Safety and transfer of veterinary drugs from substrate to black soldier fly larvae.

There is an increasing interest in edible insects in Europe for feed and food purposes. Quantitative information on the transfer of chemical hazards from substrates to larvae is needed to evaluate food and feed safety aspects. This evaluation is especially needed when organic substrates or residual streams such as manure will be applied as substrate, contributing to a circular food system. This study investigated the transfer of veterinary drugs from spiked substrate to black soldier fly larvae (Hermetia illucens). Veterinary drugs that are commonly administered to chicken, fattening pigs, and cattle and regularly detected in manure were included: three different antibiotics (enrofloxacin, oxytetracycline, sulfamethoxazole), three coccidiostats (narasin, salinomycin, toltrazuril) and one antiparasitic drug (eprinomectin). The chemicals were spiked to insect substrate to reach final concentrations of 0.5 and 5 mg/kg for the antibiotics and the antiparasitic drug, and 5 and 50 mg/kg for the coccidiostats. Black soldier fly larvae were reared for 1 week on the spiked substrates, and the transfer of the veterinary drugs to the larvae and frass was quantified using liquid chromatography coupled with tandem mass spectrometry. Only oxytetracycline and eprinomectin reduced the average weight and/or survival of the black soldier fly larvae. The transfer of the veterinary drugs to the larvae was on average 19.2% for oxytetracycline, 12% for enrofloxacin, 9.5% for narasin, 8.1% for eprinomectin, 3.9% for salinomycin, 4.2% for toltrazuril, and 0.2% for sulfamethoxazole, relative to concentrations in the substrate. Mass-balance calculations revealed that the larvae seem to metabolise veterinary drugs, and indeed, metabolites of enrofloxacin, sulfamethoxazole, and toltrazuril were detected in the larvae and frass. In conclusion, insect-rearing substrates should be evaluated for the presence of veterinary drug residues to ensure feed (and food) safety, as well as because of possible effects on insect growth.

A novel approach to identify critical knowledge gaps for food safety in circular food systems.

The transition from linear production towards a circular agro-food system is an important step towards increasing Europe's sustainability. This requires re-designing the food production systems, which inevitably comes with challenges as regards controlling the safety of our food, animals and the ecosystem. Where in current food production systems many food safety hazards are understood and well-managed, it is anticipated that with the transition towards circular food production systems, known hazards may re-emerge and new hazards will appear or accumulate, leading to new -and less understood- food safety risks. In this perspective paper, we present a simple, yet effective approach, to identify knowledge gaps with regard to food safety in the transition to a circular food system. An approach with five questions is proposed, derived from current food safety management approaches like HACCP. Applying this to two cases shows that risk assessment and management should emphasize more on the exposure to unexpected (with regards to its nature and its origin) hazards, as hazards might circulate and accumulate in the food production system. Five knowledge gaps became apparent: there's a need for (1) risk assessment and management to focus more on unknown hazards and mixtures of hazards, (2) more data on the occurrence of hazards in by-products, (3) better understanding the fate of hazards in the circular food production system, (4) the development of models to adequately perform risk assessments for a broad range of hazards and (5) new ways of valorization of co-products in which a safe-by-design approach should be adopted.

Framework for evaluation of food safety in the circular food system.

In order to minimise food waste, side streams from feed and food production are increasingly being (re-) used in food supply chains. Such reuse contributes to the desire to implement circularity in food and agricultural systems. However, the reuse of side products in circular food systems may impact food safety, for instance, contaminant residues present at low levels in biomass may accumulate when reusing streams. In order to assess potential food safety issues related to circular food systems, a framework has been developed in this study. Based on this framework, appropriate actions can be taken to prevent from human health risks. The framework consists of three steps: 1. Describing the changes in the food supply chain as a result of the circularity transition; 2. Identifying potential food safety hazards related to the change; and 3. Prioritising food safety hazards related to the circularity transition. For the prioritisation, both the presence of the hazards in final foods and the effects of the hazards on human health need to be assessed. Persistence of the hazard in the environment and potential transfer from the environment to the final food product are relevant elements to include. The framework was tested in three case studies, showing that it allows for a prioritisation between hazards. Based on the case study results, circularity not so much influences the health effects of the hazards, but rather their presence depending on the persistence and transfer of food safety hazards in a circular system.

