摘要
针对短效静脉麻醉药物异丙酚在药物代谢过程中存在强时变性、复杂非线性等特点,以及传统的群体药代非线性混合效应法在建模方法中存在工作繁杂、人为因素多等缺陷,本研究利用差分进化算法优化多因素时序灰色模型,建立基于灰色理论的异丙酚药代血药浓度预测模型,并与非线性混合效应建模法(nonlinear mixed effects modeling,NONMEM)预测效果进行比较。结果显示,差分进化-多变量灰色模型(DE-MGM)的预测结果的偏离性(MDPE)为-4.6%,NONMEM为-12.13%;DE-MGM的预测结果的精确度(MDAPE)为13.19%,NONMEM为23.12%。基于差分进化优化多因素灰色模型能稳定预测异丙酚血药浓度,且准确度高。该方法原理简单,实现便捷,可适用于异丙酚等短效静脉麻醉药物的群体药代药效学研究和分析。
Due to the characteristics of propofol of high time-varying, and complex compartment model, the traditional method of nonlinear mixed effects modeling（NONMEM） has miscellaneous of variables and plenty of artificial factors in the estimation of propofol. This study was aimed to build a propofol prediction model based on the differential evolution（DE） algorithm and grey model. DE was used to optimize the parameter of multi-variable grey model（MGM） and to build a model of prediction of the plasma concentration of propofol based on the grey model. It was compared with the results of NONMEM algorithm. In conclusion, the median performance error（MDPE） of DE-MGM was-4.6%, while the result of NONMEM is-12.13%. The median absolute performance error（MDAPE） of GA-BP neural network is 13.19%, while that of NONMEM is 23.12%. The experimental results suggest that the new method is suitable to determine the short half-life of anesthesia drug propofol with higher accuracy.
作者
李龙艳
曹扬
潘克家
LI Long-yan CAO Yang PAN Ke-jia(Department of Anesthesiology Center of Medical Engineering, Xiangya Hospital, Central South University, Changsha 410008, China School of Mathematics and Statistics, Central South University, Changsha 410083, China)
出处
《药学学报》
CAS
CSCD
北大核心
2017年第10期1599-1604,共6页
Acta Pharmaceutica Sinica
基金
国家自然科学青年基金资助项目(81601728)
关键词
差分进化算法
多因素灰色模型
异丙酚
血药浓度
differential evolution algorithm
multi-variable grey model
propofol
plasma concentration
