Comparison of Three Modelling Approaches for Predicting Deoxynivalenol Contamination in Winter Wheat

Generic Food Safety Assessment: A Framework to Evaluate Food Safety Hazards Emerging from Change(s) in the Primary Production System - A Case Study Involving Intercropping

Food safety is a shared responsibility of all actors along the food supply chain. Changes in the primary production system can affect food safety hazards along the supply chain. This highlights the need for a framework that enables primary producers (i.e., farmers) to assess the potential food safety hazards and, if needed, to apply control measures. This paper presents a generic food safety assessment (GFSA) framework that has been developed based on Hazard Analysis and Critical Control Point (HACCP). The proposed framework was applied to a case study, i.e., the transition from sole cropping of oats to intercropping of oats with lupins. The application of the GFSA framework enabled the evaluation of potential changes in food safety hazards from this transition and the establishment of appropriate control measures. In addition, GFSA users can employ the results to support decision-making process. Our case study showed that implementing GFSA can be challenging for smallholder or individual farmers and may need coordinated action. Finally, effective and transparent communication is critical for managing food safety along the food supply chain, including when changes are implemented in primary production.

Prediction of deoxynivalenol contamination in spring oats in Sweden using explainable artificial intelligence

Weather conditions and agronomical factors are known to affect Fusarium spp. growth and ultimately deoxynivalenol (DON) contamination in oat. This study aimed to develop predictive models for the contamination of spring oat at harvest with DON on a regional basis in Sweden using machine-learning algorithms. Three models were developed as regional risk-assessment tools for farmers, crop collectors, and food safety inspectors, respectively. Data included: weather data from different oat growing periods, agronomical data, site-specific data, and DON contamination data from the previous year. Results showed that: (1) RF models were able to predict DON contamination at harvest with a total classification accuracy of minimal 0.72; (2) good predictions could already be made in June; (3) rainfall, relative humidity, and wind speed in different oat growing stages, followed by crop variety and elevation were the most important features for predicting DON contamination in spring oats at harvest.

Tracing the origin and authenticity of Spanish PDO honey using metagenomics and machine learning

The Protected Designation of Origin (PDO) indication for foods intends to guarantee the conditions of production and the geographical origin of regional products within the European Union. Honey products are widely consumed due to their health-promoting properties and there is a general interest in tracing their authenticity. In this regard, metagenomics sequencing and machine learning (ML) have been proposed as complementary technologies to improve the traceability methods of foods. Therefore, the aim of this study was to analyze the metagenomic profiles of Spanish honeys from three different PDOs (Granada, Tenerife and Villuercas-Ibores), and compare them with non-PDO honeys using ML models (PLS, RF, LOGITBOOST, and NNET). According to the results obtained, non-PDO honeys and Granada PDO showed higher beta diversity values than Tenerife and Villuercas-Ibores PDOs. ML classification of honey products allowed the identification of different microbial biomarkers of the geographical origin of honeys: Lactobacillus kunkeei, Parasaccharibacter apium and Lactobacillus helsingborgensis for PDO honeys and Paenibacillus larvae, Lactobacillus apinorum and Klebsiella pneumoniae for non-PDO honeys. In addition, potential microbial biomarkers of some honey varieties including L. kunkeei for Albaida and Retama del Teide varieties, and P. apium for Tajinaste variety, were identified. ML models were validated on an independent set of samples leading to high accuracy rates (above 90 %). This work demonstrates the potential of ML to differentiate different types of honey using metagenome-based methods, leading to high performance metrics. In addition, ML models discriminate both the geographical origin and variety of products corresponding to different PDOs and non-PDO products. Results here presented may contribute to develop enhanced traceability and authenticity methods that could be applied to a wide range of foods.

Machine Learning Applied to the Detection of Mycotoxin in Food: A Systematic Review

Mycotoxins, toxic secondary metabolites produced by certain fungi, pose significant threats to global food safety and public health. These compounds can contaminate a variety of crops, leading to economic losses and health risks to both humans and animals. Traditional lab analysis methods for mycotoxin detection can be time-consuming and may not always be suitable for large-scale screenings. However, in recent years, machine learning (ML) methods have gained popularity for use in the detection of mycotoxins and in the food safety industry in general due to their accurate and timely predictions. We provide a systematic review on some of the recent ML applications for detecting/predicting the presence of mycotoxin on a variety of food ingredients, highlighting their advantages, challenges, and potential for future advancements. We address the need for reproducibility and transparency in ML research through open access to data and code. An observation from our findings is the frequent lack of detailed reporting on hyperparameters in many studies and a lack of open source code, which raises concerns about the reproducibility and optimisation of the ML models used. The findings reveal that while the majority of studies predominantly utilised neural networks for mycotoxin detection, there was a notable diversity in the types of neural network architectures employed, with convolutional neural networks being the most popular.

