摘要
目的:探讨人工神经网络技术用于肾移植术后患者环孢素A(cyclosporine A,CYA)血药浓度预测的可行性。方法:收集60例肾移植术后患者CYA血药浓度监测结果及监测当日身高、体质量、肝肾功能等13项相关指标,采用EasyNN-plus8.0f软件用于人工神经网络的训练和模型的建立,计算方法为反向传播算法(Back-Propagation)。用建立好的人工神经网络预测肾移植术后患者CYA血药浓度,同时采用多元线性回归法进行预测,并对两者的预测结果进行比较。结果:人工神经网络技术预测的血药浓度和观察值之间的相关系数为0.931 4,多元线性回归的相关系数为0.704 5,人工神经网络技术优于多元线性回归。结论:相对于传统的统计方法,人工神经网络技术在肾移植术后患者CYA血药浓度预测方面显示出良好的效能,但该方法也有待完善。
OBJECTIVE To investigate the feasibility using an artificial neural network （ANN） simulator to predict plasma level of CYA in patients after kidney transplantation. METHODS A data set of 13 physiological measurements for 60 patients was used to develop the model. The ANN structure consisted of 3 layers： an input layer comprised of 13 processing elements,a hidden layer comprised of 15 processing elements,and an output layer of one processing elements （CYA plasma level from TDM）. EasyNN-plus 8. 0f software used for the investigate, Predicted values were obtained by ＂leave-one-out＂ experiments by both ANN and multiple linear regression analysis （MLRA）. RESULTS The correlation coefficients between observed and predicted values obtained by ANN prediction using standardized data sets were r = 0. 931 4. Tbe correlation coefficients obtained by MLRA were r = 0. 704 5. These results indicate that ANN shows better performance in prediction of CYA plasma levels from patients＇physiological measurements than MLRA. CONCLUSION Prediction of plasma levels of CYA in patients after Kidney Transplantation by ANN is superior to the standard statistical method. This method needs to be improve.
出处
《中国医院药学杂志》
CAS
CSCD
北大核心
2008年第4期276-278,共3页
Chinese Journal of Hospital Pharmacy
关键词
人工神经网络
环孢素A
血药浓度
预测
artificial neural network
cyclosporine A
plasma level
prediction
