Using Mid-Infrared Spectroscopic Fingerprinting, Multivariate Analysis And Machine Learning To Differentiate Traditional Herbal Medicine
Traditional herbal medicine is an important part of the global health system with persistent issues like adulteration and misidentification. While current global standards employ chromatographic identification to combat this issue, there are some disadvantages with such methods. This study looked at...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2022
|
| Subjects: | |
| Online Access: | http://eprints.usm.my/60131/1/24%20Pages%20from%20YEAP%20ZHAO%20QIN.pdf http://eprints.usm.my/60131/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Traditional herbal medicine is an important part of the global health system with persistent issues like adulteration and misidentification. While current global standards employ chromatographic identification to combat this issue, there are some disadvantages with such methods. This study looked at five traditional herbal medicine, namely Anoectochilus roxburghii, Aristolochia manshuriensis, Dioscorea hamiltonii, Gelsemium elegans and Alisma orientalis, and ways to classify them. Infrared spectra of the herbs were collected from a total of 200 samples from 20 °C to 120 °C at 10 °C intervals. Infrared signals of functional groups on the main chemical constituents for every herb were present in the infrared spectra collected. Principal component analysis of these spectra found a limitation where the success of the analysis might require less types of herbs included. Computational combination of thermally perturbed spectra was performed to obtain two-dimensional chemical fingerprints for every sample. Machine Learning Classifier were trained to generate models. The model classified the herbs with an accuracy of 87.9 % when all the herbs were included in training. When Alisma orientalis was excluded from training to measure the robustness of the model, 91.3 % of the samples were classified correctly. |
|---|