The chemical and microbiological safety of emerging alternative protein sources and derived analogues: A review.

Climate change and changing consumer demand are the main factors driving the protein transition. This shift toward more sustainable protein sources as alternatives to animal proteins is also reflected in the rapid upscaling of meat and dairy food analogues. Such changes could challenge food safety, as new food sources could result in new and unexpected food safety risks for consumers. This review analyzed the current knowledge on chemical and microbiological contamination of emerging alternative protein sources of plant origin, including soil-based (faba bean, mung bean, lentils, black gram, cowpea, quinoa, hemp, and leaf proteins) and aquatic-based (microalgae and duckweeds) proteins. Moreover, findings on commercial analogues from known alternative protein sources were included. Overall, the main focus of the investigations is on the European context. The review aimed to enable foresight approaches to food safety concerning the protein transition. The results indicated the occurrence of multiple chemical and microbiological hazards either in the raw materials that are the protein sources and eventually in the analogues. Moreover, current European legislation on maximum limits does not address most of the "contaminant-food" pairs identified, and no legislative framework has been developed for analogues. Results of this study provide stakeholders with a more comprehensive understanding of the chemical and microbiological safety of alternative protein sources and derived analogues to enable a holistic and safe approach to the protein transition.

The role of larvae of black soldier fly and house fly and of feed substrate microbes in biotransformation of aflatoxin B1.

Over the past few years, there has been growing interest in the ability of insect larvae to convert various organic side-streams containing mycotoxins into insect biomass that can be used as animal feed. Various studies have examined the effects of exposure to aflatoxin B1 (AFB1) on a variety of insect species, including the larvae of the black soldier fly (BSFL; Hermetia illucens L.; Diptera: Stratiomyidae) and the housefly (HFL; Musca domestica L.; Diptera: Muscidae). Most of these studies demonstrated that AFB1 degradation takes place, either enzymatic and/or non-enzymatic. The possible role of feed substrate microorganisms (MOs) in this process has thus far not been investigated. The main objective of this study was therefore to investigate whether biotransformation of AFB1 occurred and whether it is caused by insect-enzymes and/or by microbial enzymes of MOs in the feed substrate. In order to investigate this, sterile and non-sterile feed substrates were spiked with AFB1 and incubated either with or without insect larvae (BSFL or HFL). The AFB1 concentration was determined via LC-MS/MS analyses and recorded over time. Approximately 50% of the initially present AFB1 was recovered in the treatment involving BSFL, which was comparable to the treatment without BSFL (60%). Similar patterns were observed for HFL. The molar mass balance of AFB1 for the sterile feed substrates with BSFL and HFL was 73% and 78%, respectively. We could not establish whether non-enzymatic degradation of AFB1 in the feed substrates occurred. The results showed that both BSFL and substrate-specific MOs play a role in the biotransformation of AFB1 as well as in conversion of AFB1 into aflatoxin P1 and aflatoxicol, respectively. In contrast, HFL did not seem to contribute to AFB1 degradation. The obtained results contribute to our understanding of aflatoxin metabolism by different insect species. This information is crucial for assessing the safety of feeding fly larvae with feed substrates contaminated with AFB1 with the purpose of subsequent use as animal feed.

Factors influencing the fate of chemical food safety hazards in the terrestrial circular primary food production system-A comprehensive review.

Food safety is recognized as a major hurdle in the transition toward circular food production systems due to the potential reintroduction and accumulation of chemical contaminants in these food systems. Effectively managing these hazardous contaminants in a risk-based manner requires quantitative insights into the factors influencing the presence and fate of contaminants in the entire circular food chain. A systematic literature review was performed to gain an up-to-date overview of the known factors and their influence on the transfer and accumulation of contaminants. This review focused on the terrestrial circular primary food production system, including the pathways between waste- or byproduct-based fertilizers, soil, crops, animal feed, and farmed animals. This review revealed an imbalance in research regarding the different pathways: studies on the soil-to-crop pathway were most abundant. The factors identified can be categorized as compound-related (intrinsic) factors, such as hydrophobicity, molecular weight, and chain length, and extrinsic factors, such as soil organic matter and carbon, pH, milk yield of cows, crop age, and biomass. Quantitative data on the influence of the identified factors were limited. Most studies quantified the influence of individual factors, whereas only a few studies quantified the combined effect of multiple factors. By providing a holistic insight into the influential factors and the quantification of their influence on the fate of contaminants, this review contributes to the improvement of food safety management for chemical hazards when transitioning to a circular food system.

