Abstract：
The price and safety of finished pharmaceutical preparations are two major concerns while prescribing medicine. In this work, machine learning-based classification models were developed with respect to the quality attributes of 258 samples covering 9 marketed amlodipine (AMLO) formulations. The quantitation of AMLO and its three sulfonate ester genotoxic impurities of besylate counter ion was settled using a validated high-performance liquid chromatography-diode-array detection method. The classification of correlation between dependent and independent variables was exercised using linear discriminant analysis models. The linear dispersion of acceptable quality attributes was significantly different for AMLO besylate formulation with unit price per tablet \"<1 Rs.\" Although the correlations between price and quality are well-understood associations group centroid distance for price group \"2-3 Rs.\" and \"1-2 Rs.\" reveal that acceptable quality dispersion was similar for both groups. Nonetheless, a higher price could allow storage of the finished formulation to be kept on the shelf for a longer period.
摘要：
成品药物制剂的价格和安全性是开药时的两个主要问题。在这项工作中,针对包含9种市售氨氯地平(AMLO)制剂的258个样品的质量属性,建立了基于机器学习的分类模型.使用经过验证的高效液相色谱-二极管阵列检测方法，确定了AMLO及其三种磺酸酯基因毒性杂质的苯磺酸盐抗衡离子的定量。使用线性判别分析模型对因变量和自变量之间的相关性进行分类。可接受的质量属性的线性分散对于AMLO苯磺酸盐制剂与每片单价“<1Rs。“尽管价格和质量之间的相关性是很好理解的关联，但价格组的质心距离”2-3Rs。\"and\"1-2Rs.“揭示两组的可接受质量分散相似。尽管如此，更高的价格可以允许成品制剂的储存在货架上保存更长的时间。
