Research of machine learning in the application of authenticity discrimination of Fritillariae Thunbergii Bulbus and Fritillariae Hupehensis Bulbus in different form with dry-process REIMS fingerpring
Objective:To study and analyze rapid evaporative ionization mass spectrometry(REIMS)fingerprints of samples of Fritillariae Thunbergii Bulbus and Fritillariae Hupehensis Bulbus in different forms for authenticity discrimination with machine learning.Methods:Aerosol formations from the samples by high temperature of dry burning method were ionized and determined by REIMS with m/z 50-1 200 as scanning range in sensitive mode and positive ion mode.The scanning time was 0.2 s and data was recorded as continuous mode.Then the basic situation of REIMS data distribution was studied and analyzed through the methods of cluster analysis,correlation analysis,similarity analysis and principal component analysis.And then logistic regression model with ridge regression(l2)as penalty parameter and quasi-Newton method(lbfgs)as optimization algorithm was established.Results:The REIMS fingerprints of the samples showed the characteristics of variety differences.Both cross vali-dation and test set validation had an accuracy of 1.0,and the logistic regression model could accurately predict and distinguish the varieties of the samples.Conclusion:The application prospect of REIMS technique combined with machine learning in the field of traditional Chinese medicine is very broad.
Fritillariae Thunbergii BulbusFritillariae Hupehensis BulbusREIMSmachine learningartificial intelligencelogistic regressionanalysis of traditional Chinese medicineauthenticity discrimination
万方数据
维普