Effects of dietary exposure to plant toxins on bioaccumulation, survival, and growth of black soldier fly (Hermetia illucens) larvae and lesser mealworm (Alphitobius diaperinus).

In their natural habitat, insects may bioaccumulate toxins from plants for defence against predators. When insects are accidently raised on feed that is contaminated with toxins from co-harvested herbs, this may pose a health risk when used for human or animal consumption. Plant toxins of particular relevance are the pyrrolizidine alkaloids (PAs), which are genotoxic carcinogens produced by a wide variety of plant species and the tropane alkaloids (TAs) which are produced by a number of Solanaceae species. This study aimed to investigate the transfer of these plant toxins from substrates to black soldier fly larvae (BSFL) and lesser mealworm (LMW). PAs and the TAs atropine and scopolamine were added to insect substrate simulating the presence of different PA- or TA-containing herbs, and BSFL and LMW were grown on these substrates. Bioaccumulation from substrate to insects varied widely among the different plant toxins. Highest bioaccumulation was observed for the PAs europine, rinderine and echinatine. For most PAs and for atropine and scopolamine, bioaccumulation was very low. In the substrate, PA N-oxides were quickly converted to the corresponding tertiary amines. More research is needed to verify the findings of this study at larger scale, and to determine the potential role of the insect and/or substrate microbiome in metabolizing these toxins.

The resilience of the pork supply chain to a food safety outbreak: The case of dioxins.

Food supply chains are constantly challenged by food safety hazards entering the chain. The ability of the supply chain to provide safe food within a reasonable time after such a food safety threat or shock can be investigated with the concept of resilience using food safety as an indicator. Resilience is then defined as the food safety performance deviation due to the shock and takes both the severity of the shock as well as the time to fully recover or reach a new equilibrium into account. This study developed a stochastic simulation model to evaluate the resilience of the Dutch pork supply chain to dioxin contamination in the feed. The resilience of the supply chain as well as the potential costs associated with the contamination are compared between several monitoring strategies with the aim to determine the optimal control points for dioxin monitoring. Model results show that collecting and analyzing samples at more than one control point along the pork supply chain, in particular at feed mills and fat melting facilities, resulted in the highest resilience and the lowest costs after a shock. This model and these results can be used by public and private decision makers to make proactive and informed decisions on the monitoring strategies to control dioxins in the pork supply chain that result in optimal resilience to a dioxin crises.

Building a resilient pork supply chain to Salmonella spp.

Salmonella spp. control in pork supply chains has always been a challenging issue and insufficient control can lead to high social and economic consequences. Conventional risk management and risk management approaches and models are not sufficient to address potential food safety shocks caused by Salmonella spp., as they mainly focus on assessing measures to reduce Salmonella spp. risks instead of developing the resilience capability (e.g., flexibility to adapt to sudden changes in the risks). Our study is the first that incorporated the resilience concept to the quantitative modeling of Salmonella spp. spread in the pork supply chain. The objective of this study was to explore the resilience performance of the pork supply chain under different food safety shocks caused by Salmonella spp., and to investigate the effectiveness of interventions on reducing the impact of these shocks on the resilience performance of the chain. Scenario analysis indicated that the effectiveness of the investigated resilience strategies or interventions depended on the risk profile (i.e., default, minimum, maximum level of Salmonella spp. contamination) of the pork supply chain. For pork supply chains with minimum and default risk profiles, more attention should be paid to increasing resilience of pigs towards Salmonella spp. infection. For supply chains with maximum risk profile, the focus should be on improving the performance of the slaughterhouse, such as careful evisceration, logistic slaughtering. To conclude, enhancing resilience performance of the pork supply chain can contribute to a safe pork supply.

Dynamic geospatial modeling of mycotoxin contamination of corn in Illinois: unveiling critical factors and predictive insights with machine learning.

