logistic回归在疾病多指标联合诊断中的应用

*3100059030410511.2260192.△[e-mail]*秦正积1沈毅1△崔晓莉2肖静1何书1【】logisticＲOC。logisticＲOC。SASlogisticＲOCAlogisticBlogisticPＲOC。。CyCＲOCZ0.57460.77940.76190.79680.77460.74760.76510.75080.7508Z=0.5P0.45140.00040.00160.00050.00120.00710.00510.00570.0037AbeforeBbeforeZ0.74800.8381P＜0.0001。BA。【】logisticＲOCＲOCreceiveroperatingcharacteristiccurve。ＲOC、———。。。logisticＲOC。logisticＲOCYnnxnjY=0j=12…nnnaxaiY=1i=12…namX=X1X2…Xm。X=PY=1|X1X2…Xmlogistic1lnP1－P=β0+β1X1+β2X2+…+βmXm。ＲOClogisticPPkPi≥PkYi=1Pi＜PkYi=0。ＲOC。SASlogisticＲOCＲOC。AKI2、、、AKIAKI。“CCyCC”before、ICU6icu6、1～7onedaytwodaythreedayfourdayfivedaysixdaysevendayCyC。1.logisticＲOCSASCyClogistciＲOC3ＲOC。logisticＲOC0.5。1beforeS=0.5476P=0.4514。ＲOCS0.74～0.8。。2．ＲOC1logisticＲOCAKI。·611·201423111ＲOCszpLCIUCIbefore0.57460.75310.45140.38050.7688icu60.77943.54460.00040.62490.9338oneday0.76193.15240.00160.59910.9247twoday0.79683.50300.00050.63080.9629threeday0.77463.22840.00120.60790.9413fourday0.74762.69060.00710.56720.9280fiveday0.76512.79950.00510.57950.9507sixday0.75082.76580.00570.57310.9285sevenday0.75082.90040.00370.58130.9203*。。。logis-ticＲOCSAS。2beforebefore。2beforebefore。2logisticＲOC2ＲOCszpLCIUCIAbefore0.74808.1600＜0.00010.68880.8082before0.72907.7872＜0.00010.67160.7870Bbefore*0.435551.92673－1.945990.369980.50111before0.838104.46147＜0.00010.689570.98662*。YCyClogisticＲOC2。2beforeBA0.091。be-forelogistic。SASlogisticＲOCＲOC。logisticPXXＲOC4。AlogisticP。BlogisticＲOC。。logistic-54。。logisticiＲOCB。01PＲOPＲOC6。1.．22008205-206．2．MehtaＲLPascualMTSorokoSetal．SpectrumofacuterenalfailureintheintensivecareunitthePICAＲDexperience．KidneyInt20046641613-1621．3．．ＲOC19991604241-244．4．．logisticＲOC24122-24．5．MetzCEHermanBAShenJH．Maximumlikelihoodestimationofre-ceiveroperatingcharacteristicＲOCcurvesfromcontinuouslydistribu-teddata．statistmed1998171033-1053．6．．ＲOC200057-59．·711·ChineseJournalofHealthStatisticsFeb2014Vol．31No．1