利用判别分析预测慢加急性肝衰竭患者的治疗效果  

Study of discriminant analysis to predict the therapeutic effect of acute-on-chronic liver failure

作　　者：李娟[1] 张云丽[1] 王越[1] 徐厚君 LI Juan;ZHANG Yunli;WANG Yue(Beijing Jingmei Group General Hospital,Beijing 102300,China)

机构地区：[1]北京京煤集团总医院消化科,北京102300 [2]华北理工大学公共卫生学院

出　　处：《华北理工大学学报（医学版）》2021年第6期450-455,共6页Journal of North China University of Science and Technology：Health Sciences Edition

摘　　要：目的回顾性研究慢加急性肝衰竭(Acute-on-chronic liver failure,ACLF)患者临床资料,筛选ACLF相关的预后影响因素。通过判别分析法建立ACLF判别分析模型,对未知ACLF样本治疗效果进行判定。方法收集北京佑安医院人工肝中心住院的149例ACLF患者的病例资料进行回顾性研究。根据临床结局,分为76例好转组和73例无效组。通过回顾性研究,借助单因素分析、Logistic回归分析筛选出有意义指标,再进行判别分析,建立判别分析模型,探讨判别函数对ACLF患者的预后判别价值。结果对ACLF患者两组指标经单因素分析得到年龄、总胆红素(TBIL)、白蛋白(ALB)、胆碱酯酶(CHE)、总胆固醇(CHO)、尿素氮(BUN)、肌酐(CR)、凝血酶原时间(PT)、降钙素原(PCT)差异具有统计学意义(P<0.1);上述有统计学意义的指标统一赋值后行Logistic回归分析,最终得到TBIL、PT、PCT、ALB、CHO差异有统计学意义(P<0.05)。TBIL、PT、PCT的OR值>1,因此,TBIL、PT、PCT是ACLF的危险因素。ALB、CHO的OR值<1,因此,CHO和ALB是ACLF的保护因素。通过对Logistic分析有统计学意义的指标进行判别分析得到TBIL、ALB、CHO、PT、PCT建立的Fisher线性判别函数为:Z 1(好转)=-33.231+0.008×TBIL+1.437×ALB+1.631×CHO+0.457×PT+0.557×PCT;Z 1(无效)=-33.250+0.016×TBIL+1.296×ALB+1.139×CHO+0.542×PT+0.750×PCT。对Logistic回归分析有意义指标进行判别分析的试验结果采用回顾法检验其判别效果,判别检验的误判率是23.5%,正确率是76.5%。结论在本次实验中所建立的ACLF逐步判别分析函数,在评估ACLF患者预后状况方面具有较好的判别价值。Objective To conduct a retrospective study of patients with acute-on-chronic liver failure(ACLF)to screen for ACLF related prognostic factors.To establish a discriminant analysis model of the patient's treatment effect and to determine the treatment effect of the unknown ACLF sample by using the discriminant analysis method.Methods A total of 149 cases of ACLF patients were collected that hospitalized in the artificial liver center of Beijing Youan Hospital.According to the clinical outcome,they were divided into 76 cases of improvement group and 73 cases of invalid group.Through retrospective studied,single-factor analysis and Logistic regression analysis were used to screen out meaningful indicators,and then discriminant analysis was conducted to establish a discriminant analysis model to explore the prognostic value of discriminant functions for ACLF patients.Results Univariate analysis of the indicators of two groups of ACLF patients showed that age,TBIL,ALB,CHE,CHO,BUN,Cr,PT and PCT were P<0.1,and the differences were statistically significant.The above indicators with statistical significance were assigned uniformly.Logistic regression analysis was performed after the assigned values,and finally TBil,PT,PCT,ALB and CHO were obtained with P<0.05,the differences were significant.The OR value of TBIL,PT and PCT was>1.Therefore,TBIL,PT and PCT were risk factors for ACLF.The OR value of ALB and CHO was<1,so CHO and ALB were the protective factors of ACLF.The Fisher linear discriminant function of TBil,ALB,CHO,PT and PCT obtained by the discriminant analysis of indicators with statistical significance in Logistic analysis was as follows:Z 1(improvement)=-33.231+0.008×TBIL+1.437×ALB+1.631×CHO+0.457×PT+0.557×PCT;Z 1(invalid)=-33.250+0.016×TBIL+1.296×ALB+1.139×CHO+0.542×PT+0.750×PCT.The retrospective method was used to test the discriminant effect of the function.In this experiment,the retrospective method was used to test the discriminant effect of the discriminant function.The misjudgment rate of the dis

关 键 词：判别分析 慢加急性肝衰竭 治疗效果 

分 类 号：R575.3[医药卫生—消化系统]