1. Research on Using K-Means Clustering to Explore High-Risk Products with Ethylene Oxide Residues and Their Manufacturers in Taiwan
- Author
-
Li-Ya Wu, Fang-Ming Liu, Wen-Chou Lin, Jing-Ting Qiu, Hsu-Yang Lin, and King-Fu Lin
- Subjects
K-means clustering ,food safety ,ethylene oxide ,unsupervised learning ,Chemical technology ,TP1-1185 - Abstract
Considering the frequency of ethylene oxide (EtO) residues found in food, the health effects of EtO have become a concern. Between 2022 and 2023, 489 products were inspected using the purposive sampling method in Taiwan, and nine unqualified products were found to have been imported; subsequently, border control measures were enhanced. To ensure the safety of all imported foods, the current study used the K-means clustering method for identifying EtO residues in food. Data on finished products and raw materials with EtO residues from international public opinion bulletins were collected for analysis. After matching them with the Taiwan Food Cloud, 90 high-risk food items with EtO residues and 1388 manufacturers were screened. The Taiwan Food and Drug Administration set up border controls and grouped the manufacturers using K-means clustering in the unsupervised learning algorithm. For this study, 37 manufacturers with priority inspections and 52 high-risk finished products and raw materials with residual EtO were selected for inspection. While EtO was not detected, the study concluded the following: 1. Using international food safety alerts to strengthen border control can effectively ensure domestic food safety; 2. K-means clustering can validate the risk-based purposive sampling results to ensure food safety and reduce costs.
- Published
- 2024
- Full Text
- View/download PDF