人工神经网络应用于骨架型缓释片处方设计

APPLICATION OF ARTIFICAL NEURAL NETWORKS FOR FORMULATION DESIGNING OF MATRIX TABLETS

建立了一个设计二氢吡啶类钙拮抗剂骨架片的人工神经网络，该系统由输入层、隐含层、输出层组成，输入层有４个节点（药物的溶解度、骨架材料的水化速率和用量、表面活性剂的用量），隐含层有３个节点，输出层有３个节点（第２、４、６ｈ药物的释放量）。以卡波姆为骨架材料研究了该系统在设计硝苯地平与尼莫地平骨架片的可行性。

Application of artificial neural networks for designing matrix tablets of nifedipine and nimodipine was established. The networks is composed with three layers, i.e., input layer(four processing elements: water solubility of drug, hydrating rate of matrix materials and the amount applied, applied amount of surfactant), hidden layer and output layer (three processing elements: released amount of drug at 2,4,6 h). The matrix materials were Carbopol 934P, 971P, and 974P. The networks was used to predict the released amount of the drug. The result suggested that the predicted values simulated the tested values well.