Food Control 162 (2024) 110471

Contents lists available at ScienceDirect

Food Control
journal homepage: www.elsevier.com/locate/foodcont

Quantitative and qualitative approach for accessing and predicting food

safety using various web-based tools
Hafiz Muhammad Rizwan Abid a, Nimrah Khan a, Athar Hussain a, Zainab Bintay Anis a,
Muhammad Nadeem a, Nauman Khalid a, b, *
a
Department of Food Science and Technology, School of Food and Agricultural Sciences, University of Management and Technology, Lahore, 54000, Pakistan
b
College of Health Sciences, Abu Dhabi University, Abu Dhabi, 59911, United Arab Emirates

A R T I C L E I N F O A B S T R A C T

Keywords: Food safety and quality are pivotal to a country’s economy and consumer health, with the whole food chain
Food safety depending on its success. Hence, with immense progress in industrialization, the need arose to develop an in­
Internet of things tegrated system constituting various essential elements to track, trace, and ensure food safety. The review reports
Big data
various web-based tools and their integration for assessing food safety during each step of the food chain. Various
Food processing
research resources were ransacked to find articles on the current review. These included Google Scholar, journal
Web-based tools
Predictive modeling articles, and other relevant sources with typical keywords and titles. Review data report about the feasibility of
artificial intelligence and allied techniques, i.e., vision-based technology and machine learning, that includes the
development of tailored algorithms for the food industry and food supply chain management. It also reports
physical objects like sensors, devices, and food processing tools, which can be paired with the Internet to form
the Internet of Things (IoT). Further, Big data pertains to a vast set of information from which complete data,
patterns, and outcomes can be extracted. Blockchain is another important thing to share, store, and process
authentic information, which securely reports every food processing step accordingly. Three is a network of tools
called web-based resources, which are helpful for qualitatively and quantitatively predicting food safety and
quality, thereby necessitating computer technology in food and nutrition science. Food-related hazards like
toxins, pesticides, and heavy metals from food commodities like fruits, vegetables, and meats can be traced,
reported, and eliminated. The review also presents insight into food fraud mitigation and related tools, including
blockchain technology, Comptox, ComBase, FDAiRISK, and others.

1. Introduction processing involves mixing corresponding food additives, and food hy­
giene must guarantee no mistakes (Knorr & Augustin, 2021). Food
According to news reports, the annual cost of foodborne illness in the shipping includes the shipment and preservation of food, which
United States will range from 10 to 83 billion dollars (Gemechu & Aliyo, frequently degrades due to fluctuations in the external environment.
2024). The risks to food safety in every step of food production, food Food retail is primarily used to prevent spoiled food from being sold by
preparation, food shipping, and food retail are further aggravated by the oversight (Casson et al., 2020). The four processes above describe the
global demand for food (Chapman & Gunter, 2018). Thus, ensuring the complete "life" of a kind of food, and they are all prone to impacting food
safety of the food is significant but complicated. Because of the safety (Soon, Brazier, & Wallace, 2020).
complexity of the food industry processes, concerns regarding food The importance of food safety is increased as global trade and eco­
safety might occur at any time. In agriculture, food production mainly nomic advancements continue to unfold. The potential to enhance food
refers to the rearing of livestock and the establishment of crops. As a safety is closely tied to the progression of technologies such as artificial
result, ensuring food safety at this point is essential and demanding intelligence (A.I.) and big data (Kumar, Rawat, Mohd, & Husain, 2021).
(Ayyildiz & Erdal, 2021). Food processing is the transformation of crops People experienced difficulty in realizing where the food they bought
or animals into consumables meant for human consumption. Food had come from in the past. Food traceability makes obtaining

* Corresponding author. Department of Food Science and Technology, School of Food and Agricultural Sciences, University of Management and Technology,
Lahore, 54000, Pakistan.
E-mail addresses: [email protected], [email protected] (N. Khalid).

https://doi.org/10.1016/j.foodcont.2024.110471
Received 12 December 2023; Received in revised form 6 March 2024; Accepted 24 March 2024
Available online 25 March 2024
0956-7135/© 2024 Elsevier Ltd. All rights reserved.
H.M.R. Abid et al. Food Control 162 (2024) 110471

