A Voting-Based Ensemble Deep Learning Method Focused on Multi-Step Prediction of Food Safety Risk Levels: Applications in Hazard Analysis of Heavy Metals in Grain Processing Products

A safety risk assessment method based on conditionally constrained game theory and adaptive ensemble learning: Application to wheat flour and rice

Food safety risk control and comprehensive assessment are crucial measures to ensure food safety. However, existing food safety risk assessment methods face challenges, such as unreasonable weight distribution of hazard factors and poor adaptability. Therefore, a safety risk assessment model based on conditionally constrained game theory and adaptive ensemble learning is proposed in this paper. Firstly, new constraints are established on the traditional game theory combination weighting method and solved using the augmented Lagrange multiplier method to obtain the optimal linear combination coefficients and actual composite risk values of the samples, which are taken together with the hazard factor detection data as inputs to the adaptive ensemble learning model. Then, an adaptive ensemble learning model is constructed, which prefers the base learner based on the combined measure of stability and accuracy, and predicts the composite risk value by using robust weighted random forest as the meta-learner. Finally, the model's validity was verified using wheat flour and rice hazard factor detection data. The experimental results indicate that the model's fit on the two datasets is 0.996 and 0.991, respectively, demonstrating strong generalization ability and high prediction accuracy. Meanwhile, unqualified products in wheat flour and rice can be effectively identified through risk thresholds, which helps to provide early warning of potential safety risks.

Lipidomics in forensic science: a comprehensive review of applications in drugs, alcohol, latent fingermarks, fire debris, and seafood authentication

Forensic science, an interdisciplinary field encompassing the collection, examination, and presentation of evidence in legal proceedings, has recently embraced lipidomics as a valuable tool. Lipidomics, a subfield of metabolomics, specializes in the analysis of lipid structures and functions, offering insights into biological processes that can aid forensic investigations. While not a substitute for DNA analysis in personal identification, lipidomics complements this technique by focusing on small biological molecules, with distinct sample requirements. This review comprehensively explores the current applications of lipidomics in forensic science. The review commences with an introduction to the concept and historical background of lipidomics, subsequently delving into its utilization in diverse areas such as drug analysis, ethyl alcohol and substitute assessment, latent fingermark detection, fire debris analysis, and seafood authentication. By showcasing the various biological materials and methods employed, this review underscores the potential of lipidomics as a powerful adjunct in forensic investigations.

Microplastics from face masks: Unraveling combined toxicity with environmental hazards and their impacts on food safety

Microplastics (MPs) refer to tiny plastic particles, typically smaller than 5 mm in size. Due to increased mask usage during COVID-19, improper disposal has led to masks entering the environment and releasing MPs into the surroundings. MPs can absorb environmental hazards and transfer them to humans and animals via the food chain, yet their impacts on food safety and human health are largely neglected. This review summarizes the release process of MPs from face masks, influencing factors, and impacts on food safety. Highlights are given to the prevalence of MPs and their combined toxicities with other environmental hazards. Control strategies are also explored. The release of MPs from face masks is affected by environmental factors like pH, UV light, temperature, ionic strength, and weathering. Due to the chemical active surface and large surface area, MPs can act as vectors for heavy metals, toxins, pesticides, antibiotics and antibiotic resistance genes, and foodborne pathogens through different mechanisms, such as electrostatic interaction, precipitation, and bioaccumulation. After being adsorbed by MPs, the toxicity of these environmental hazards, such as oxidative stress, cell apoptosis, and disruption of metabolic energy levels, can be magnified. However, there is a lack of comprehensive research on both the combined toxicities of MPs and environmental hazards, as well as their corresponding control strategies. Future research should prioritize understanding the interaction of MPs with other hazards in the food chain, their combined toxicity, and integrating MPs detection and degradation methods with other hazards.

