应用人工神经网络—遗传算法优化多西紫杉醇微球制备工艺参数

Application of Artificial Neuralnetwork and Genetic Algorithm to the Process Parameters Optimization of Docetaxel Chitosan Microspheres

关键词:  多西紫杉醇  壳聚糖微球  人工神经网络  遗传算法

Key words:  docetaxel  chitosan microspheres  artificial neural network  genetic algorithm

摘要:

目的 优化多西紫杉醇壳聚糖微球的制备工艺参数。方法 应用人工神经网络对微球制备工艺参数与考察指标之间的关系进行模型拟合，并结合遗传算法优化微球的制备工艺参数。结果 模型参数优化结果为：壳聚糖浓度3.730 8%、乳化剂用量0.500 4 g、油水体积比1.843 3、药载比25.027 7、交联剂用量2.246 5 mL、搅拌乳化时间63.419 1 min、搅拌速率611.922 8 r·min^-1。考察指标预测结果是：微球的载药量43.653 8%、粒径8.168 5μm、 跨距0.594 0。验证实验数据与网络模型优化结果基本吻合。结论 应用人工神经网络建模结合遗传法寻优，可以实现多西紫杉醇壳聚糖微球制备工艺参数的优化。

Abstract:

OBJECTIVE To optimize process parameters of docetaxel chitosan microspheres. METHODS The preparation was selected by L18(37) orthogonal design, and a mathematical model of relationship between the independent variables and dependent variables was established by using back-propagation(BP) artificial neural networks(ANN) and the process parameters were optimized with genetic algorithm(GA). RESULTS The optimum process parameters GA-predicted was established as follows: 3.730 8% as concentration percentage of chitosan, 0.500 4 g as amount of emulsifier, 1.843 3 as volume percentage ratio of organic phase to water phase, 25.027 7 as drug loading ratio, 2.246 5 mL as volume of glutaral, 63.419 1 min as duration of rotation and 611.922 8 r·min^-1 as rotation speed with the maximum drug loading 43.65 38%, the minimum span dispersity 0.594 0, and 8.168 5 μm as the mean diameter of docetaxel chitosan microspheres. Bias between observed and predicted values of the drug loading, the mean diameter and span dispersity of Docetaxel chitosan microspheres had no significant difference. CONCLUSION The multi-objective simultaneous optimization of process parameters in docetaxel chitosan microspheres preparation could be achieved by combining BP ANN modeling with GA. The models developed in this study were proved to be predictable and feasible.