information about a food’s whole manufacturing chain simpler. Sensors analytics processes and analyzes the vast data generated by IoT sensors
and other gadgets are used in the manufacturing process of food to and facilitates actionable insights (Kalinowska, Wojnowski, & Tobis­
monitor food safety and produce data for traceability. zewski, 2021). Additionally, blockchain technology offers immutable
Then, to process data, records for food manufacturing, food shipping, and transparent record-keeping capabilities, ensuring the integrity and
and food retailing are retained. During these stages, lots of data is traceability of food products from farm to fork. By exploring these
collected, and processing is done using big data and A.I. technology. A.I. innovative technologies synergistically, qualitative approaches for
makes it feasible for systems to "understand" the meanings obtained accessing and predicting food safety enhance risk assessment and
from data. (Kumar et al., 2021). The capacity of A.I. to automatically management strategies and contribute to the overarching goal of safe­
acquire knowledge from an extensive quantity of data is one of its most guarding public health and consumer confidence in the food supply.
significant benefits, necessitating big data for A.I. (Dong et al., 2021).
Big Data is a word used to denote vast amounts of quickly created and 2.1. Artificial intelligence
collected sophisticated online data via multiple places (research, in­
dustrial, organizational, online databases, social media). DSS (Decision A.I. stands at the forefront of cutting-edge research within computer
Support System) is a computer-based decision-making system that de­ science. This dynamic field encompasses diverse ideas, including ML,
tects and evaluates problems within a specific domain of information deep learning (DL), and more. A.I. seeks to replicate human intelligence
and management systems. The current landscape of food safety research within machines, equipping them with the capacity to reason and
underscores a burgeoning interest in harnessing the potential of Big Data behave in ways reminiscent of humans. (Zhou, Zhang, & Wang, 2021).
and DSS for enhancing food safety risk assessments (Liu, Li, Yang, & A.I. presents a promising opportunity within the food industry, funda­
Guo, 2018). Big Data and DSS are pivotal in tackling intricate, un­ mentally transforming various aspects of our food system (Kumar et al.,
structured, and semi-structured challenges within food safety. They 2021). Its role spans from revolutionizing precision farming to
open doors to pioneering data-driven approaches to bolster the accuracy enhancing food production and consumption. Furthermore, A.I. con­
and efficiency of food safety risk assessments (G. Liu et al., 2018). As a tributes significantly to quality control measures across the food sector,
result, food industries and risk analysts for food safety have become reshaping how we perceive food production, quality assessment, and
more interested in applying these web-based technologies for predicting delivery, mainly through the rise of intelligent mobile applications
food fraud, incidents, and risk assessments (Talari, Cummins, McNa­ (Bendre et al., 2022).
mara, & O’Brien, 2022). The first section of the review examines the Integrating AI-powered systems can be leveraged effectively to build
extensive body of work surrounding the utilization of A.I., Big Data, and and manage extensive food databases and conduct in-depth analyses.
blockchain technology in food safety, while the second section explores This has the potential to foster a healthier and more cost-effective food
how these cutting-edge technologies are being applied to detect and industry, benefiting both workers and consumers. Traditional methods
mitigate a wide spectrum of food safety hazards. The third section re­ for ensuring food safety, such as detecting food adulteration, are often
views the web-based technologies for assessing food safety risks and costly and intricate, requiring specialized infrastructure and intensive
their predictions entailing pesticides, heavy metals, allergens, and other manual labor. These procedures can be tedious and inefficient. A.I.
food processing-related hazards. Section four provides overview of (Fig. 1) offers an avenue to develop cost-effective automated systems
integration of web-based tools in food safety offering an analysis of these that end-users can employ to detect food safety hazards in industries
tools by considering their applications, domains of expertise, and the dealing with fruits, vegetables, and dairy products (D’Amore, Di Vaio,
outcomes they deliver. This analysis is a valuable resource for stake­ Balsalobre-Lorente, & Boccia, 2022; Miyazawa et al., 2022).
holders (food manufacturers, regulatory agencies, and researchers) Two forms of data are essential to construct an A.I. system capable of
seeking to navigate the dynamic landscape of food safety technology and detecting food safety hazards. The first type revolves around developing
make informed choices regarding the most suitable tools for their spe­ a vision-based model, focusing on assessing the quality of food items
cific needs (Bendre, Shinde, Kale, & Gilda, 2022; Casino et al., 2021; Liu based on parameters such as color, texture, size, shape, defects, and
et al., 2022; Sood & Singh, 2021). Fifth and sixth section contributes to morphological features. Machine learning models are trained on this
advancing food safety research by showcasing how emerging technol­ data to classify food items, such as fruits, vegetables, or dairy products,
ogies can be leveraged to mitigate a wide range of food safety issues in based on their physical characteristics (Septiarini et al., 2021). The
the food safety domain encompassing a wide range of food processing second data type encompasses moisture content, pH levels, temperature,
and research-relevant premises. pressure, humidity, and viscosity. These parameters explore the chem­
ical composition of food products. IoT can be harnessed to gather data
2. Qualitative approaches for accessing and predicting food from various sensors, enabling the detection of food safety hazards
safety by web-based tools based on the chemical composition of food products (Liu et al., 2022).