Application and effectiveness of artificial intelligence for the border management of imported frozen fish in Taiwan

In Taiwan, the number of applications for inspecting imported food has grown annually and noncompliant products must be accurately detected in these border sampling inspections. Previously, border management has used an automated border inspection system (import food inspection (IFI) system) to select batches via a random sampling method to manage the risk levels of various food products complying with regulatory inspection procedures. Several countries have implemented artificial intelligence (AI) technology to improve domestic governmental processes, social service, and public feedback. AI technologies are applied in border inspection by the Taiwan Food and Drug Administration (TFDA). Risk management of border inspections is conducted using the Border Prediction Intelligent (BPI) system. The risk levels are analyzed on based on the noncompliance records of imported food, the country of origin, and international food safety alerts. The subjects of this study were frozen fish products, which have been under surveillance by the BPI system. The purpose of this study was to investigate the relevance between the noncompliant trend of frozen fish products using the adoption of the BPI system and the results of postmarket sampling inspections. The border inspection and postmarket sampling data were divided into two groups: IFI and BPI groups (corresponding to before and after the adoption of the BPI system, respectively). The Chi-square test was employed to analyze the noncompliant differences in products between before and after the BPI system adoption. Despite the number of noncompliance batches being statistically insignificant after the adoption of the BPI system, the noncompliance rate of frozen fish products at the border increased from 3.0% to 4.7%. Meanwhile, the noncompliance rate in the postmarket decreased from 2.1% to 1.9%. The results indicate that the BPI system improves the effectiveness of interception of noncompliant products at the border, thereby preventing the entrance of noncompliant products to the postmarket. The variables were further classified and organized according to the scope of this study and product characteristics. Furthermore, ordinal logistic regression (OLR) was employed to determine the correlations among border, postmarket, and major influencing factors. Based on the analysis of major influencing factors, small fish and fish internal organ products exhibited significantly high risk for fish body type and product type, respectively. The BPI system effectively utilizes the large amount of data accumulated from border inspections over the years. Additionally, real-time information on bilateral data obtained from the border and postmarket should be bidirectionally shared for effectively intercepting noncompliance products and used for improving the border management efficiency.

Advancing food safety risk assessment in China: development of new approach methodologies (NAMs)

Novel techniques and methodologies are being developed to advance food safety risk assessment into the next-generation. Considering the shortcomings of traditional animal testing, new approach methodologies (NAMs) will be the main tools for the next-generation risk assessment (NGRA), using non-animal methodologies such as in vitro and in silico approaches. The United States Environmental Protection Agency and the European Food Safety Authority have established work plans to encourage the development and application of NAMs in NGRA. Currently, NAMs are more commonly used in research than in regulatory risk assessment. China is also developing NAMs for NGRA but without a comprehensive review of the current work. This review summarizes major NAM-related research articles from China and highlights the China National Center for Food Safety Risk Assessment (CFSA) as the primary institution leading the implementation of NAMs in NGRA in China. The projects of CFSA on NAMs such as the Food Toxicology Program and the strategies for implementing NAMs in NGRA are outlined. Key issues and recommendations, such as discipline development and team building, are also presented to promote NAMs development in China and worldwide.

Prediction of Safety Risk Levels of Benzopyrene Residues in Edible Oils in China Based on the Variable-Weight Combined LSTM-XGBoost Prediction Model

To assess and predict the food safety risk of benzopyrene (BaP) in edible oils in China, this study collected national sampling data of edible oils from 20 Chinese provinces and their prefectures in 2019, and constructed a risk assessment model of BaP in edible oils with consumption data. Initially, the k-means algorithm was used for risk classification; then the data were pre-processed and trained to predict the data using the Long Short-Term Memory (LSTM) and the eXtreme Gradient Boosting (XGBoost) models, respectively, and finally, the two models were combined using the inverse error method. To test the effectiveness of the prediction model, this study experimentally validated the model according to five evaluation metrics: root mean square error (RMSE), mean absolute error (MAE), precision, recall, and F1 score. The variable-weight combined LSTM-XGBoost prediction model proposed in this paper achieved a precision of 94.62%, and the F1 score value reached 95.16%, which is significantly better than other neural network models; the results demonstrate that the prediction model has certain stability and feasibility. Overall, the combined model used in this study not only improves the accuracy but also enhances the practicality, real-time capabilities, and expandability of the model.

