The application of artificial neural network(ANN) and near-infrared spectroscopy for pharmaceutical nondestructive quantitative analysis of Paracetamol was investigated. The artificial neural network patterns of Parac...The application of artificial neural network(ANN) and near-infrared spectroscopy for pharmaceutical nondestructive quantitative analysis of Paracetamol was investigated. The artificial neural network patterns of Paracetamol tablet medicines, powder medicines, first derivative spectra and second derivative spectra were established, and they were compared each other. The uncertain specimens were predicted. The parameters affecting ANN were discussed. A new network evaluation criterion, i.e., the degree of approximation, was employed, and the predicted results were reliable.展开更多
目的:建立测定对乙酰氨基酚片中有关物质含量的方法。方法:采用高效液相色谱法。色谱柱为Agilent 5HC-C8,流动相A、B分别为甲醇-水-冰醋酸(50∶950∶1,V/V/V)和甲醇-水-冰醋酸(500∶500∶1,V/V/V)(梯度洗脱),流速为0.9 m L/min,检测波长...目的:建立测定对乙酰氨基酚片中有关物质含量的方法。方法:采用高效液相色谱法。色谱柱为Agilent 5HC-C8,流动相A、B分别为甲醇-水-冰醋酸(50∶950∶1,V/V/V)和甲醇-水-冰醋酸(500∶500∶1,V/V/V)(梯度洗脱),流速为0.9 m L/min,检测波长为254 nm,柱温为40℃,进样量为5μL。结果:该色谱条件下,对乙酰氨基酚片中的主药(对乙酰氨基酚)、6个已知杂质(对氨基酚、对氯苯乙酰胺和杂质A、B、D、F)、3个制剂特定辅料(羟苯甲酯、羟苯乙酯和羟苯丙酯)和1个未知杂质的分离度均大于1.5。6个已知杂质检测质量浓度的线性范围分别为0.539~1.617、0.026~0.384、0.237~17.799、0.257~19.271、0.239~17.955、0.246~18.462μg/mL(r≥0.9998),杂质A、B、D、F的校正因子分别为2.9、1.0、1.2、6.2;检测限分别为0.0096、0.0242、0.1640、0.0511、0.0559、0.4220 ng,定量限分别为0.0320、0.0806、0.5460、0.1700、0.1860、1.4060 ng;平均回收率为95.96%~111.09%(RSD为0.05%~2.42%);精密度试验的RSD均小于15%,且耐用性良好。3批样品均检出了对氨基酚(均为0.006%)、杂质B(0.016%~0.017%)、未知杂质(0.0020~0.0021%),未检出对氯苯乙酰胺和杂质A、D、F。结论:该方法专属性强、准确度高,可用于对乙酰氨基酚片有关物质的测定。展开更多
文摘The application of artificial neural network(ANN) and near-infrared spectroscopy for pharmaceutical nondestructive quantitative analysis of Paracetamol was investigated. The artificial neural network patterns of Paracetamol tablet medicines, powder medicines, first derivative spectra and second derivative spectra were established, and they were compared each other. The uncertain specimens were predicted. The parameters affecting ANN were discussed. A new network evaluation criterion, i.e., the degree of approximation, was employed, and the predicted results were reliable.
