【摘要】 目的探讨使用BP神经网络技术用于评估肾小球滤过率的意义。方法根据神经网络原理,进行BP网络建模,并进行训练,直到均方误差(MSE)<10-8。选择我科慢性肾病患者1 330例,分别使用MDRD简化方程和训练好的BP网络对肾小球滤过率进行计算,结果使用方差分析和线性回归分析,比较两种方法的优缺点。结果两种方法方差分析P=0.590,均值比较差异无统计学意义。线性相关系数0.999(P<0.001),两种方法计算结果高度相关。结论BP神经网络可以用于评估肾小球滤过率,为探索更优的评估方程提供了一种不同于回归方程的方法。更多还原

【Abstract】 Objective To study the significance of MDRD equation from BP neural network in predicting renal glomerular filtration rate(GFR).Methods On the basis of the theory of neural network,we built a BP network,and trained it until the MSE was smaller then 10-8.1 330 patients of chronic kidney disease were selected.We calculated the values of GFR by the two methods: MDRD equation,BP network.The results were compared by analysis of variance and linear regression.Results The mean of MDRD equation was 59.412 ml/min.And the mean of BP network was 58.787 ml/min.It was no statistically significant between the two methods.Conclusion BP network can estimate renal glomerular filtration rate.更多还原