Qualitative approaches exploiting web-based tools have emerged as 2.1.1. Vision-based artificial intelligence
indispensable assets by offering insights into the complex interplay of Assessing food quality relies significantly on its visual attributes.
factors influencing food quality and integrity. Key technologies driving Using vision-based models involves using images of food products to
these approaches include A.I., machine learning (ML), Internet of Things identify potential hazards through training ML and DL models. These
(IoT), big data analytics, and blockchain technology. By harnessing the models play a crucial role in evaluating food quality by analyzing textual
power of A.I. and ML algorithms, researchers can sift through vast data. Additionally, computer vision systems are instrumental in auto­
amounts of data to identify patterns, anomalies, and predictive in­ mating and accurately classifying food products based on adulteration
dicators of food safety risks. These technologies enable the development levels. The process begins with acquiring digital images of the food,
of sophisticated models capable of forecasting potential hazards, followed by employing ML models for predictive analysis (Sood & Singh,
contamination events, and adulteration incidents with high accuracy 2021).
(Qian, Murphy, Orsi, & Wiedmann, 2023). Artificial Neural Networks (ANNs) have recently gained substantial
The integration of IoT devices, i.e., sensors embedded in food pro­ popularity for their ability to extract meaningful features from data with
cessing facilities, vehicles, and storage units, continuously capture minimal prior knowledge and independent feature selection (Dobchev &
critical parameters such as temperature, humidity, pH levels, and mi­ Karelson, 2016). ANNs draw inspiration from the human neural system,
crobial activity. This granular data facilitates early detection of de­ operating like a black box that adjusts its internal weights through
viations from optimal conditions, allowing stakeholders to intervene training. ANN consists of a single layer of neurons, known as percep­
promptly and mitigate potential safety threats. Moreover, big data trons, interconnected to each input feature. ANNs are particularly

2
H.M.R. Abid et al. Food Control 162 (2024) 110471

Fig. 1. An IoT ecosystem ensuring food safety with user devices, cloud networks, predictive analysis, and decision-making components.

valuable in cases where the exact underlying model is unknown. Moving on to the second phase, the focus shifts to developing and
Training these neural networks with food product ingredients using implementing an A.I. system. This A.I. system harnesses the data ac­
images allows for extracting textural, shape, and morphological prop­ quired from the IoT sensors. Its primary mission is to oversee the envi­
erties (Olaniyi, Adekunle, Odekuoye, & Khashman, 2017). Therefore, ronmental conditions in which food products are stored and transported.
ANN classifiers can be successfully implemented to inspect and grade By continuously monitoring these conditions, detecting deviations that
food product quality. However, due to their linear nature, ANNs with a might compromise food quality is possible. This proactive approach
single layer of perceptrons are limited in their ability to handle complex significantly reduces the risk of food deterioration. This A.I. system
relationships within the data. They are suitable for relatively simpler (Fig. 2) monitors environmental factors and uses machine learning
tasks or datasets with linearly separable classes. This vision-based models to analyze the collected data. (Pal & Kant, 2018).
approach is considered environmentally friendly because it eliminates In the final phase, the system provides intelligent recommendations
the need for chemical reagents. It offers numerous advantages, including and predictions based on its analysis. It becomes adept at discerning
swift and efficient results, cost-effectiveness, and broad applicability whether a food product is safe for consumption or if it harbors any po­
within the food industry for classification and grading purposes (Sharma tential hazards. Moreover, it can gauge the overall quality of the food,
& Sawant, 2017). making it a valuable tool for monitoring food authenticity and detecting
adulteration (Magnus, Virte, Thienpont, & Smeesters, 2021). In essence,
2.1.2. Internet of Things sensor-based artificial intelligence developing a web-based food safety hazard detection system (Table 1)
The IoT refers to a network of interconnected physical objects or hinges on the seamless integration of IoT devices for data sensing and
"things" embedded with sensors, software, and other technologies to AI-driven machine learning models for predictive analysis. This holistic
collect and exchange data over the Internet or other communication approach empowers us to create a safer and more secure food supply
networks. The key idea behind IoT is to enable these objects to chain, assuring consumers of the quality and safety of the products they
communicate with each other and with centralized systems, allowing consume.
them to gather data, make intelligent decisions, and perform actions
based on the information they receive. IoT has a wide range of appli­ 2.1.3. Machine learning
cations across various industries, including smart homes, smart cities, ML is a vital subfield within artificial intelligence, empowering
industrial automation, healthcare, agriculture, and more, where it can computers to learn autonomously without explicit programming. At its
enhance efficiency and automation and provide valuable insights by core, it revolves around crafting algorithms that can glean insights from
analyzing collected data (D’Amore et al., 2022). data and subsequently make informed predictions based on that data
Leveraging the synergy between A.I. and the IoT has proven potent in (Datta & Sahoo, 2021). The two primary branches of ML are supervised
bolstering food security and safety measures. This innovative pairing and unsupervised learning. In the category of supervised ML, a pre­
transforms the concept of food safety hazard detection into an intelli­ defined set of categories exists into which various items, such as food
gent, interconnected system (Smith, 2018). Through a sequence of three products, are sorted. Extensive training data is accessible for each of
essential phases, the IoT plays a pivotal role in this process: data these distinct categories. Within this framework, the system harnesses
collection, data analysis, and data prediction. Data acquisition is classification algorithms like Naive Bayes (N.B.), Support Vector Ma­
accomplished through a diverse array of sensors in the initial phase. chines (SVM), Decision Trees (D.T.), and k-Nearest Neighbors, among
These sensors are strategically deployed to capture a spectrum of data others, to meticulously train a model. Once trained, this model becomes
points vital for food safety assessment. They serve as the eyes and ears of proficient in categorizing food items by making predictions based on the
the system, collecting information on variables such as temperature, features it has learned from the provided data (Piccialli, Casolla, Cuomo,
humidity, alcohol levels, and light exposure, all of which can influence Giampaolo, & Di Cola, 2019). In the food industry, through supervised
food quality (Stracener, Samelson, Mackie, & Ihaza, 2019). ML, foodborne pathogens can be detected in food samples based on