Mycotoxin contamination of corn is a pervasive problem that negatively impacts human and animal health and causes economic losses to the agricultural industry worldwide. Historical aflatoxin (AFL) and fumonisin (FUM) mycotoxin contamination data of corn, daily weather data, satellite data, dynamic geospatial soil properties, and land usage parameters were modeled to identify factors significantly contributing to the outbreaks of mycotoxin contamination of corn grown in Illinois (IL), AFL >20 ppb, and FUM >5 ppm. Two methods were used: a gradient boosting machine (GBM) and a neural network (NN). Both the GBM and NN models were dynamic at a state-county geospatial level because they used GPS coordinates of the counties linked to soil properties. GBM identified temperature and precipitation prior to sowing as significant influential factors contributing to high AFL and FUM contamination. AFL-GBM showed that a higher aflatoxin risk index (ARI) in January, March, July, and November led to higher AFL contamination in the southern regions of IL. Higher values of corn-specific normalized difference vegetation index (NDVI) in July led to lower AFL contamination in Central and Southern IL, while higher wheat-specific NDVI values in February led to higher AFL. FUM-GBM showed that temperature in July and October, precipitation in February, and NDVI values in March are positively correlated with high contamination throughout IL. Furthermore, the dynamic geospatial models showed that soil characteristics were correlated with AFL and FUM contamination. Greater calcium carbonate content in soil was negatively correlated with AFL contamination, which was noticeable in Southern IL. Greater soil moisture and available water-holding capacity throughout Southern IL were positively correlated with high FUM contamination. The higher clay percentage in the northeastern areas of IL negatively correlated with FUM contamination. NN models showed high class-specific performance for 1-year predictive validation for AFL (73%) and FUM (85%), highlighting their accuracy for annual mycotoxin prediction. Our models revealed that soil, NDVI, year-specific weekly average precipitation, and temperature were the most important factors that correlated with mycotoxin contamination. These findings serve as reliable guidelines for future modeling efforts to identify novel data inputs for the prediction of AFL and FUM outbreaks and potential farm-level management practices.

Effects of Climate Change on Areas Suitable for Maize Cultivation and Aflatoxin Contamination in Europe.

The climate is changing in Europe: average temperatures are increasing, and so is the frequency of extreme weather events. Climate change has a severe impact on areas suitable for growing certain crops and on food safety, for example, affecting the occurrence of the aflatoxin contamination of maize. The aim of this study was to obtain insights into the impact of climate change on possible changes in land use in Europe, particularly in areas suitable for maize cultivation, and on the probability of the mycotoxin contamination of maize in order to give directions for long-term adaptation to climate change. By combining a land use model and a mycotoxin prediction model, the suitability of land for maize cultivation and the probability of aflatoxin contamination were estimated for suitable areas in Europe, comparing the current climate with the 2050 scenario. In 2050, the occurrence of aflatoxin contamination in Europe is predicted to severely increase, especially in Central and Southern Europe. More northern regions, presently unsuitable for maize cultivation, will become suitable for maize cultivation in 2050. In the baseline scenario, most regions suitable for maize cultivation have a low probability of aflatoxin contamination, whereas in 2050, about half of the regions suitable for maize cultivation have a medium to high probability of aflatoxin contamination. Regions for safely growing maize for human consumption will shift from the southern to the northern half of Europe.

Comparison of commercial allergen ELISA kits for egg detection in food matrices.

Consumption of low levels of egg already can evoke harmful physiological responses in humans in those allergic to eggs. By detection of egg in food products, using Egg ELISA kits to determine its unintended presence, food producers can respond to avoid potential safety or quality risks of their products. Selection of an ELISA kit fit for the issue at hand is challenging due to, amongst others, lack of information on assay performances with specified matrices. In this study, performances of seven commercial egg ELISA kits are compared for nine different relevant matrices: cookie, chocolate, pasta, dressing, stock cube, wine, vegetable drink and milk, ice cream and meat/meat replacers. The presence of egg was unified for all ELISA kits to mg total egg protein kg-1 food product. In every matrix, kit performances for recovery, intra- and interassay were compared, and also processing is accounted for by determination of egg in incurred samples. All seven kits were able to detect egg qualitatively at the VITAL3 ED01 level of 0.2 mg total egg protein and the corresponding relevant portion size for each matrix. For quantitative results, each ELISA kit showed an increase in detected egg concentration with increased egg levels and performed within the set criteria for recovery for the cookie, chocolate, stock cube and wine. For pasta, vegetable drink and milk, ice cream, and salad dressing, recovery of egg was within the set criteria for at least 4 ELISA kits. Most challenging matrices were meat/meat replacers, showing high matrix effects which could not be explained by the possible egg presence in the cognate blank. Only one ELISA kit was able to recover egg within the set criteria for the meat/meat replacer matrix. Results enable food industry to choose for ELISA kits suitable for egg detection in the matrix of interest.

