论文部分内容阅读
本文旨在研究儿童青少年肺通气功能预测的后向传播神经网络(backpropagation neural network,BPNN)方法,以期得到更准确的肺通气功能预计值。样本数据包括内蒙古自治区10~18岁汉族健康儿童青少年999人(男性500人,女性499人),测量身高和体重,使用肺功能仪检测肺通气功能。利用BPNN和多元逐步回归,对用力肺活量(forced vital capacity,FVC)、用力呼气一秒量(forced expiratory volume in one second,FEV1)、最大呼气流量(peak expiratory flow,PEF)、用力呼出25%肺活量时呼气流量(forced expiratory flow at25%of forced vital capacity,FEF25%)、用力呼出50%肺活量时呼气流量(forced expiratoryflow at50%of forced vital capacity,FEF50%)、最大呼气中段流量(maximal mid-expiratory flow,MMEF)、用力呼出75%肺活量时呼气流量(forced expiratory flow at75%of forced vital capacity,FEF75%),分性别建立BPNN预测模型和预计方程式,并利用均方差异(mean squared difference,MSD)和相关系数(R)评价BPNN、基于本工作所建立的线性回归方程(LR方程)、香港Ip等报道的Ip方程和国外较常用的Zapletal方程的预测准确程度。结果显示:无论性别,由BPNN所得各指标的预计值与实测值的MSD均小于其它各个预计方程式,且其预计值与实测值的R均大于其它各个预计方程式;由LR方程所得各个指标的预计值与实测值的MSD均小于Ip方程和Zapletal方程,且其R均大于Ip方程和Zapletal方程。FEF50%、MMEF、FEF75%等3个指标的变异系数(coefficient of variance,CV)均大于其它肺通气功能指标,而这3个指标由BPNN所得预计值和实测值的R较LR方程所得R的增幅ΔR(%)也相应大于其它指标。综上所述,进行肺通气功能预测的BPNN方法要优于传统的多元线性回归方法。肺通气指标的CV越大时,BPNN较传统回归方法的预测优势也越明显。
The purpose of this paper is to study the BP neural network (BPNN) method for predicting pulmonary ventilatory function in children and adolescents with a view to obtaining more accurate prediction of pulmonary function. The sample data included 999 healthy Han children (500 males and 499 females) from 10 to 18 years of age in the Inner Mongolia Autonomous Region. Height and weight were measured. Pulmonary function tests were performed on lung function. Using forced vital capacity (FVC), forced expiratory volume in one second (FEV1), peak expiratory flow (PEF), exhaled by using BPNN and multivariate stepwise regression Forced expiratory flow at 50% of forced vital capacity (FEF 50%), forced expiratory flow at 25% of forced vital capacity (FEF 25%), exhaled 50% of forced vital capacity forced expiratory flow at75% of forced vital capacity (FEF75%) was used to establish BPNN prediction model and prediction equation by gender, and the mean square difference Squared difference (MSD) and correlation coefficient (R) were used to evaluate BPNN. Based on the linear regression equation (LR equation) established by this work, the Ip equation reported by Hong Kong Ip and the prediction accuracy of the more commonly used Zapletal equations abroad. The results show that regardless of gender, the MSDs of the predicted and measured values of each index obtained by BPNN are less than those of other expected equations, and the predicted and measured values of R are greater than those of other expected equations. The MSDs of both the value and the measured value are smaller than those of the Ip equation and the Zapletal equation, and their Rs are larger than the Ip equation and the Zapletal equation. The coefficient of variance (CV) of three indicators such as FEF50%, MMEF and FEF75% were higher than those of other lung ventilation indexes. The predicted and measured R values of BPNN were higher than those of R The increase in ΔR (%) is also correspondingly greater than other indicators. In conclusion, the BPNN method for predicting pulmonary ventilatory function is superior to the traditional multiple linear regression method. The greater the CV of the index of lung ventilation, the more obvious the prediction advantage of BPNN over the traditional regression method.