3
H.M.R. Abid et al. Food Control 162 (2024) 110471

Fig. 2. Food Adulteration Detection with IoT: Sensor data transmitted to the cloud via an IoT gateway, processed by machine learning models, and relayed to users,
ensuring informed food choices.

input variables such as temperature, pH, and water activity. potential of ML in enhancing food safety measures by providing pre­
Conversely, unsupervised ML takes a different approach, operating dictive capabilities and actionable insights derived from comprehensive
without the luxury of labeled data. Here, models must decipher patterns data analysis.
and structures within the data independently, without prior guidance. In
food establishments, unsupervised ML (Table 2) is used to cluster food 2.1.4. Practical applications
samples based on their chemical composition or microbial profiles and The practical application of A.I. involves the AI-assisted detection of
identify patterns in food safety incidents or outbreaks based on data Escherichia coli using optical imaging techniques, enabling swift and
from multiple sources, such as social media, news reports, and public accurate identification of harmful pathogens in food samples (Liu &
health records (Piccialli et al., 2019). Chen, 2018). Additionally, AI facilitates vegetable growers’ manage­
DL, a subset of machine learning, mirrors the human brain’s neural ment of potential issues in leafy greens production by analyzing weather
network architecture. Recently, it has emerged as a formidable tech­ patterns, geographical data, and water quality metrics, thereby enabling
nique for pattern recognition. DL models are constructed with multiple proactive interventions to mitigate risks. Moreover, AI is instrumental in
layers of interconnected neurons, offering the advantage of automatic monitoring food workers’ behaviors within workplace settings,
feature extraction and the potential to tackle highly intricate problems providing insights into adherence to hygiene and sanitation protocols.
with remarkable accuracy. Multiple layers of the DL model distinguish it For instance, a Japanese company has developed an AI system to oversee
from other neural networks. More layers enable DL models to capture slaughterhouse staff activities, ensuring consistent application of good
and process highly complex data distributions, improving performance practices and proper handwashing in kitchens (Talari et al., 2022).
on challenging tasks. (Chander, 2020). The food industry has an Furthermore, AI-supported initiatives by regulatory bodies, such as the
emerging trend of employing machine learning models. These models US FDA, aid in identifying problematic batches of imported seafood like
are increasingly being utilized across various domains, including pre­ shrimps, enhancing overall food safety surveillance efforts.
dicting food safety, grading food and agricultural products, and opti­ By analyzing the behavioral patterns of customers and staff, AI sys­
mizing food packaging processes (Liu & Chen, 2018). tems can identify potential risks and implement preventive measures to
Various studies have harnessed these models for monitoring and ensure product quality. Identifying and characterizing microbes through
predicting food safety parameters, underlining their potential as valu­ AI systems can streamline processes, yet the accuracy and reliability of
able tools in this critical endeavor. Thus, ML offers a promising avenue such methodologies must be thoroughly evaluated. Moreover, while
for enhancing food safety protocols (Vithu & Moses, 2016). In another beneficial, modeling microbial population dynamics and growth re­
study, ML methods were effectively applied in the dairy sector, quires robust data inputs and validation to ensure its effectiveness in
including detecting unknown hazards through word embedding tech­ real-world scenarios.
niques and predicting hazards based on changes in influential factors
(Liu et al., 2022).ML techniques such as Bayesian networks, neural
networks, random forests, and decision trees have revolutionized the 2.2. Food recognition and classification
identification of early warning signals and emerging risks in food safety.
For example, Bayesian networks have successfully predicted the occur­ Object detection is a classical computer vision application, particu­
rence of chemical food hazards like pesticide residues and mycotoxins in larly in developing automated food sorting systems. This involves
fruits and vegetables across different regions. This approach integrates recognizing objects in an image and determining their locations. Con­
data from various sources, including E.U. food law enforcement alerts, volutional Neural Networks (CNNs) are commonly used for object
agricultural statistics from the Food and Agriculture Organization detection. They work by extracting features from input images and using
Corporate Statistical Database (FAOSTAT), and meteorological data them to differentiate objects (Gong et al., 2019).
from the National Oceanic and Atmospheric Administration (NOAA) Food recognition and classification are critical for tracking daily
(Bouzembrak & Marvin, 2019). These advancements highlight the diets, as dietary habits significantly impact people’s health. Diabetic and
allergic individuals closely monitor and control their diets. Information