Weighted Bayesian network for the classification of unbalanced food safety data: Case study of risk-based monitoring of heavy metals

Historical data on food safety monitoring often serve as an information source in designing monitoring plans. However, such data are often unbalanced: a small fraction of the dataset refers to food safety hazards that are present in high concentrations (representing commodity batches with a high risk of being contaminated, the positives) and a high fraction of the dataset refers to food safety hazards that are present in low concentrations (representing commodity batches with a low risk of being contaminated, the negatives). Such unbalanced datasets complicate modeling to predict the probability of contamination of commodity batches. This study proposes a weighted Bayesian network (WBN) classifier to improve the model prediction accuracy for the presence of food and feed safety hazards using unbalanced monitoring data, specifically for the presence of heavy metals in feed. Applying different weight values resulted in different classification accuracies for each involved class; the optimal weight value was defined as the value that yielded the most effective monitoring plan, that is, identifying the highest percentage of contaminated feed batches. Results showed that the Bayesian network classifier resulted in a large difference between the classification accuracy of positive samples (20%) and negative samples (99%). With the WBN approach, the classification accuracy of positive samples and negative samples were both around 80%, and the monitoring effectiveness increased from 31% to 80% for pre-set sample size of 3000. Results of this study can be used to improve the effectiveness of monitoring various food safety hazards in food and feed.

A perspective on data sharing in digital food safety systems

In this age of data, digital tools are widely promoted as having tremendous potential for enhancing food safety. However, the potential of these digital tools depends on the availability and quality of data, and a number of obstacles need to be overcome to achieve the goal of digitally enabled "smarter food safety" approaches. One key obstacle is that participants in the food system and in food safety often lack the willingness to share data, due to fears of data abuse, bad publicity, liability, and the need to keep certain data (e.g., human illness data) confidential. As these multifaceted concerns lead to tension between data utility and privacy, the solutions to these challenges need to be multifaceted. This review outlines the data needs in digital food safety systems, exemplified in different data categories and model types, and key concerns associated with sharing of food safety data, including confidentiality and privacy of shared data. To address the data privacy issue a combination of innovative strategies to protect privacy as well as legal protection against data abuse need to be pursued. Existing solutions for maximizing data utility, while not compromising data privacy, are discussed, most notably differential privacy and federated learning.

Nutritional impact of mycotoxins in food animal production and strategies for mitigation

Mycotoxins are toxic secondary metabolites produced by filamentous fungi that are commonly detected as natural contaminants in agricultural commodities worldwide. Mycotoxin exposure can lead to mycotoxicosis in both animals and humans when found in animal feeds and food products, and at lower concentrations can affect animal performance by disrupting nutrient digestion, absorption, metabolism, and animal physiology. Thus, mycotoxin contamination of animal feeds represents a significant issue to the livestock industry and is a health threat to food animals. Since prevention of mycotoxin formation is difficult to undertake to avoid contamination, mitigation strategies are needed. This review explores how the mycotoxins aflatoxins, deoxynivalenol, zearalenone, fumonisins and ochratoxin A impose nutritional and metabolic effects on food animals and summarizes mitigation strategies to reduce the risk of mycotoxicity.

Effective approaches for early identification and proactive mitigation of aflatoxins in peanuts: An EU-China perspective

