Background: Authentication methods are necessary to guarantee the integrity of botanical supplements and their ingredients. In 2012, AOAC International published "Guidelines for Validation of Botanical Identification Methods" however these guidelines proved rather cumbersome.
Objective: Develop a simpler method for validatation based on one-class modeling that only considers authentic materials.
Methods: One-class modeling uses chemometric analysis based on soft independent modeling of class analogy and the specific pre-processing steps of sample vector normalization and autoscaling.
Results: Any unknown sample can be judged authentic or adulterated based on its agreement with the profile of the authentic samples. The sensitivity and accuracy of one-class modeling is improved using sample vector normalization and autoscaling. The limit of detection for any variable is statistically predictable.
Conclusion: One-class modeling offers a simple approach to authentication and is applicable to any non-targeted analytical method. Only the characteristics of the authentic samples are necessary and no specification of an adulterant is necessary.
Highlights: One-class modeling offers a simple approach to authentication and is easily implemented using any chemometrics platform.
Published by Oxford University Press on behalf of AOAC INTERNATIONAL 2023.