4
H.M.R. Abid et al. Food Control 162 (2024) 110471

Table 1 Table 2
IoT and its applications in food Safety. Application of machine learning in food safety.
Techniques Description Conclusion Reference Methodology Description Results and Reference
and Application Accuracy
IoT and RFID Tags Proposing a method Establishment of a Jia et al.
for Food Quality to create a quality food quality (2011) SVM with Gas Detecting sour skin Validation: 85%, Li et al. (2009)
Supervision supervision platform supervision system Sensor Array in onions using Training: 81%
for food production and PCA SVM and gas sensor
utilizing IoT. The array.
approach ANN for Beer Classifying beer Classification Debska et al.
incorporates RFID Sample samples into good Accuracy: Almost (2011)
tags and 1-D codes, Classification quality and 100%
establishes food unsatisfactory using
quality models based 12 features.
on ontology-based ML Techniques Authenticating Best Accuracy: Gomez-Meire
context modeling for Wine wine samples from Random Forest et al. (2014)
theory, and outlines Sample 6 different origins
methods for Authentication using ML
combining and techniques.
presenting service SVM and KNN Detecting tomato Tomato SVM: Bandyopadhyaya
functions for various for Ripeness and ladyfinger 64.23%, KNN: et al. (2011)
users. Detection in ripeness with SVM 92.86%
Oxygen and Evaluating the Monitoring Eom et al. Tomato and (64.23%) and KNN
Carbon Dioxide freshness of concentrations of (2012) Ladyfinger (92.86%).
Sensors with vegetables using an oxygen and carbon SVM and Partial Identifying sesame RMSEP: 19%– Dan Peng et al.
RFID Tags oxygen and carbon dioxide gases Least Square oil adulteration 4.29% (2014)
dioxide Method with vegetable oils
concentration (PLSM) for using gas
monitoring system Sesame Oil chromatography.
incorporating radio Adulteration
RFID technology. ANN and SVM Building a model to Effective Mu et al. (2016)
IoT for Food Utilizing IoT to Development of a Chanthini for Oils predict adulteration Differentiation
Quality remotely monitor the food quality et al. Classification concentration, then
Monitoring quality of perishable monitoring system (2017) classifying pure and
food. Employing using IoT mixed oils.
Raspberry Pi as a ELM and K-ELM Predicting the K-ELM: Best Zhang et al.
sensor node and for Dairy quality risks of Accuracy (2018)
gateway node to Product dairy products
collect data from Quality using ELM and K-
various sensors, Prediction ELM.
including ANN for Milk Classifying milk Used for Milk Kobek et al.
temperature, Adulteration samples based on Adulteration (2017)
humidity, and Detection images captured
moisture. with bromothymol
Internet of Things Detecting poor food Development of a Gupta et al. blue.
for Food Quality quality using various food adulteration (2018) CNN and SVM SVM used to detect CNN Accuracy: Al Sarayreh et al.
Monitoring sensors to record monitoring system for Red Meat red meat 4.4% (2018)
data and Raspberry using IoT Adulteration adulteration, CNN
Pi for system control. Detection for spectral and
Development of a spatial features.
food adulteration Image Processing Detection of the The defects of the Rong et al. (2017)
monitoring system. with algorithm surface defects in oranges were
Detection and Utilizing atrous ResNet-101 Kang et al. the oranges in grey- identified using
Segmentation spatial pyramid backbone achieved (2020) level images. to the image
Network for pooling and gate superior segment the various processing and
Apple feature pyramid performance: F1 defects window local
Harvesting network to enhance score of 0.832 for segmentation
feature extraction for apple detection and algorithm
real-time fruit 87.6% and 77.2% for
detection and apple semantic
segmentation during segmentation. different cases of olives, including normal ones, empty ones, inade­
apple harvesting
quately de-pitted at various angles, and atypical instances, including
using a visual sensor
in orchards.
extraneous components such as leaves and rocks. These studies show­
case the versatility of CNNs in food sorting and hazard detection,
highlighting their effectiveness in automating tasks essential for main­
about food products can be captured through images, a cost-effective taining food safety and dietary control (Pariente, Cancilla, Wierzchos, &
and straightforward method for food recognition and classification Torrecilla, 2018).
(Mezgec, Eftimov, Bucher, & Seljak, 2019). Several ML and DL methods
are available for this purpose, with CNNs being a popular choice for 2.2.1. Practical applications
image identification. CNNs were utilized by researchers to analyze im­ Computer vision systems offer versatile applications across various
ages of walnuts to identify a range of extraneous items, including food industry sectors, including the meat industry, where they play a
packing materials, remnants of leaves, dust, paper, and metal and plastic crucial role in pathogen identification. These systems are adept at
fragments (Rong, Xie, & Ying, 2019). Image-based methods are detecting pathological lesions and fecal contamination on abattoir car­
commonly used to generate data for analyzing physical hazards, and casses, aiding in maintaining stringent hygiene standards throughout
neural networks are frequently employed as machine learning algo­ meat production. In a study by Zhong, Reibman, Mina, and Deering
rithms for modeling. In another study, researchers used CNNs to classify (2021), a computer vision system was employed to identify hygiene