Nearly 700,000 tonnes of peanuts are consumed annually in Europe. In the last 5 years, peanuts imported from China exceeded legal European Union (EU) aflatoxin limits more than 180 times. To prevent and mitigate aflatoxin contamination, the stages of the peanut chain most vulnerable to contamination must be assessed to determine how to interrupt the movement of contaminated produce. This paper discusses effective approaches for early identification and proactive mitigation of aflatoxins in peanuts to reduce a contaminant that is an impediment to trade. We consider (i) the results of the EU Commission's Directorate-General (DG) for Health and Food Safety review, (ii) the Code of Practice for the prevention and reduction of aflatoxins in peanuts issued by Food and Agriculture Organization/World Health Organization, (iii) the results from previous EU-China efforts, and (iv) the latest state-of-the-art technology in pre- and postharvest methods as essential elements of a sustainable program for integrated disease and aflatoxin management. These include preharvest use of biocontrol, biofertilizers, improved tillage, forecasting, and risk monitoring based on analysis of big data obtained by remote sensing. At the postharvest level, we consider rapid testing methods along the supply chain, Decision Support Systems for effective silo management, and effective risk monitoring during drying, storage, and transport. Available guidance and current recommendations are provided for successful practical implementation. Food safety standards also influence stakeholder and consumer trust and confidence, so we also consider the results of multiactor stakeholder group discussions.

Application of machine learning to the monitoring and prediction of food safety: A review

Machine learning (ML) has proven to be a useful technology for data analysis and modeling in a wide variety of domains, including food science and engineering. The use of ML models for the monitoring and prediction of food safety is growing in recent years. Currently, several studies have reviewed ML applications on foodborne disease and deep learning applications on food. This article presents a literature review on ML applications for monitoring and predicting food safety. The paper summarizes and categorizes ML applications in this domain, categorizes and discusses data types used for ML modeling, and provides suggestions for data sources and input variables for future ML applications. The review is based on three scientific literature databases: Scopus, CAB Abstracts, and IEEE. It includes studies that were published in English in the period from January 1, 2011 to April 1, 2021. Results show that most studies applied Bayesian networks, Neural networks, or Support vector machines. Of the various ML models reviewed, all relevant studies showed high prediction accuracy by the validation process. Based on the ML applications, this article identifies several avenues for future studies applying ML models for the monitoring and prediction of food safety, in addition to providing suggestions for data sources and input variables.

Key Global Actions for Mycotoxin Management in Wheat and Other Small Grains

Mycotoxins in small grains are a significant and long-standing problem. These contaminants may be produced by members of several fungal genera, including Alternaria, Aspergillus, Fusarium, Claviceps, and Penicillium. Interventions that limit contamination can be made both pre-harvest and post-harvest. Many problems and strategies to control them and the toxins they produce are similar regardless of the location at which they are employed, while others are more common in some areas than in others. Increased knowledge of host-plant resistance, better agronomic methods, improved fungicide management, and better storage strategies all have application on a global basis. We summarize the major pre- and post-harvest control strategies currently in use. In the area of pre-harvest, these include resistant host lines, fungicides and their application guided by epidemiological models, and multiple cultural practices. In the area of post-harvest, drying, storage, cleaning and sorting, and some end-product processes were the most important at the global level. We also employed the Nominal Group discussion technique to identify and prioritize potential steps forward and to reduce problems associated with human and animal consumption of these grains. Identifying existing and potentially novel mechanisms to effectively manage mycotoxin problems in these grains is essential to ensure the safety of humans and domesticated animals that consume these grains.

Decision support system for integrated management of mycotoxins in feed and food supply chains

Mycotoxins present a global food safety threat of our feed and food. Mycotoxins are toxic metabolites of certain fungi in agricultural products that are harmful to animal and human health. The presence of mycotoxins in these products depends on a variety of management and environmental factors in the field, during storage and/or processing of feed and food commodities. To date, information on mycotoxin management is available, but is not easy to access by supply chain actors. This study aimed to design, build and test a Decision Support System (DSS) that can help decision making on mycotoxin management by various actors along the feed and food supply chains. As part of this, available knowledge and data on mycotoxin prevention and control were collected and synthesised into easy to understand guidelines and tools for various groups of end-users. The DSS consists of four different modules: (a) static information module and (b) scenario analysis module, (c) dynamic module for forecasting mycotoxins, and (d) dynamic module for real-time monitoring of moulds/mycotoxins in grain silos. Intended end-users are all end-user groups for modules (a) and (b); growers and collectors for module (c) and; post-harvest storage managers for module (d). The DSS is user-friendly and accessible through PCs, tablets and smartphones (see https://mytoolbox-platform.com/ ). In various phases of the DSS development, the tool has been demonstrated to groups of end-users, and their suggestions have been taken into account, whenever possible. Also, a near final version has been tested with individual farmers on the easiness to use the system. In this way we aimed to maximise the DSS uptake by actors along the chain. Ultimately, this DSS will improve decision making on mycotoxin management; it will assist in reducing mycotoxin contamination in the key crops of Europe, thereby reducing economic losses and improving animal and human health.

