应用人工神经网络预测ESWL单次治疗肾结石的成功率  

Prediction of success rate for single ESWL-treated renal calculi by artificial neural network

作　　者：张鹏[1] 吴芃[1] 张辉见[1] 陈壮飞[1] 郑少斌[1] 刘成山[1] 

机构地区：[1]南方医科大学南方医院泌尿外科,广州510515

出　　处：《临床泌尿外科杂志》2012年第11期803-806,共4页Journal of Clinical Urology

基　　金：广东省医学科研基金项目(No:WSTJJ2008111232012319780420951X);南方医院院长基金(No:2009C024)

摘　　要：目的:建立预测ESWL治疗肾结石成功率的人工神经网络,并转化为临床应用。方法:2008年1月～2010年2月接受ESWL治疗的肾结石患者325例,将治疗前的临床资料共10项(性别、尿路刺激症、血尿、肾绞痛、结石位置、结石患侧、年龄、体重指数、病程时间、结石大小)纳入预测参数,建立人工神经网络。绘制预测拟概率的ROC曲线,计算曲线下面积(area under curve,AUC),计算约登指数,结合临床要求,得到适宜的概率分界值。结果:人工神经网络得到预测参数重要性的前5位依次是结石大小、病程时间、血尿、结石位置、体重指数。ANN的AUC为0.856,95%可信区间为0.774～0.917,与AUC为0.5进行显著性检验,P<0.001。确定概率分界值为0.595时,ANN的敏感度和特异度达到较为理想状态,分别为92%和60%。结论:人工神经网络预测ESWL单次治疗肾结石成功率有较好的准确性,选择恰当的概率分界值,可提高人工神经网络的预测效能。Objective：To construct artificial neural network（ANN） for prediction of success rate for single ES- WL-treated renal calculi, and to facilitate to use clinically. Method.. Between January 2008 and February 2010, 325 patients underwent ESWL, ten pretreatment clinical data such as gender, urinary irritation symptoms, hematuria, renal colic, stone location, laterality, age, body mass index（BMI）, disease duration, stone size were included in predictive parameter of ANN, then ANN was constructed. ROC curve of predictive pseudo-probabilities was draw and its AUC was calculated. The Youden index and clinical demand were taken in account for obtaining an adap- tive cutoff value. Result：Stone size, disease duration, hematuria, stone location, BMI were found to be the top 5 important impact parameters. The AUC is 0. 856 （95 %$ confidence interval： 0. 774-0. 917）, Comparing with AUC = 0.5, P〈0. 001. While cutoff value is set to be 0. 595, the sensitivity and the specificity of ANN attained a opti- mal status, which is 92% and 60%, respectively. Conclusion：ANN for prediction of success rate for single ESWL- treated renal calculi has a good predictive ability. Appropriate selection of cutoff value could improve the efficacy of ANN.

关 键 词：肾结石 体外冲击波碎石 人工神经网络 预测 

分 类 号：R692.4[医药卫生—泌尿科学]