5
H.M.R. Abid et al. Food Control 162 (2024) 110471

using a hand hygiene action detection mechanism. Hence, by engaging Critical Control Points (HACCP), facilitating real-time monitoring at
computer vision technology, potential incidents related to food safety critical control points.
can be identified and mitigated, thereby ensuring the well-being of Utilizing biosensors, alongside other sensor types, provides situa­
workers and the food being handled. tional awareness within food production environments, enabling the
timely detection of abnormal behaviors, such as deviations in the
2.3. Internet of Things (IoT) physical activity of livestock. Additionally, the IoT finds application in
enhancing meat cold chain logistics, with technologies like RFID-based
IoT represents the interconnectedness of various entities such as tracking and humidity sensors ensuring optimal conditions during
sensors, devices, machinery, and computing tools through the Internet transportation and storage. For instance, RFID tags enable the individual
or other communication mediums like Wi-Fi, Bluetooth, and RFID tracing of pigs throughout their productive lifecycle, providing valuable
(Gomez, Chessa, Fleury, Roussos, & Preuveneers, 2019). One of the most data from farm to slaughterhouse (Al Rasyid et al., 2020).
significant applications of IoT has been in food supply chains, primarily
for tracking food products and ensuring food safety and quality (Wit­ 2.4. Big data
jaksono, Saeed Rabih, Yahya, & Alva, 2018). Most IoT applications in
food safety are centered on valuable food items like meat, perishable Big Data refers to vast and complex sets of information that go
goods, and agricultural products. These applications utilize sensors to beyond what traditional data-handling approaches can manage. It
monitor crucial parameters such as temperature, humidity, and location stands out due to its immense size, its ability to accumulate quickly, and
(Pal & Kant, 2018). The emergence of IoT-based technologies promises its diverse composition, which includes organized data, unorganized
to create safer, more efficient, and sustainable food supply chains soon. data, and partially structured data of different kinds (Desai, 2018).
Through IoT, mobile phones, digital cameras, and sensors can efficiently Applying Big Data technologies (Table 3) and analytics is instrumental
collect and transmit data to centralized data infrastructures using Wi-Fi in extracting valuable insights, patterns, and trends from this extensive
or other data transfer channels. This facilitates real-time monitoring and and varied data pool.
control, enhancing overall operational efficiency (Al Rasyid, Mubarrok, In food safety, Big Data plays a pivotal role by providing the means to
& Hasim, 2020). collect, analyze, and interpret data from all stages of the food supply
An innovative IoT system has recently been designed to address food chain (Liu et al., 2018). Through Big Data analytics, patterns and
fraud by monitoring product adulterants. Their proposed system is anomalies within the data can be detected, which might signal potential
characterized as straightforward yet highly effective, making it acces­ food safety risks. For example, abrupt fluctuations in temperature during
sible to various stakeholders in the food supply chain, including farmers, transportation could result in spoilage or contamination. Early identi­
consumers, and regulatory authorities. The system incorporates a range fication of such issues allows for swift intervention and mitigation.
of sensors capable of detecting factors such as temperature, oil content, Through historical and real-time data analysis, predictive models can
humidity, salt levels, metal contamination, color variations, pH levels, forecast potential food safety hazards (Jin et al., 2020).
and viscosity, all of which contribute to identifying potential food As a result, current methods for early warning and emerging risk
adulteration (Song, Jiang, Wang, & Vincent, 2020). identification in food safety extend beyond traditional monitoring con­
ducted by food inspectors and companies. Innovative technologies such
2.3.1. Practical applications as precision agriculture, radiofrequency identification (RFID), and
The IoT technology emerges as a valuable tool for providing rapid wireless sensor networks are revolutionizing data collection in primary
information at points of care within the food industry. These technolo­ food production, enabling real-time monitoring of safety and quality
gies encompass various forms, such as lateral flow devices and RFID parameters. These data can then be used for the early warning systems
technology, enabling the qualitative, semiquantitative, and quantitative and emerging risk tools to identify and mitigate potential hazards.
detection of pathogens (e.g., Salmonella), toxins (e.g., mycotoxins), and
even bacteriophages, serving as indirect indicators for certain pathogens 2.4.1. Practical applications
(Witjaksono et al., 2018). Moreover, IoT-based biosensors can be inte­ In meat production and dairy chains, digital devices like RFID tags,
grated into packaging materials, constituting "Intelligent packaging," readers, and GPS tracking devices are crucial in enhancing surveillance
which not only enhances product safety but also acts as a compliance against food safety threats. For instance, biosensors deployed along the
tool for food safety management systems based on Hazard Analysis and production chain and on farms can detect specific pathogens utilizing