Simultaneous quantitation of 3ADON and 15ADON chemotypes of DON-producing Fusarium species in Chinese wheat based on duplex droplet digital PCR assay

Pathogens within Fusarium species are the primary agents of Fusarium head blight (FHB) of wheat, which bring about yield reduction and deoxynivalenol (DON) contamination and are of great concern worldwide. DON-producing Fusarium species can be classified into 3-acetyldeoxynivalenol (3ADON) and 15-acetyldeoxynivalenol (15ADON) chemotypes according to the trichothecene metabolites they produce. The detection of these two chemotypes of pathogens is paramount to the successful implementation of disease management strategies and pathogen-related DON forecasting models. In this study, a duplex droplet digital PCR (duplex ddPCR) assay was developed that allowed for the simultaneous quantitation of 3ADON and 15ADON chemotypes of DON-producing Fusarium species. The assay specificity was tested against 30 isolates of target Fusarium species and several non-target Fusarium species that are frequently isolated from wheat in China. Analyzing 90 wheat samples collected from the North China plain and Yangtze River plain demonstrated that the duplex ddPCR assay coupled with magnetic bead-based DNA extraction was competent for investigating composition of 3ADON and 15ADON chemotypes in Chinese wheat. This assay will be useful for monitoring the epidemic and geographic distribution of 3ADON and 15ADON chemotypes of FHB pathogens, which will help with the disease control and DON management.

Improved Aflatoxins and Fumonisins Forecasting Models for Maize (PREMA and PREFUM), Using Combined Mechanistic and Bayesian Network Modeling-Serbia as a Case Study

Contamination of maize with aflatoxins and fumonisins is one of the major food safety concerns worldwide. Knowing the contamination in advance can help to reduce food safety risks and related health issues and economic losses. The current study aimed to develop forecasting models for the contamination of maize grown in Serbia with aflatoxins and fumonisins. An integrated modeling approach was used, linking mechanistic modeling with artificial intelligence, in particular Bayesian network (BN) modeling. Two of such combined models, i.e., the prediction model for aflatoxins (PREMA) and for fumonisins (PREFUM) in maize, were developed. Data used for developing PREMA were from 867 maize samples, collected in Serbia during the period from 2012 to 2018, of which 190 were also used for developing PREFUM. Both datasets were split randomly in a model training set and a model validation set. With corresponding geographical and meteorological data, the so-called risk indices for total aflatoxins and total fumonisins were calculated using existing mechanistic models. Subsequently, these risk indices were used as input variables for developing the BN models, together with the longitudes and latitudes of the sites at which the samples were collected and related weather data. PREMA and PREFUM were internally and externally validated, resulting in a prediction accuracy of PREMA of, respectively, 83 and 70%, and of PREFUM of 76% and 80%. The capability of PREMA and PREFUM for predicting aflatoxins and fumonisins contamination using data from the early maize growth stages only was explored as well, and promising results were obtained. The integrated approach combining two different modeling techniques, as developed in the current study, was able to overcome the obstacles of unbalanced data and deficiency of the datasets, which are often seen in historical observational data from the food safety domain. The models provide predictions for mycotoxin contamination at the field level; this information can assist stakeholders of the maize supply chain, including farmers, buyers/collectors, and food safety authorities, to take timely decisions for improved mycotoxin control. The developed models can be further validated by applying them into practice, and they can be extended to other European maize growing areas.

Aflatoxin contamination and recommendations to improve its control: a review

Aflatoxin producing fungi cause contamination of food and feed resulting in health hazards and economic loss. It is imperative to develop workable control measures throughout the food chain to prevent and reduce aflatoxin contamination. This is a critical review of contemporary published papers in the field. It is a review of reports from the original aflatoxin researches conducted on foods, from 2015-2020. Most of the reports show high aflatoxin contaminations in food at levels that exceed a regulatory limit of 20 μg/kg and 4 μg/kg set for foods for human consumption in the USA and European Union, respectively. The highest aflatoxin concentration (3,760 μg/kg) was observed in maize. Some of the strategies being deployed in aflatoxin control include application of biocontrol agents, specifically of Aflasafe™, development of resistant crop varieties, and application of other good agricultural practices. We recommend the adoption of emerging technologies such as combined methods technology (CMT) or hurdle technology, one health concept (OHC), improved regulations, on-line monitoring of aflatoxins, and creative art intervention (CAI) to prevent or restrict the growth of target aflatoxin causative fungi.

