Logratio变换与偏最小二乘法在多目标混料均匀设计药物处方配比优化中的应用  

The Application of Logratio Transform and Partial Least Squares in the Multi-objective Mixture Uniform Design of Drug Prescription Ratio Optimization

作　　者：王祥[1] 冯瑞梅[1,2] 魏珍[1] 王华芳[1] 范瑾[1] 武鹏[1] 张洁[1] 仇丽霞[1] 

机构地区：[1]山西医科大学公共卫生学院卫生统计学教研室,030001 [2]北京协和医院,中国医学科学院肿瘤医院

出　　处：《中国卫生统计》2016年第4期571-574,共4页Chinese Journal of Health Statistics

基　　金：山西省自然科学基金项目(2013011059)

摘　　要：目的研究logratio变换与偏最小二乘法在多目标混料均匀设计药物处方配比优化中的应用。方法对混料数据进行logratio变换之后,用偏最小二乘法进行建模,再用小生境遗传算法进行多目标寻优。结果小生境遗传算法所得相对最优方案:保湿剂、含湿量、胶黏剂、赋型剂、温度和时间分别为36.61%、5.14%、52.51%、5.74%、69℃和8h,与原文献结果相比,评价指标为初粘力15.50cm、持粘力14.52s、载药量6.49mg、剥离度45.58s,分别增加了1.84cm、3.38s、1.25mg、6.48s。结论将logratio变换和偏最小二乘法结合应用于混料均匀设计所得试验数据的模型构建中,并采用多目标遗传算法获得最佳配比和最佳工艺条件,理论是可行且合理的。Objective To research the application of logratio transform and partial least squares in the multi-objective mixture uniform design of drug prescription ratio optimization. Methods After the logratio transformation of mixture data,using partial least squares modeling, then using niched genetic algorithm to optimize the multi-objective. Results The relative optimal solution were obtained by using niched pareto genetic algorithm： moisturizing factor, moisture content, adhesive, excipient, tem- perature and time were 36. 61% ,5.14% ,52. 51% ,5.14% ,69~C and 8h, comparing the results with the original literature, the e- valuation index were 15.50cm initial adhesion, 14. 52s durable adhesion, 6.49mg drug loading and 45.58s stripped degree, respectively increased by 1.84cm ,3.38s, 1.25mg, and 6.48s. Conclusion Using the combination of logratio transform and partial least squares on experimental data model building of mixing uniform design, and using multi-objective genetic algorithm to obtain the optimum ratio and optimum craft condition, it was feasible and reasonable.

关 键 词：混料均匀设计 logratio变换 偏最小二乘法 小生境遗传算法 

分 类 号：R91[医药卫生—药学]