摘要
目的依据BP神经网络技术建立环孢素A用量的预测模型。方法收集服用环孢素A的肾移植病人16例,共65组样本。其中51组样本作为训练样本,14组样本作为测试样本,建立两层BP神经网络预测模型。结果BP神经网络的仿真预测结果正确率为97.1%。结论BP神经网络模型在环孢素A用量上能够取得良好的预测效果,具有一定的研究性。
Objective To build the prediction model of cyclosporine dosage based on BP neural networks. Method The kidney transplant data of 14 patients was collected,total 65 group data. group data of 51 were selected randomly to form training samples set,on which two-layer BP prediction models were built. About 14 group data were test set to prove the efficiency of prediction model. Result The preciseness of the model is 97. 1% . Conclusion BP model shows high performance in the prediction of cyclosporine dosage. It is worthy of further study.
出处
《北京生物医学工程》
2008年第4期380-381,388,共3页
Beijing Biomedical Engineering
基金
江苏省高校自然科学基础研究项目(07KJB310075)
南京医科大学校基金(06NMUM007)
江苏省大学生科技创新(KY101J27024)资助