The MyToolbox EU-China Partnership-Progress and Future Directions in Mycotoxin Research and Management

Affordable and practical tools for farmers and food processors along the chain are required to efficiently reduce the risk of mycotoxin contamination of crops, feeds and foods. Developing new tools and enhancing existing ones was the mission of MyToolBox-a four-year EU-project that included important Chinese partners and joint research efforts. To identify future directions in mycotoxin research and management in China and their role in China-EU relations, a unique stakeholder workshop including group discussions was organized in Beijing. Six related topics: biocontrol, forecasting, sampling and analysis, silo management, detoxification, and the development of safe use options for contaminated materials were covered. The discussions clearly identified a critical need for smart, integrated strategies to address mycotoxin issues to attain safer food and feed, and to minimize losses and export rejections. Managing data on when, where and the size of mycotoxin contamination events and identifying the institution(s) to manage them are complex issues in China. Studies of microbes and novel, genetically-altered enzymes to limit pre-harvest contamination and to manage post-harvest product detoxification and alternate uses of contaminated materials are in the early stages in China. Further efforts are needed to increase the visibility of mycotoxin problems beyond the scientific and research communities.

Factors underlying Dutch farmers' intentions to adapt their agronomic management to reduce Fusarium species infection in wheat

Infection of wheat by Fusarium species can lead to Fusarium Head Blight (FHB) and mycotoxin contamination, thereby reducing food quality and food safety, and leading to economic losses. Agronomic management through the implementation of various pre-harvest measures can reduce the probability of Fusarium spp. infection in the wheat field. To design interventions that could stimulate wheat farmers to (further) improve their agronomic management to reduce FHB, it is key to understand farmers’ behaviour towards adapting their management. The aim of this paper was to understand the intention, underlying behavioural constructs, and beliefs of Dutch wheat farmers to adapt their agronomic management to reduce FHB and mycotoxin contamination in wheat, applying the Theory of Planned Behaviour (TPB). Data were collected from 100 Dutch wheat farmers via a questionnaire. The standard TPB analysis was extended with an assessment of the robustness of the belief results to account for the statistical validity of the analysis on TPB beliefs (i.e. to address the so-called expectancy-value muddle). Forty-six percent of the farmers had a positive intention to change their management in the next 5 years. The two behavioural constructs significantly related to this intention were attitude and social norm, whereas association with the perceived behavioural control construct was insignificant indicating that farmers did not perceive any barriers to change their behaviour. Relevant attitudinal beliefs indicated specific attributes of wheat, namely yield, quality and safety (lower mycotoxin contamination). This indicates that strengthening these beliefs—by demonstrating that a change in management will result in a higher yield and quality and lower mycotoxin levels—will result in a stronger attitude and, subsequently, a higher intention to change management. Interventions to strengthen these beliefs should preferably go by the most important referents for social norms, which were the buyers and the farmer cooperatives in this study.

AFLA-PISTACHIO: Development of a Mechanistic Model to Predict the Aflatoxin Contamination of Pistachio Nuts

In recent years, very many incidences of contamination with aflatoxin B₁ (AFB₁) in pistachio nuts have been reported as a major global problem for the crop. In Europe, legislation is in force and 12 μg/kg of AFB₁ is the maximum limit set for pistachios to be subjected to physical treatment before human consumption. The goal of the current study was to develop a mechanistic, weather-driven model to predict Aspergillus flavus growth and the AFB₁ contamination of pistachios on a daily basis from nut setting until harvest. The planned steps were to: (i) build a phenology model to predict the pistachio growth stages, (ii) develop a prototype model named AFLA-pistachio (model transfer from AFLA-maize), (iii) collect the meteorological and AFB₁ contamination data from pistachio orchards, (iv) run the model and elaborate a probability function to estimate the likelihood of overcoming the legal limit, and (v) manage a preliminary validation. The internal validation of AFLA-pistachio indicated that 75% of the predictions were correct. In the external validation with an independent three-year dataset, 95.6% of the samples were correctly predicted. According to the results, AFLA-pistachio seems to be a reliable tool to follow the dynamic of AFB₁ contamination risk throughout the pistachio growing season.