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+1.350 x1 (age) - 6.329x3 (corpse)+27.073 x6 (swell)+11.094x6 (arthralgia) - 42. 288x7 (medication)+15.821x8 (time) + 5.444x9 ( blood pressure) - 0.978 x10( temperature) + 14.824x15 (kalium) + 14.038x16 (BUN) +41.856X17 (Cr) +13.105 x18 (urine) ; for acute tubular necrosis : y2=- 111.755+3.678x1(age) +0.267 x8(corpse) +4.845 x6(swell)+0.106 x6 (arthralgia) -37.255 X7 (medication)+8.382x8 (time) +5.827x9 ( blood pressure) - 2.l57x10 (temperature) +8.l77x15 (kalium) + 13.757x16 (BUN) +38.798 x17(Cr)+9.600 x18 (urine) ; for nephrotoxicity of calcineurin inhibitor : ys =-120.512+2.690 nx1(age) - 2.245 x8 (corpse) +18.833 x5 (swell) -1.353 x6(arthralgia) +41.266 x7 ( medication) +9.192 x8(time) + 1.341x9 (blood pressure) +0.724x10 (temperature) +5.739 x15 (kalium)+8.995 x16(BUN)+24.532x17(Cr)+20.531 x18 (urine) . The accuracy rate of Bayes equation for acute rejection was 90％, for acute tubular necrosis was lOO％, and for nephrotoxicity of calcineurin inhibitor was 90. 9％ . The total accuracy rate was 93.6％.Conclusion Bayes discriminatory analysis may be useful in the diagnosis of the 3 complications mentioned above which commonly occur in the early-stage after kidney transplantation.%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)-0.978x,10(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y2=-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.
展开▼
