Bayes判别分析对鉴别肾移植术后早期尿量骤减原因的临床价值  

Bayes discriminatory analysis for identifying the causes of abrupt decrease in urine production at the early stage after kidney transplantation

作　　者：赵豫刚[1] 石炳毅[1] 

机构地区：[1]解放军309医院全军器官移植中心,北京100091

出　　处：《解放军医学杂志》2011年第6期569-572,共4页Medical Journal of Chinese People's Liberation Army

基　　金：全军"十一五"医学科研计划科技攻关课题(06G115);国家科技支撑计划课题(2008BAI60B04)

摘　　要：目的探讨Bayes判别分析对尽早发现并鉴别肾移植术后1个月内导致早期尿量骤减或无尿的3种常见症(急性排斥反应、急性肾小管坏死和钙调磷酸酶抑制剂肾毒性损伤)的临床价值。方法从677例肾移植中收集术后1个月内伴上述3种并发症的125例患者资料,采用随机数字表法将其分为方程组100例和检验组25例。对方程组临床指标进行分析,利用Bayes判别分析进行指标筛选,建立诊断模型,得到3种并发症诊断公式。将检验组临床资料带入建立的数字模型进行判别诊断并与临床最终诊断进行对比,验证模型的准确率。结果采用Bayes判别分析法建立肾移植术后1个月内3种并发症的诊断模型:急性排斥反应y1=-162.384+1.350x1(年龄)-6.329x3(尸肾)+27.073x5(肾肿胀)+11.094x6(关节痛)-42.288x7(血药)+15.821x8(发生时间)+5.444x9(血压)-0.978x10(体温)+14.824x15(血钾)+14.038x16(血尿素氮)+41.856x17(血肌酐)+13.105x18(尿量);急性肾小管坏死y2=-111.755+3.678x1(年龄)+0.267x3(尸肾)+4.845x5(肾肿胀)+0.106x6(关节痛)-37.255x7(血药)+8.382x8(发生时间)+5.827x9(血压)-2.157x10(体温)+8.177x15(血钾)+13.757x16(血尿素氮)+38.798x17(血肌酐)+9.600x18(尿量);钙调磷酸酶抑制剂肾毒性损伤y3=-120.512+2.690x1(年龄)-2.245x3(尸肾)+18.833x5(肾肿胀)-1.353x6(关节痛)+41.266x7(血药)+9.192x8(发生时间)+1.341x9(血压)+0.724x10(体温)+5.739x15(血钾)+8.995x16(血尿素氮)+24.532x17(血肌酐)+20.531x18(尿量)。经检验组验证判别函数赋值对急性排斥反应、急性肾小管坏死和钙调磷酸酶抑制剂肾毒性损伤鉴别诊断的正确率分别为90%、100%和90.9%,总正确率为93.6%。结论通过分析肾移植术后1个月内3种常见的导致尿量突然减少的并发症的临床指标,建立了Bayes判别分析模型,有助于这3种并发症的早期鉴别和诊断。Objective To investigate the clinical significance of Bayes discriminatory analysis on identification of the causes（acute rejection,AR;acute tubular necrosis,ATN;and nephrotoxicity of calcineurin inhibitor,NOCIN） of inducing an abrupt decrease in urine production at the early stage after kidney transplantation.Methods Of 677 patients undergone kidney transplantation,125 of them who suffered from one of the 3 complications listed above within one month after transplantation were included,and assigned either to equation group（n=100） or verify group（n=25） with randomized table.The clinical and laboratory data of equation group were retrospectively analyzed.Bayes discriminatory analysis was performed to screen the valuable parameters and construct the diagnostic model.A total of 3 Bayes equations were established for the 3 complications.The clinical data of verify group were entered into the established equations to acquire the differential diagnosis,which were then compared with the final clinical diagnosis to verify the accuracy of the models.Results Three equations were constructed with Bayes discriminatory analysis for diagnosis of 3 common complications,which caused an abrupt decrease of patients＇ urine production,emerged within one month after kidney transplantation.The equations were expressed as following.For acute rejection： y1=-162.384＋1.350x1（age）-6.329x3（corpse）＋27.073x5（swell）＋11.094x6（arthralgia）-42.288x7（medication）＋15.821x8（time）＋5.444x9（blood pressure）-0.978x10（temperature）＋14.824x15（kalium）＋14.038x16（BUN）＋41.856x17（Cr）＋13.105x18（urine）;for acute tubular necrosis： y2=-111.755＋3.678x1（age）＋0.267x3（corpse）＋4.845x5（swell）＋0.106x6（arthralgia）-37.255x7（medication）＋8.382x8（time）＋5.827x9（blood pressure）-2.157x10（temperature）＋8.177x15（kalium）＋13.757x16（BUN）＋38.798x17（Cr）＋9.600x18（urine）;for nephrotoxicity of calcineurin inhibitor： y3=-120.512＋2.690x1（ag

关 键 词：肾移植 移植物排斥 急性肾小管坏死 钙调磷酸酶抑制剂 Bayes判别分析 

分 类 号：R617[医药卫生—外科学]