摘要
目的通过探索中药升降浮沉药性与拉曼图谱相关关系,建立中药升降浮沉辨识模型。方法以功效作为升降浮沉药性判定标准,选取升浮药101种、沉降药138种,经处理后分别进行拉曼光谱检测。基于特征筛选后的中药拉曼图谱数据结合支持向量机(SVM)、随机森林(RF)、极度梯度提升(XGBoost)和自适应提升算法(AdaBoost)4种分类器建立中药升降浮沉辨识模型。选用曲线下面积(AUC)、准确率、精确度、召回率及F1值作为模型的评价指标。结果4种辨识模型的AUC均达80%以上,其中XGBoost模型表现最优,AUC达90%以上,准确率、精确度和召回率均优于其他模型,可以实现对升浮中药和沉降中药的准确辨识。结论基于特征筛选后的中药拉曼图谱数据结合XGBoost分类器建立中药升降浮沉辨识模型,可以实现对升浮中药和沉降中药的准确辨识。
Objective To establish an identification model of the ascending,descending,floating and sinking of Chinese materia medica by exploring the correlation between the properties of the ascending,descending,floating and sinking of Chinese materia medica and Raman spectrum.Methods The efficacy was used as the criterion to determine the medicinal properties of ascending,descending,floating and sinking.Totally 101 kinds of rising and floating Chinese materia medica and 138 kinds of descending and sinking Chinese materia medica were selected.After processing,Raman spectroscopy detection was conducted separately.Based on the feature filtered Raman spectrum data of Chinese materia medica,four classifiers,namely support vector machine(SVM),random forest(RF),extreme gradient lifting(XGBoost)and adaptive lifting algorithm(AdaBoost),were used to establish the identification model of the ascending,descending,floating and sinking of Chinese materia medica.The area under the curve(AUC),accuracy,precision,recall,and F1 value were selected as the evaluation indicators of the model.Results The AUC of the four identification models all reached more than 80%,among which XGBoost model showed the best performance.AUC reached more than 90%,and the accuracy,precision and recall were higher than other models,which could realize the accurate identification of rising and floating Chinese materia medica and descending and sinking Chinese materia medica.Conclusion The ascending,descending,floating and sinking identification model of rising and floating Chinese materia medica and descending and sinking Chinese materia medica based on the Raman spectrum data of Chinese materia medica after feature selection,combined with XGBoost clssifier can realize the accurate identification of rising and floating Chinese materia medica and descending and sinking Chinese materia medica.
作者
刘淑明
梁浩
程虹
纪徐维晟
王耘
LIU Shuming;LIANG Hao;CHENG Hong;JI Xuweisheng;WANG Yun(Research Center of TCM-Information Engineering,School of Chinese Materia Medica,Beijing University of Chinese Medicine,Beijing 102488,China;School of Life Sciences,Beijing University of Chinese Medicine,Beijing 102488,China)
出处
《中国中医药信息杂志》
CAS
CSCD
2023年第10期134-138,共5页
Chinese Journal of Information on Traditional Chinese Medicine
基金
国家自然科学基金(81973495)。
关键词
中药
药性
升降浮沉
拉曼光谱
辨识模型
Chinese materia medica
medicinal properties
ascending,descending,floating and sinking
Raman spectrum
identification model
