RS_LMBP神经网络在心脏病诊断中的应用研究.pdfVIP

28 2 2011 2 : 1006 - 9348( 2011) 02 - 0236- 04 RS- LMBP 1 1 2 徐 冬 , 陈彩霞 , 王翰虎 ( 1. , 571158; 2. , 550025) : , , , , , ( RS) LM BP ( RS - LMBPNN ), , , LM BP , , , , RS - LMBP, , , : ; ; ; : TP183 : B Research on HeartDiseaseDiagnosis Basd on RS- LMBP NeuralNetwork 1 1 2 XU Don , CHEN Cai- x ia , WANG H an- hu ( 1. H ainan Norm al University, H aikou H ainan 571158, China; 2. Guizhou University, Guiyan Gu izhou 550025, China) ABSTRACT: H eart diseaseps test indexes have redundancy in form ation, the traditional dia nosticm ethods cannot e- lmi inate the redundant inform ation and have certain subjectivity, leadin to the low dia nostic accuracy. In order to mi prove the correct rate of the clinical d ia nosis of d isease, a heart disease dia nosism odel ( RS- LMBPNN) is pro- posed based on rou h set theory ( RS) and LMBP neural network. F irstly, RS- LMBPNN uses rou h sets theory to reduce finical dia nosis index attribute, smi plified network structure, then the reduced attribu te was trained neural network, and finally them odew as tested. S mi u lation resu lts show that RS- LMBPNN- s trainin speeds up, and has stron learn ability to mi prove the d ia nostic accuracy compared w ith a sin le LM BP neural netw ork model. K YWORDS: Rou h set theory; Neural network; Deart d isease; D ia nosis [ 1] 1 , , ,