Table 3
Big Data applications in food safety.
Name Data Type Data Description Country Source

GEMS/Food Monitoring Data Monitoring Data collected through biological and chemical Global GEMS/Food (2021)
analyses
RASFF Food Alerts Alert Information Information sourced from the Rapid Alert System for European RASFF Portal (2021)
Food and Feed Union
CompTox Chemical Database Chemical A repository of chemical structures and specifications Global CompTox Chemical Dashboard (2021)
Specifications globally
European Food Consumption Consumption Data Information on food consumption across the EU EFSA Panel on Biological Hazards
Repository European Union (BIOHAZ) (2012)
ComBase Quantitative Quantitative parameters related to food microbiology USA ComBase (2021)
Microbiology in the USA
World Bank Open Data Country Information Extensive database containing information on Global The World Bank (2021)
countries worldwide
USDA National Nutrient Food Product Nutritional data for various food products in the USA USA Food Data Central (2021)
Repository Information
EU Pesticides Repository Pesticide Approval Repository of authorized pesticide listings EU Sante (2021)
FSANZ Food Standards Codes Legislative Documents Legal documents concerning food safety standards in AU & NZ Food Standards Code (2021)
AU & NZ
Chemspider Chemical Database Chemical Global chemical composition database Global Chemspider (2008)
Specifications

6
H.M.R. Abid et al. Food Control 162 (2024) 110471

techniques such as antibody-binding assays (Desai, 2018). Integrating validate the authenticity of high-value or specialty food items. For
these biosensors with smartphones offers a convenient and accessible example, it can affirm the legitimacy of organic, fair-trade, or premium
means of monitoring safety parameters. Additionally, camera surveil­ products by providing an immutable record of their distinctive attributes
lance in slaughterhouses provides a means to visually inspect carcasses (Casino et al., 2021).
for pathological conditions like lesions, further bolstering hazard Blockchain technology holds significant promise as a solution to
detection efforts. meet the stringent requirements of food traceability, including
By harnessing the capabilities of these technologies, stakeholders in comprehensive transparency and traceability, extending even to con­
the food industry can proactively monitor and address food safety risks sumers. Its ability to swiftly trace products and data in case of recalls
at various stages of production. This integrated approach improves early adds a layer of efficiency crucial for ensuring food safety and regulatory
warning capabilities and enhances overall food safety management, compliance.
ultimately safeguarding public health and ensuring the integrity of the However, the widespread adoption of blockchain systems in the food
food supply chain. industry faces several challenges that necessitate careful consideration.
Scalability emerges as a critical concern, as blockchain networks must
2.5. Blockchain technology accommodate large volumes of transactions while maintaining perfor­
mance. Additionally, ensuring robust security measures to safeguard
Blockchain technology represents a decentralized and distributed sensitive data, addressing privacy concerns, and managing storage ca­
digital ledger system designed to record and validate transactions across pacity is imperative to uphold the integrity of blockchain-based trace­
a network of computers. Its core purpose is to establish a secure and ability systems. It is essential to establish standardized protocols and
transparent environment for storing and sharing information without a ensure the accessibility of blockchain technologies to enable seamless
central authority (Zhang, 2020). Unlike conventional centralized sys­ participation of operators within diverse supply chain ecosystems.
tems, blockchain operates on a decentralized network of computers
known as nodes (Mounika, Anusha, Narayana, & Lakshmi, 2020). Every 3. Food safety hazards and associated WBTs
node maintains a complete copy of the entire blockchain, ensuring no
single entity holds sway over the entire system. Once a block joins the 3.1. Heavy metals
blockchain (Fig. 3), altering the data it contains becomes exceedingly
challenging (Gurtu & Johny, 2019). Pesticide and heavy metal residues pose significant challenges in
Blockchain technology holds significant promise, particularly in food ensuring the safety of plant-based foods. Contamination by heavy metals
safety and traceability. Blockchain can enhance the traceability of edible primarily occurs during the cultivation phase, significantly if the soil
items along the complete distribution network (Casino et al., 2021). and water sources are polluted (Ngabirano & Birungi, 2022). To mitigate
Every step in food production, processing, and distribution can be these risks, assessing soil heavy metal content, pesticide contamination,
meticulously recorded as transactions on the blockchain. This trans­ water quality, and air quality before planting is crucial. While pesticides
parency enables swift identification of a food item’s origin and facili­ are necessary for high-quality and high-yield crops, their improper use
tates tracking potential contamination or quality issues. With can result in harmful residues, impacting human health. The growth
blockchain, the provenance of a food product becomes verifiable. Con­ cycle of crops and the timing of harvest greatly influence the quality of
sumers gain access to comprehensive information about where a product plant-based foods (Camatari et al., 2017; Damalas & Khan, 2017).
originated, its processing journey, and its path to the store shelf. This Introducing blockchain technology offers a powerful solution to
fosters trust and serves as a deterrent against fraudulent activities (Xu, control the quality and safety of plant-based foods right from the source.
Li, Zeng, Cao, & Jiang, 2022). Moreover, blockchain can be harnessed to It can be applied to record and share critical agricultural data, including

