神经网络多目标同步优化HPMC缓释片处方

An artificial neural network for multi-objective simultaneous optimization of HPMC sustained release tablet formulations

目的 通过建立BP神经网络模型,预测HPMC缓释片中药物释放;并和优化算法结合实现缓释片处方的多目标同步优化。方法 选取溶解度为难溶到略溶的5种药物(别嘌醇,甲氧苄氨嘧啶,阿昔洛韦,替硝唑,对乙酰氨基酚)作为模型药物,压制HPMC骨架片,并进行体外释放情况考察,考察52个处方中药物的溶解度、含药量、HPMC的量、HPMC的固有黏度、辅料的量、黏合剂的浓度、溶出仪的转速对药物释放情况的影响。将各因素作为神经网络的输入,药物的累计释放量作为输出,对网络进行训练,建立BP神经网络模型,并和优化算法相结合实现以对乙酰氨基酚、甲氧苄氨嘧啶、米诺地尔、氧氟沙星为模型药物,在不同的含药量、不同转速的条件下对处方进行优化。结果 利用神经网络预测药物的释放,训练处方和测试处方的实测值和预测值能很好吻合,4个优化处方的释放值均和目标值很接近。结论 神经网络可用于预测不同药物不同处方组成的HPMC缓释片中药物的释放,并能同步优化HPMC缓释骨架片的处方。

OBJECTIVE: To use the artificial neural network (ANN) to predict drug release from HPMC matrix tablets and optimize the formulations of sustained release tablets. METHODS: 52 formulations of five model drugs(trimethoprim, paracetamol, allopurinol, aciclovir, tinidazole)with different solubility were prepared. Drug solubility, the amount of loaded drug in tablets, the amounts of HPMC, the intrinsic viscosity of HPMC, the amounts of MCC, the concentration of PVP in the binder, and the rotation speed of dissolution machine were selected as independent factors. The accumulation release at 6 sampling times was worked as dependent factors. The series of dependent factors and independent factors were used as tutorial data for ANN to train BP network. The trained network was used to predict drug release. combined with optimization tool, ANNS optimized four sustained release tablets formulations of 4 model drugs(minoxidilum, ofloxacin, trimethoprim and paracetamol). RESULTS: There was very good agreement between the ANN predicted and observed release profiles of both training formulations and testing formulations. The drug release parameters of four optimal formulations were close to the target values. CONCLUSION: ANN is able to precisely predict the release of drug with various solubility in different formulations and different rotation speeds of dissolution machine. ANN is suitable for multi-objective simultaneous optimization of HPMC sustained release tablet formulations.