盐酸度洛西汀颗粒粒径近红外光谱定量模型的建立与应用  

Establishment and application of near-infrared spectroscopic quantitative model of duloxetine hydrochloride particle size

作　　者：张钰婷 陈鹏 王晓雨 冯玉贞 于治国[1] 陆峰 ZHANG Yu-ting;CHEN Peng;WANG Xiao-yu;FENG Yu-zhen;YU Zhi-guo;LU Feng(School of Pharmacy,Shenyang Pharmaceutical University,Shenyang 110016,China;School of Pharmacy,Naval Medical University,Shanghai 200433,China;SPH Zhongxi Pharmaceutical Co.,Ltd.,Shanghai 201806,China)

机构地区：[1]沈阳药科大学药学院,沈阳110016 [2]中国人民解放军海军军医大学药学院,上海200433 [3]上海上药中西制药有限公司,上海201806

出　　处：《药物分析杂志》2022年第3期411-418,共8页Chinese Journal of Pharmaceutical Analysis

基　　金：国家重点研发计划“国家质量基础的共性技术研究与应用”(2017YFF0210103);上海市产业转型升级发展专项(JJ-YJCX-01-17-2237)。

摘　　要：目的:建立盐酸度洛西汀颗粒粒径的远程、快速近红外光谱定量模型,为盐酸度洛西汀片剂生产提供实时监测系统。方法:采集车间生产的多批次盐酸度洛西汀颗粒样品的近红外光谱,使用激光粒度仪测定颗粒样品参考值。选择使用原光谱、光程恒定的预处理方法,建模波段5550.00~7400.00 cm^(-1),使用偏最小二乘法建立颗粒粒径定量模型。使用建立的模型预测不同批次的独立样品集,并用激光粒度仪测定结果进行验证。结果:所建模型的校正误差均方根为0.0059,相关系数r_(c)为0.9657,预测误差均方根为0.0073,相关系数r_(p)为0.9465,验证集标准偏差与预测标准偏差的比值为3.85。结论:建立的基于近红外光谱技术的盐酸度洛西汀颗粒粒径定量模型可即时、准确预测颗粒粒径,为在线实时监测制剂粒径提供了可行路径。Objective:To establish a remote and fast near-infrared spectrum(NIRS)quantitative model of duloxetine hydrochloride particle size,and to provide a real-time monitoring system for the production of duloxetine hydrochloride tablets.Methods:The NIRS spectra of duloxetine hydrochloride granules samples were collected and the reference values were determined by laser particle size analyzer.The pretreatment method of constant spectrum and no smoothing was selected.The modeling band was 5550.00-7400.00cm^(-1).The model was established by partial least square method.The model was used to predict the independent sample sets of different batches,and the results were verified by laser particle size analyzer.Results:The root mean square error of themodel was 0.0059,the correlation coefficient r;was 0.9657,the root mean square error of prediction was 0.0073,the correlation coefficient r;was 0.9465,and the ratio of prediction to deviation was 3.85.The prediction error of the model for other batch of independent sample set is less than 5%.Conclusion:The quantitative model of duloxetine hydrochloride granules based on NIR spectrum can predict the particle size immediately and accurately,which provides a feasible path for online real-time monitoring of the particle size of preparation and stable uniformity of produced drugs.

关 键 词：盐酸度洛西汀 近红外光谱技术 片剂 粒径 偏最小二乘法 

分 类 号：K917[历史地理—人文地理学]