BP神经网络和决策树分析在重症手足口病临床早期预警指标中的应用
|
摘要
目的探讨BP神经网络(back propagation neural network,BPNN)模型和穷举卡方自动交互诊断器(Exhaustive CHI-squared Automatic Interaction Detector,Exhaustive CHAID)算法模型在重症手足口病(hand-foot-mouth disease,HFMD)临床早期预警指标分析中的应用价值。方法收集2017年4-7月河南郑州某医院收治的469例HFMD患儿流行病学资料,采用SPSS Modeler 18.0软件进行单因素logistic回归分析筛选出有统计学意义的变量,构建BPNN模型、Exhaustive CHAID算法模型与logistic回归模型,比较三种方法的差异。结果在训练样本和测试样本中,BPNN模型预测正确率(95.28%,93.33%)、灵敏度(0.971,0.905)、约登指数(0.907,0.863)与受试者工作特征曲线下面积(receiver operating characteristic,ROC)(0.992、0.967)均高于Exhaustive CHAID算法模型和logistic回归模型。重症HFMD临床早期预警指标重要性依次为:易惊(0.18)、手足抖动(0.13)、呕吐(0.12)、精神差(0.07)、心率≥140次/min(0.05)、呼吸≥30次/min(0.05)、中性粒细胞比率升高(0.05)、血糖升高(0.05)、四肢发凉(0.04)、热峰≥39℃(0.04)。结论 BPNN模型预测能力优于Exhaustive CHAID算法模型和logistic回归模型。
引文
[1] Zhang D,Li R,Zhang W,et al.A Case-control Study on Risk factors for severe hand,foot and mouth disease.Sci Rep,2017,13(7):1-7.
[2] Owatanapanich S,Wutthanarunqsan R,Jaksupa W,et al.Risk factors for severe hand,foot and mouth disease.Southeast Asian J Trop Med Public Health,2015,46(3):449-459.
[3] 马晓梅,刘颖,杨梦利,等.手足口病月发病率ARIMA乘积季节模型预测探讨.现代预防医学,2017,44(9):1541-1544.
[4] Li H,Luo M,Zheng J,et al.An artificial neural network prediction model of congenital heart disease based on risk factors:a hospital-based case-control study.Medicine(Baltimore),2017,96(6):1-7.
[5] Thishya K,Vattam KK,Naushad SM,et al.Artificial neural network model for predicting the bioavailability of tacrolimus in patients with renal transplantation.PLoS One,2018,13(4):1-7.
[6] Chen C,Yan Y,He Q,et al.Risk factors for delayed breastfeeding initiation based on decision tree and logistic regression model.J Cent South Univ(Med Sci),2018,43(3):306-312.
[7] 朱亚楠,侯玉梅,朱立春.决策树模型在2型糖尿病患病风险预测中的应用.中国卫生统计,2016,33(6):976-982.
[8] Sui M,Huang X,Li Y,et al.Application and Comparison of Laboratory Parameters for Forecasting Severe Hand-Foot-Mouth Disease Using Logistic Regression,Discriminant Analysis and Decision Tree.Clin Lab,2016,62(6):1023-1031.
[9] 马晓梅,隋美丽,段广才,等.手足口病重症化危险因素BP神经网络模型预测分析.中国公共卫生,2014,30(6):758-761.
[10] 宋健,吴学森,张杰,等.三种统计学模型在糖尿病个体患病风险预测中的应用.中国卫生统计,2017,34(2):312-314.
[11] 薛薇,陈欢歌.SPSS Modeler 数据挖掘方法及应用.第2版.北京:电子工业出版社,2016:105-175.
[12] 冯慧芬,赵秋民,朱光.重症手足口病并发神经源性肺水肿危险因素的Meta分析.现代预防医学,2015,42(21):3975-3978.
[2] Owatanapanich S,Wutthanarunqsan R,Jaksupa W,et al.Risk factors for severe hand,foot and mouth disease.Southeast Asian J Trop Med Public Health,2015,46(3):449-459.
[3] 马晓梅,刘颖,杨梦利,等.手足口病月发病率ARIMA乘积季节模型预测探讨.现代预防医学,2017,44(9):1541-1544.
[4] Li H,Luo M,Zheng J,et al.An artificial neural network prediction model of congenital heart disease based on risk factors:a hospital-based case-control study.Medicine(Baltimore),2017,96(6):1-7.
[5] Thishya K,Vattam KK,Naushad SM,et al.Artificial neural network model for predicting the bioavailability of tacrolimus in patients with renal transplantation.PLoS One,2018,13(4):1-7.
[6] Chen C,Yan Y,He Q,et al.Risk factors for delayed breastfeeding initiation based on decision tree and logistic regression model.J Cent South Univ(Med Sci),2018,43(3):306-312.
[7] 朱亚楠,侯玉梅,朱立春.决策树模型在2型糖尿病患病风险预测中的应用.中国卫生统计,2016,33(6):976-982.
[8] Sui M,Huang X,Li Y,et al.Application and Comparison of Laboratory Parameters for Forecasting Severe Hand-Foot-Mouth Disease Using Logistic Regression,Discriminant Analysis and Decision Tree.Clin Lab,2016,62(6):1023-1031.
[9] 马晓梅,隋美丽,段广才,等.手足口病重症化危险因素BP神经网络模型预测分析.中国公共卫生,2014,30(6):758-761.
[10] 宋健,吴学森,张杰,等.三种统计学模型在糖尿病个体患病风险预测中的应用.中国卫生统计,2017,34(2):312-314.
[11] 薛薇,陈欢歌.SPSS Modeler 数据挖掘方法及应用.第2版.北京:电子工业出版社,2016:105-175.
[12] 冯慧芬,赵秋民,朱光.重症手足口病并发神经源性肺水肿危险因素的Meta分析.现代预防医学,2015,42(21):3975-3978.