| Project/Area Number |
22KJ0403
|
|---|
| Project/Area Number (Other) |
22J11143 (2022)
|
|---|
| Research Category |
Grant-in-Aid for JSPS Fellows
|
| Allocation Type | Multi-year Fund (2023)
Single-year Grants (2022) |
|---|
| Section | 国内 |
|---|
| Review Section |
Basic Section 41040:Agricultural environmental engineering and agricultural information engineering-related
|
|---|
| Research Institution | University of Tsukuba |
|---|
Principal Investigator |
BUI Thi Bao Chau 筑波大学, 生命環境系, 特別研究員(PD)
|
|---|
| Project Period (FY) |
2023-03-08 – 2024-03-31
|
| Project Status |
Completed (Fiscal Year 2023)
|
| Budget Amount *help |
¥1,700,000 (Direct Cost: ¥1,700,000)
Fiscal Year 2023: ¥800,000 (Direct Cost: ¥800,000)
Fiscal Year 2022: ¥900,000 (Direct Cost: ¥900,000)
|
|---|
| Keywords | fluorescence fingerprint / quality assessment / chemical attributes / machine learning / extraction method / wet milling / phytochemicals |
|---|
| Outline of Research at the Start |
Aiming to establish a workflow to prepare high quality extracts from hard materials like the spices, this research starts with extracting the chosen spices by a new method using electric wet stone-mill, then continues to evaluate the quality of the extracts by using the Excitation and Emission Matrix (EEM) model and confirm their practical usage by the biochemical assays.
|
| Outline of Annual Research Achievements |
The quick quality assessment method incorporating Fluorescence Fingerprints (FFs) and machine learning (ML) was successfully established. FFs of different spice extracts were obtained at multiple dilution levels. All FFs were combined into a meta-data matrix used as input to ML models. Different algorithms were considered and optimized, including Partial Least Squares, Support Vector Machine, Artificial Neural Network, Random Forest, and the ensemble models to predict several chemical attributes of spice extracts such as scavenging and reducing antioxidant ability, total polyphenol content and total flavonoid content. Results showed that different chemical attributes needed different optimized models. While scavenging antioxidant ability was adequately predicted with the simple Least Squares model, other chemical attributes including total polyphenol and flavonoid contents as well as reducing antioxidant ability required the more complex ensemble model. In addition, although the meta-data comprised of a very large number of variables, we showed that the pre-treatment feature selection was not always necessary for successful regression optimization with the use of ensemble model. These results were published in an original paper (Bui et al., 2024) and presented as an invited lecture at the Tsukuba Conference 2023.
|
