Related Experiment Videos
A Machine Learning Framework for Meat Safety Risk Assessment with Cross-Domain Data Fusion
Liying Wu1, Yanna Ke1, Yiming Huang2
1Shanghai Institute of Quality Inspection and Technical Research Co., Ltd., Shanghai 201114, China.
Journal of Food Protection
|May 7, 2026
Summary
A new framework assesses meat supply chain risks using advanced data analysis. It identifies pork and fresh meat as high-risk, with a growing threat from biological hazards like pathogens.
Area of Science:
- Food safety
- Risk assessment
- Supply chain management
Background:
- Global meat supply chains face increasing complexity and consumption, elevating public health risks.
- High-volume consumption regions are particularly vulnerable to these escalating risks.
Purpose of the Study:
- To develop a novel risk assessment framework for global meat supply chains.
- To integrate multi-source heterogeneous data for comprehensive risk evaluation.
Main Methods:
- Implementation of Large Language Model (LLM)-enhanced data governance.
- Utilizing a Delphi-Analytic Hierarchy Process (AHP) dual-dimensional indicator system.
- Employing an optimized Random Forest model for predictive analysis.
Main Results:
- Achieved predictive accuracy ranging from 88.25% to 99.24%.
- Identified pork as the highest-risk meat category and fresh meat as the riskiest processing type.
- Observed a shift towards biological hazards (pathogens) dominating over chemical contaminants since 2022.
Conclusions:
- The data-driven framework enables proactive and targeted risk management in meat safety.
- The tool supports regulatory decision-making for enhanced public health protection.
- Highlights the evolving nature of meat safety risks, emphasizing biological over chemical threats.