摘要
建立川崎病并发冠状动脉病变预测模型。从电子病历数据库中收集1 000例(343例患冠状动脉病变)川崎病患儿的人口学资料、实验室检验数据、超声心动图数据,对数据进行预处理后用关联规则筛选川崎病并发冠状动脉病变的危险指标,划分训练集和测试集分别为总样本集的70%和30%,分别建立神经网络模型和Logistic回归模型,并用灵敏度及特异性等指标对模型的预测效果予以评估。结果显示,神经网络模型的灵敏度=0.718,特异性=0.746,准确率=0.737及AUC(ROC曲线下面积)=0.796,优于Logistic回归模型[灵敏度=0.175,特异性=0.893,准确率为0.647及AUC=0.624]。研究结果表明,神经网络模型对川崎病并发冠状动脉病变的预测效果优于Logistic回归模型。
The objective of the study is to find out the risk factors for coronary artery lesions( CAL) in Kawasaki disease( KD) and build the predictive model. The electronic medical record( EMR) data of 1000 KD patients( 343 KD with CAL) was collected including the demographic data, laboratory test data,echocardiography and diagnosis data,which were pre-processed for analysis. The risk factors for CAL in KD were selected using association rules. The data set was divided into training set( 70%) and testing set( 30%),and the neural network( NN) model and logistic regression( LR) model were built. The predictive performance of the two models was evaluated. Results showed that the sensitivity,specificity,accuracy and AUC( Area Under the ROC Curve) of NN model was 0. 718,0. 746,0. 737 and 0. 796 respectively,which was better than those obtained from LR model[0. 175,0. 893,0. 647 and 0. 624 respectively]. Thus,the performance of NN model to predict CAL in KD is better than that of LR model.
引文
[1]Newburger JW,Takahashi M,Gerber MA,et al.Diagnosis,treatment,and long-term management of Kawasaki disease:A statement for health professionals from the committee on rheumatic fever,endocarditis,and Kawasaki Disease,council on cardiovascular disease in the young,American Heart Association.[J].Pediatrics,2004,110(6):1708-1733.
[2]王宏伟,程佩萱.小儿风湿病诊治进展[J].中国实用儿科杂志,2002,17(5):266-268.
[3]Kim T,Choi W,Woo C,et al.Predictive risk factors for coronary artery abnormalities in Kawasaki disease[J].European Journal of Pediatrics,2007,166(5):421-425.
[4]Maric LS,Knezovic I,Papic N,et al.Risk factors for coronary artery abnormalities in children with Kawasaki disease:A 10-year experience.[J].Rheumatology International,2015,35(6):1053-1058.
[5]段泓宇,华益民,王晓琴.川崎病患儿并发冠状动脉损害的高危因素分析[J].临床儿科杂志,2010,28(7):640-643
[6]Tsumoto S,Hirano S.Risk mining in medicine:application of data mining to medical risk management[J].Fundamenta Informaticae,2010,98(1):107-121.
[7]刘尚辉,王露,郑德禄.Graves眼病相关因素的关联分析[J].中国医科大学学报,2011,40(5):472-474.
[8]陈若珠,杨紫娟,韦哲.基于BP神经网络的骨质疏松疾病的诊断分类研究[J].医疗卫生装备,2011,32(8):9-11.
[9]刘莉,徐玉生,马志新.数据挖掘中数据预处理技术综述[J].甘肃科学学报,2003,15(1):117-119.
[10]Ohri A.Data mining using R[J].R for Business Analytics,2012,170(7):193-223.
[11]Agrawal R,Imieliński T,Swami A.Mining association rules between sets of items in large databases[C]//ACM SIGMOD International Conference on Management of Data.ACM,1999:207-216.
[12]朱大奇,史慧.人工神经网络原理及应用[M].北京:科学出版社,2006:41-42.
[13]Melo F.Area under the ROC Curve[M].New York:Springer,2013:37-38.
[14]Kim TY,Choi WS,Woo CW,et al.Predictive risk factors for coronary artery abnormalities in Kawasaki disease[J].European Journal of Pediatrics,2007,166(5):421-425.
[15]Honkanen VEA,Mccrindle BW,Laxer RM,et al.Clinical relevance of the risk factors for coronary artery inflammation in Kawasaki disease[J].Pediatric Cardiology,2003,24(2):122-126.
[16]Azhar AS,Alattas A.Risk factors for coronary artery lesions in Kawasaki disease.[J].Medicinski Glasnik,2013,10(2):254-256.
[17]Takahashi K,Oharaseki T,Naoe S,et al.Neutrophilic involvement in the damage to coronary arteries in acute stage of Kawasaki disease[J].Pediatrics International,2005,47(3):305-310.
[18]郭奕瑞,李玉倩,王高帅,等.人工神经网络模型在2型糖尿病患病风险预测中的应用[J].郑州大学学报(医学版),2014,57(2):180-183.
[19]周水红,聂绍发,王重建,等.应用人工神经网络预测个体患原发性高血压病危险度[J].中华流行病学杂志,2008,29(6):614-617.
[20]徐新,谭红专,周书进,等.BP人工神经网络在早产预测模型中的应用[J].中华流行病学杂志,2014,35(9):1028-1031.