Effects of pyrethroid and organophosphate insecticides on reared black soldier fly larvae (Hermetia illucens).

Black soldier fly larvae (Hermetia illucens) receive growing interest as a potential alternative animal feed source. These insects may be exposed to insecticide residues in the rearing substrate. This study aimed to investigate the effects of six different pyrethroid and organophosphate insecticides on this insect species' performance. The toxicity of two "model" substances for each of these classes (cypermethrin; pirimiphos-methyl) was quantified, with and without the synergist piperonyl butoxide (PBO). Critical effect doses corresponding to 10% yield (CED10) for cypermethrin (0.4 mg/kg) and pirimiphos-methyl (4.8 mg/kg) were determined. The addition of PBO to cypermethrin enhanced its relative potency with a factor 2.6. These data were compared against the relative toxicity of two analogue substances in each class (permethrin, deltamethrin; chlorpyrifos-methyl, malathion). Results suggest that exposure to concentrations complying with legal limits can cause significant reductions in yield. Exposure to multiple substances at lower concentrations resulted in negative additive and synergistic effects. Of the tested substances, deltamethrin was most toxic, causing 94% yield at 0.5 mg/kg. Analytical results suggested that transfer of tested substances to the larval biomass was substance- and concentration-specific, but appeared to be correlated to reduced yields and the presence of PBO. Transfer of organophosphates was overall low (<2%), but ranged from 8% to 75% for pyrethroids. Due to very low limits in insect biomass (∼0.01 mg/kg), high transfer may result in noncompliance. It is recommended that rearing companies implement lower contractual thresholds, and that policymakers consider adjusting legally allowed maximum residue levels in insect feed.

Impact of environmental factors on the presence of quinolizidine alkaloids in lupins: a review.

Lupin seeds have a high potential as an alternative for animal proteins in feed and food. However, the possible presence of alkaloids hinders the usage of lupins in human diets. This review aims to identify the main factors that influence the presence of alkaloids in lupins. A literature study covering English-published scientific papers in Scopus from 1980 to 2022 was performed. Biotic, abiotic, and genotypic factors influence the production of these toxic secondary metabolites by lupines. In particular, sweet cultivars with high 13-hydroxylupanine and 13-tigloyloxylupanine concentrations, abundant light exposure and standard diurnal cycles, well-watering procedures, relatively cold environment, N-deficient fertilizer with 240 mg K kg-1 and 60 mg P kg-1, high soil pH, and organic growing system conditions, are the best options to avoid high global alkaloid content. Results of this study can be used to develop predictive mechanistic models, although there is still the necessity to collect additional data by performing multi-variate studies.

Methodology for risk-based monitoring of contaminants in food - A case study in cereals and fish.

In this study, a methodology was developed that can be used as input for risk-based monitoring plans for chemical contaminants in food products. The novel methodology was applied to a case study in which cereals and fish were evaluated simultaneously for the possible presence of mycotoxins and heavy metals. The methodology was based on hazard quotients that were estimated by dividing the daily intake - using concentrations of the contaminants in the different food products and consumption of the respective products combined per product group - by the health based guidance value (HBGV) or reference points used for assessing potential health concerns (RPHC). The most relevant hazard-product combinations were further ranked based on the volume of import of the ingredients, per import country and a defined contaminant prevalence level per country. For fish, the hazard quotients were around ten times lower compared to the highest hazard quotients in cereals. Consumption of molluscs, mackerel-type fish and herring-type fish contaminated with mercury contributed most to the HBGV or RPHC. The top 25 hazard-product combinations for various age groups included: aflatoxin B1 in combination with wheat, rice (products), maize (products), and pasta, zearalenone in combination with wheat (products), T2/HT2-toxin in combination with rice (products), and DON in combination with wheat (products). The methodology presented showed to be useful in identifying the most relevant hazard-food-age group combinations and the most relevant import countries linked to these that should be included in the monitoring. As such, the method can help risk managers in establishing risk-based monitoring programs.

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.

Occurrence of dietary advanced glycation end-products in commercial cow, goat and soy protein based infant formulas.

