基于MATLAB的冠心病中医证候BP神经网络实现-孙贵香

29820118CHINESEARCHIVESOFTRADITIONALCHINESEMEDICINEVol．29No．8Aug．20111774MATLABBP121211．4100072．410007。MATLAB。。49690．5%。13291．36%。。BPR259．414A1673－7717201107－1774－03TheRealizationoftheBPNeuralNetworkModelBasedontheMATLABCoronaryHeartDiseaseofTCMSyndromeSUNGui-xiangYUANZhao-kai1．HunanUniversityoftraditionalChinesemedicineChangsha410007HenanChina2．TheFirstHospitalofHunanUniversityofTCMChangsha410007HunanChinaAbstractobjectivetostudytheartificialneuralnetworktechnologyintheapplicationinthefieldofTCMsyn-drome．MethodsBasedontheclinicalepidemiologyinvestigationincoronaryheartdiseaseweconstructedanartificialneuralnetworkmodelofthecoronaryheartdiseaseofTCMsyndromeonthebasisofneuralnetworkMATLABtoolboxthenwetesttheperformanceofthismodelusingaretrospectiveofinspectionandprospectivetestingmethod．ResultsThediagnosticaccuracyofrateis90．5%in496casesofalreadycollectionofretrospectiveexaminationshowedandspecificsyndrometypesofdiscriminationandsamplenumberaccuracywerepositivelycorrelated．Thediagnosticaccuracyofrateis91．36%innewcollectionof132casesoftheprospectiveexaminationshowed．ConclusionTheartificialneuralnet-workcanbettertheforensicsofsyndromebecauseitcanexploretheinternalrulesofTCMsyndrome．AndthereareagoodprospectusingartificialneuralnetworkinthestudyofstandardizationoftraditionalChinesemedicalsyndrome．KeywordsBPneuralnetworkcoronaryheartdiseaseTCMsyndrome2011－03－2207C4762007061976－。ArtificialNeuralNetworksANN1。。MATLABNeuralNetworkToolboxNNTool、2。MATLABBP“”。1BP3－5BP。BPBP。。1．1MATLABBP1。1BPtraingdtraingdmtraingdatrainscgtrainlm1．2BPMATLABBP51～4。1traingd0．025326410000。2traingdm0．02132164771DOI10.13193/j.archtcm.2011.08.72.sungx.01829820118CHINESEARCHIVESOFTRADITIONALCHINESEMEDICINEVol．29No．8Aug．20111775。1BPtraingd2BPtraingdm3traingda3traingda0．00021967710000。4traingscg4traingscg9．9987×10－82655。LMtrainlmPC。。Trainscg。TrainscgTrainscgLM。Trainscg。TrainscgBP。2MATLABBP2．1。9921097、、、、、、。100。2．210。0．01－0．810－71091167y1=1000000y2=0100000y3=0010000y4=0001000y5=0000100y6=0000010y7=0000001。“0”0000000。2．3BP1955。BP。3MATLABBP3．13．1．1496y197、y2109、y38、y453、y567、y662、y7100。3．1．2。。。49644990．5%。。93．0%、92．6%、91．8%、91．0%、88．7%、88．6%37．5%。8。BP。577129820118CHINESEARCHIVESOFTRADITIONALCHINESEMEDICINEVol．29No．8Aug．20111776。2。2y197y2109y38y453y567y662y71008910134761559391．8%92．6%37．5%88．6%91．0%88．7%93．0%3．2BP。3．2．1①、。②《》。③132y125y221、y32、y46、y59、y611、y778151。3．2．23MATLABBP817491．36%。。3y125y221y32y46y59y611y77232015810792．0%95．2%50．0%83．3%88．9%90．9%100%3．2．347。42h。41。。。。5BP。54MatlabBPBP。4．1。322150%、、194、218、200。。。4．2。。、、。。《》6《》7、。。。、。。MATLAB。。。1．J．20072571450－1452．2MinXilinLiuGuohua．DevelopingBPnetworksapplicationwithMatlabToolboxJ．ApplicationResearchofComputers2002191163－164．3HowardDemuthMarkBeale．Neuralnetworktoolboxuser’sguideVersion4M．TheMathworksInc2001194－198．4．MatlabBPJ．．2006235142－144．5．BPMatlabJ．200422245－47．6．S．1997．7M．．1978．6771