首页|以药品监管理念为导向的基于近红外光谱与机器学习联用的艾叶真伪判别研究

以药品监管理念为导向的基于近红外光谱与机器学习联用的艾叶真伪判别研究

A Study Guided by Drug Regulatory Philosophy on the Authenticity Discrimination of Artemisiae Argyi Fo-lium Based on the Combination of Near-Infrared Spectroscopy and Machine Learning

扫码查看
原文链接
  • 维普
  • 万方数据
目的 以药品监管理念为导向,建立基于近红外光谱与机器学习联用的艾叶真伪品判别的方法.方法 使用近红外光谱仪测定艾叶真伪样品的近红外光谱,并采用特征工程中的特征筛选、特征衍生等相关技术对实验数据进行处理.随机划分训练集和测试集,使用训练集样品数据对机器学习领域经典的逻辑回归模型进行2分类模式训练,测试集样品进行模型的评估.结果 使用逻辑回归模型对测试集样品的判别准确率为97％,其他各项评价指标也均在92％以上.同时,对于正品与伪品混合样品的判别也较准确.相较于传统化学计量学方法,判别准确率更高.结论 本研究所建立的逻辑回归模型可以实现鉴别艾叶的真伪,对药品监管工作具有技术支撑作用.
OBJECTIVE To establish a method for identifying the authenticity of Artemisiae Argyi Folium suitable for use in drug regulatory work.METHODS The near-infrared spectra of samples of Artemisiae Argyi Folium and counterfeit were determined,and the experimental data was processed using feature engineering related techniques,such as feature screening and feature derivation.The training set and test set were divided randomly,and the logistic regression model,a classic model in the field of machine learning,was trained in 2-class mode and evaluated with the training set data and the test set data used,respectively.RESULTS The discrimina-tion accuracy of the samples in the test set was 97％,and the other evaluation indicators were also above 92％with the logistic regres-sion model.In addition,the results of discrimination between genuine and counterfeit mixed samples were also relatively accurate.Compared with traditional chemometrics methods,the machine learning used in the study had higher discrimination accuracy.CONCLUSION The logistic regression model established in this study can achieve the authenticity identification of Artemisiae Argyi Folium,providing technical support for actual drug regulatory work.

Artemisiae Argyi Foliumnear-infrared spectrummachine learninglogistic regressionfeature engineering

张凯笑、熊婧、郭涛、王晓伟、王海波、李彦超、张文静、石岩

展开 >

河南中医药大学药学院,郑州 450046

河南省药品医疗器械检验院(河南省疫苗批签中心),国家药品监督管理局中药材及饮片质量控制重点实验室,郑州 450008

中国食品药品检定研究院,北京 102629

艾叶 近红外光谱 机器学习 逻辑回归 特征工程

河南省科技厅科技攻关项目资助中国药品监管科学行动计划第二批重点项目资助河南省高层次人才国际化项目资助国家药品监督管理局药品监管科学体系建设重点项目

222102310110NMPAJGKX-2023-0302021-72RS2024Z006

2024

10.11669/cpj.2024.13.009

中国药学杂志
中国药学会

中国药学杂志

CSTPCD北大核心
影响因子：0.957
ISSN：1001-2494
年,卷(期)：2024.59(13)