用无创参数辨识急性呼吸窘迫综合征(acute respiratory distress syndrome,ARDS)患者疾病严重程度的可能性。方法:访问重症医学信息数据库(Medical Information Mart for Intensive Care,MIMIC-Ⅲ),获取患者相关生理参数,将患者数据随机分为训练集和测试集。使用训练集,利用广义线性回归模型,建立lg(S/F)与lg(P/F)之间的线性关系,选择最优的回归方程作为对数线性回归模型,并将lg(S/F)与lg(P/F)之间的线性关系和S/F与P/F之间的线性关系进行对比研究。使用测试集,比较2种模型对于P/F在100(重度ARDS)、200(中度ARDS)、300(轻度ARDS)处的辨识效果。结果:在训练集(n=61 634)上推导出lg(P/F)与lg(S/F)之间的线性关系:lg(S/F)=1.277+0.437×lg(P/F)(r=0.66,P<0.000 1),并确定了P/F在100、200、300处对应的S/F阈值为131、201、271。在测试集(n=26 758)上使用S/F阈值验证了其辨识效果,且较传统线性回归模型辨识结果有了明显改善。结论:通过研究,认为在患者血气分析结果缺失的情况下,可以使用无创参数SpO_2/FiO_2代替PaO_2/FiO_2作为ARDS患者病情诊断的辅助手段。
Objective To explore the relationship between SpO2/FiO2(S/F)and PaO2/FiO2(P/F)so as to determine the possibility of ARDS severity identification based on noninvasive parameters.Methods The physiological parameters of corresponding patients were acquired from Medical Information Mart for Intensive Care(MIMIC-Ⅲ),and then divided into a training set and a test set randomly.In the training set the linear relationship between lg(S/F)and lg(P/F)was established with generalized linear regression model,and a log linear regression model was formed with the optimal regression equation;the linear relationship between lg(S/F)and lg(P/F)was compared with that between S/F and P/F.In the test set,the two models were compared on the identification of ARDS in case P/F values were 100(mild ARDS),200(moderate ARDS)and 300(severe ARDS)respectively.Results In the training set(n=61 634)the linear relationship between lg(S/F)and lg(P/F)was deduced as lg(S/F)=1.277+0.437×lg(P/F)(r=0.66,P<0.000 1),and the S/F thresholds in case P/F values were 100,200 and 300 respectively were 131,201 and 271.In the test set(n=26 758)the identification effect was verified with the acquired S/F thresholds,which proved better than that of traditional regression model.Conclusion Noninvasive parameter SpO2/FiO2 can replace PaO2/FiO2 for the auxiliary diagnosis of ARDS in case the result of blood gas analysis is absent.[Chinese Medical Equipment Journal,2018,39(3):6-9,14]
作者
赵惠军
杨鹏程
王亚林
ZHAO Hui-jun;YANG Peng-cheng;WANG Ya-lin(Medical Engineering Department,Naval General Hospital,Beijing 100048,China;nstitute of Medical Support Technology,Academy of System Engineering of Academy of Military Science of Chinese PLA,Tianjin 300161,China)
出处
《医疗卫生装备》
CAS
2018年第3期6-9,14,共5页
Chinese Medical Equipment Journal
基金
国家自然科学基金项目(81501551)
国家重点研发计划(2017YFC0806402)
军事医学创新工程专项(16CXZ034)