Thermal treatment is a key step during infant formula (IF) processing which causes protein glycation and formation of dietary advanced glycation end-products (dAGEs). This study aimed to evaluate the glycation degree in IF in relation to the ingredients of the formula. dAGEs concentrations have been determined by UPLC-MS/MS in a range of commercial cow-based, goat-based, and soy-based IF. Results indicated that the protein source, protein composition, and amount and type of carbohydrates determines the level of protein glycation in IFs. The investigated soy-based formula had significant higher concentrations of arginine and arginine-derived dAGEs than cow-based and goat-based formulas. IF containing hydrolyzed proteins had higher dAGEs concentrations than those containing intact proteins. Lactose-containing formula was more prone to glycation than those containing sucrose and maltodextrin. Data showed glycation degree in IF cannot be estimated by a single compound, but the complete picture of the dAGEs should be considered.

Chemical food safety hazards in circular food systems: a review.

Food production has increasingly become effective but not necessarily sustainable. Transitioning toward circular production systems aiming to minimize waste and reuse materials is one of the means to obtain a more sustainable food production system. However, such a circular food production system can also lead to the accumulation and recirculation of chemical hazards. A literature review was performed to identify potential chemical hazards related to the use of edible and non-edible resources in agriculture and horticulture, and edible plant and animal by-products in feed production. The review revealed that limited information was available on the chemical hazards that could occur when reusing crop residues in circular agriculture. Frequently mentioned hazards present in edible and non-edible resources are heavy metals, process and environmental contaminants, pesticides and pharmaceuticals. For feed, natural toxins and pharmaceutical residues are of potential concern. Studies, furthermore, indicated that plants are capable of taking up chemical hazards when grown on contaminated soil. The presence of chemical hazards in manure, sewage sludge, crop residues, and animal by-products may lead to accumulation in a circular food production system. Therefore, it is relevant to identify these hazards prior to application in food production and, if needed, take precautionary measures to prevent food safety risks.

A predictive model on deoxynivalenol in harvested wheat in China: Revealing the impact of the environment and agronomic practicing.

Deoxynivalenol (DON) in wheat is one of the major food safety concerns worldwide. In this study, 70 characteristic precursive factors associated with environment and 6 agronomic practicing factors were explored, using historical data of 479 wheat fields in the Huang-Huai-hai, China. Results showed that DON concentrations influenced by air temperature, relative humidity, precipitation, and sunshine duration in the period from 17 days before flowering to 10 days before harvest. Rice crop rotation, straw returning, larger density of sowing, and lower latitude planting increased DON risk. Furthermore, an empirical model of DON prediction was established. The classification accuracy of internal and external validation were 87.73% (R2 = 0.62) and 80.21% (R2 = 0.60), respectively. This model is the first large-scale prediction of mycotoxin contamination in grain at harvest in China. It can be used to predict the risk of DON contamination for nearly 14 % of the global wheat supply.

Designing a monitoring program for aflatoxin B1 in feed products using machine learning.

Agricultural commodities used for feed and food production are frequently contaminated with mycotoxins, such as Aflatoxin B1 (AFB1). In Europe, both the government and companies have monitoring programs in place for the presence of AFB1. With limited resources and following risk-based monitoring as prescribed in EU Regulation 2017/625, these monitoring programs focus on batches with the highest probability of being contaminated. This study explored the use of machine learning algorithms (ML) to design risk-based monitoring programs for AFB1 in feed products, considering both monitoring cost and model performance. Historical monitoring data for the presence of AFB1 in feed products (2005-2018; 5605 records in total) were used. Four different ML algorithms, including Decision tree, Logistic regression, Support vector machine and Extreme gradient boosting (XGB), were applied and compared to predict the high-risk feed batches to be considered for further AFB1 sampling and analysis. The monitoring cost included the cost of: sampling and analysis, disease burden, storage, and of recalling and destroying contaminated feed batches. The ML algorithms were able to predict the high-risk batches, with an AUC, recall, and accuracy higher than 0.8, 0.6, and 0.9, respectively. The XGB algorithm outperformed the other three investigated ML. Its incorporation would result into up to 96% reduction in monitoring cost in 2016-2018, as compared to the official monitoring program. The proposed approach for designing risk based monitoring programs can support authorities and industries to reduce the monitoring cost for other food safety hazards as well.