EL V.2 Model for Predicting Food Safety Risks at Taiwan Border Using the Voting-Based Ensemble Method

Border management serves as a crucial control checkpoint for governments to regulate the quality and safety of imported food. In 2020, the first-generation ensemble learning prediction model (EL V.1) was introduced to Taiwan's border food management. This model primarily assesses the risk of imported food by combining five algorithms to determine whether quality sampling should be performed on imported food at the border. In this study, a second-generation ensemble learning prediction model (EL V.2) was developed based on seven algorithms to enhance the "detection rate of unqualified cases" and improve the robustness of the model. In this study, Elastic Net was used to select the characteristic risk factors. Two algorithms were used to construct the new model: The Bagging-Gradient Boosting Machine and Bagging-Elastic Net. In addition, F_β was used to flexibly control the sampling rate, improving the predictive performance and robustness of the model. The chi-square test was employed to compare the efficacy of "pre-launch (2019) random sampling inspection" and "post-launch (2020-2022) model prediction sampling inspection". For cases recommended for inspection by the ensemble learning model and subsequently inspected, the unqualified rates were 5.10%, 6.36%, and 4.39% in 2020, 2021, and 2022, respectively, which were significantly higher (p < 0.001) compared with the random sampling rate of 2.09% in 2019. The prediction indices established by the confusion matrix were used to further evaluate the prediction effects of EL V.1 and EL V.2, and the EL V.2 model exhibited superior predictive performance compared with EL V.1, and both models outperformed random sampling.

Data-Driven Analysis of Risk-Assessment Methods for Cold Food Chains

The problem of cold-chain food safety is becoming increasingly prominent. Cold food chain risk assessment is an important way to ensure cold-chain food safety. Using CiteSpace, this study analyzes the knowledge map of research hotspots in the field of cold-chain food safety over the past 18 years, identifies the research keywords, presents the centrality statistics, and calculates the cluster values and average cluster contour values. Adopting a data-driven perspective, risk-assessment methods for cold food chains are summarized based on qualitative risk assessment, quantitative risk assessment, and comprehensive qualitative and quantitative risk assessment. The advantages and disadvantages of each are summarized. Finally, the problems and challenges in current cold food chain risk-assessment research are summarized in three aspects: the data credibility of cold food chain traceability systems, cold-chain food safety audit methods, and nontraditional cold food chain risk assessment. Suggestions are given for strengthening the cold food chain risk-assessment system to provide a decision-making reference to help regulatory authorities take risk prevention and control measures.

Research on the Construction of Grain Food Multi-Chain Blockchain Based on Zero-Knowledge Proof

As the main food source of the world's population, grain quality safety is of great significance to the healthy development of human beings. The grain food supply chain is characterized by its long life cycle, numerous and complex business data, difficulty defining private information, and difficult managing and sharing. In order to strengthen the ability of information application processing and coordination of the grain food supply chain under many risk factors, an information management model suitable for the grain food supply chain is studied based on the blockchain multi-chain technology. First, the information on key links in the grain food supply chain is analyzed to obtain privacy data classifications. Second, a multi-chain network model of the grain food supply chain is constructed, and based on this model, the hierarchical encryption and storage mode of private data as well as the relay cross-chain communication mode, are designed. In addition, a complete consensus process, including CPBFT, ZKP, and KZKP algorithms, is designed for the global information collaborative consensus under the multi-chain architecture. Finally, the model is verified through performance simulation, theory analysis, and prototype system verification in terms of its correctness, security, scalability, and consensus efficiency. The results show that this research model effectively reduces the storage redundancy and deals with problems of data differential sharing in traditional single-chain research, as well as provides a secure data protection mechanism, a credible data interaction mechanism, and an efficient multi-chain collaborative consensus mechanism. By attempting to apply blockchain multi-chain technology to the grain food supply chain, this study provides new research ideas for the trusted protection of data and information collaborative consensus in this field.

A Rice Hazards Risk Assessment Method for a Rice Processing Chain Based on a Multidimensional Trapezoidal Cloud Model

Rice is common in the human diet, making rice safety issues important. Moreover, rice processing safety is key for rice security, so rice processing chain risk assessment is critical. However, methods proposed to assess the rice processing chain risk have issues, such as the use of unreasonable thresholds for the rice processing chain and fixed weight. To solve these problems, we propose a risk assessment method for the rice processing chain based on a multidimensional trapezoidal cloud model. First, an evaluation model based on a multidimensional trapezoidal cloud model was established. Based on the historical evaluation results, Atanassov’s interval-valued intuition language numbers (AIVILNs) were introduced to determine the cloud model’s parameters. Second, the concept of dynamic weight was introduced to integrate the static and dynamic weights. An exponential function was used to construct dynamic weighting mechanisms, and the analytic hierarchy stage (AHP) was used to construct a static weight. The proposed method was validated by 104 sets of rice processing chain data, and the results show that the method could evaluate the risk level of the rice processing chain more accurately and reasonably than other methods, indicating that it can provide a sound decision-making basis for food safety supervision authorities.

Stacking- and voting-based ensemble deep learning models (SEDL and VEDL) and active learning (AL) for mapping land subsidence

This contribution presents a novel methodology based on the feature selection, ensemble deep learning (EDL) models, and active learning (AL) approach for prediction of land subsidence (LS) hazard and rate, and its uncertainty in an area involving two important plains - the Minab and Shamil-Nian plains - in the Hormozgan province, southern Iran. The important features controlling LS hazard were identified by ridge regression. Then, two EDL models were constructed by stacking (SEDL) and voting (VEDL) five dense deep learning (DL) models (model 1 to model 5) for mapping LS hazard. Thereafter, the predictive model performance was assessed by a precision-recall curve and Kolmogorov-Smirnov (KS) plot. A partial dependence plot (PDP), individual conditional expectation plots (ICEP), game theory, and a sensitivity analysis were used for the interpretability of the predictive DL model. According to SEDL - a model with higher accuracy - 34% (1624 km²), 14.7% (698 km²), and 19.2% (912 km²) of the total area were classified as being of very low, low, and moderate hazards, whereas 17.7% (845 km²) and 14.4% (683 km²) of area were classified as being of high and very high hazards, respectively. Based on all interpretability techniques, aquifer loss or groundwater drawdown is the most important feature controlling LS hazard, and it having the greatest impact on the SEDL model output. Based on a Taylor diagram and R² as model performance assessment indicators, SEDL-AL (with R² > 95% for training and test datasets) performed better than SEDL for quantify LS rate, the rate of LS ranging between 0 and 48.1 cm. The highest rate of LS occurred in the Minab plain - an area located downstream of the Minab Esteghlal dam. SEDL-AL was used to quantify the uncertainty associated with the LS rate. The observed values fell within predictions provided by SEDL-AL, which indicates a high accuracy of our predictive model. Overall, our newly developed modeling techniques are helpful tools for the spatial mapping of LS susceptibility and rate, and its uncertainty.

Anomaly Score-Based Risk Early Warning System for Rapidly Controlling Food Safety Risk

Food safety is a high-priority issue for all countries. Early warning analysis and risk control are essential for food safety management practices. This paper innovatively proposes an anomaly score-based risk early warning system (ASRWS) via an unsupervised auto-encoder (AE) for the effective early warning of detection products, which classifies qualified and unqualified products by reconstructing errors. The early warning analysis of qualified samples is carried out by early warning thresholds. The proposed method is applied to a batch of dairy product testing data from a Chinese province. Extensive experimental results show that the unsupervised anomaly detection model AE can effectively analyze the dairy product testing data, with a prediction accuracy and fault detection rate of 0.9954 and 0.9024, respectively, within only 0.54 s. We provided an early warning threshold-based method to conduct the risk analysis, and then a panel of food safety experts performed a risk revision on the prediction results produced by the proposed method. In this way, AI improves the panel’s efficiency, whereas the panel enhances the model’s reliability. This study provides a fast and cost-effective, food safety early warning method for detection data and assists market supervision departments in controlling food safety risk.

Prediction of Safety Risk Levels of Veterinary Drug Residues in Freshwater Products in China Based on Transformer

Early warning and focused regulation of veterinary drug residues in freshwater products can protect human health and stabilize social development. To improve the prediction accuracy, this paper constructs a Transformer-based model for predicting the safety risk level of veterinary drug residues in freshwater products in China to conduct a comprehensive assessment and prediction of the three veterinary drug residues with the maximum detection rate in freshwater products, including florfenicol, enrofloxacin and sulfonamides. Using the national sampling data and consumption data of freshwater products from 2019 to 2021, this paper constructs a self-built dataset, combined with the k-means algorithm, to establish the risk-level space. Finally, based on a Transformer neural network model, the safety risk assessment index is predicted on a self-built dataset, with the corresponding risk level for prediction. In this paper, comparison experiments are conducted on the self-built dataset. The experimental results show that the prediction model proposed in this paper achieves a recall rate of 94.14%, which is significantly better than other neural network models. The model proposed in this paper provides a scientific basis for the government to implement focused regulation, and it also provides technical support for the government’s intervention regulation.