Fig. 3. Image showcasing the transformative impact of block chain technology on plant food safety across various stages, from harvest to retailing and trans­
portation, benefiting both consumers and industry stakeholders.

7
H.M.R. Abid et al. Food Control 162 (2024) 110471

climate information, soil conditions, water quality, crop varieties, and 3.3. Food processing hazards
the supply of agricultural inputs such as pesticides, herbicides, and
fertilizers. This information is transmitted through various platforms Throughout the food supply chain, there is a risk of food products
and loaded onto the blockchain, which network nodes can verify and becoming contaminated by various safety hazards, including biological,
authenticate using a consensus mechanism (Zhang, 2020). In blockchain chemical, and physical threats. These hazards can compromise food and
technology, nodes first verify their legitimacy when a transaction enters feed safety, harming human and animal health (Kleter & Marvin, 2009).
the peer-to-peer network. The transaction is confirmed if there is a AI, IoT, ML, and blockchain technologies can collectively play a pivotal
consensus among nodes regarding its authenticity, and a new block is role in preventing these hazards during food processing. They offer
added to the blockchain, making the information immutable. Recording real-time monitoring, data analysis, traceability, and advanced
origin information in the blockchain gives consumers a deeper under­ decision-making capabilities (Kumar et al., 2021). IoT sensors provide
standing of the products they purchase. Moreover, transactions are live data on critical parameters, which AI and ML can analyze to detect
shared with all participants, enabling the detection of abnormal activ­ patterns and anomalies. Predictive models and adaptive control mech­
ities and automatic enforcement of conditions set in smart contracts. anisms allow swift responses to deviations from safety standards.
Thus, blockchain enhances the transparency and authenticity of food Blockchain technology ensures transparency and traceability, making
safety and quality information and instills greater consumer trust and identifying the source of contamination easier and enabling targeted
confidence (Köhler & Pizzol, 2020). product recalls. AI’s risk assessment capabilities and collaborative ef­
A new study has explored the application of ML models to predict forts are facilitated by blockchain. Together, these technologies estab­
heavy metal concentrations in water based on physicochemical param­ lish a robust system that enhances food safety, improves supply chain
eters. This research aimed to develop a reliable and interpretable model efficiency, and fosters consumer trust (Smith, 2018).
for forecasting heavy metal levels in groundwater, eliminating the need IoT systems have been designed to avoid the introduction of con­
for time-consuming laboratory tests. The researchers employed the taminants and the deterioration of food. These systems can monitor the
Random Forest model (RFM) and quick-measure parameters to predict overall food quality and be customized for specific commodities. It relies
groundwater arsenic, iron, and manganese concentrations. Their find­ on various sensors to detect indicators of food degradation, such as
ings demonstrated the effectiveness of this approach, emphasizing the temperature, moisture, and GPS locations (Popa et al., 2019). A
importance of considering low-cost feature predictability and explain­ cloud-based IoT system called Smart Wine was developed for monitoring
ability assessments during the modeling process (Sihag, Keshavarzi, & the wine supply chain. The primary goals of this system are resource
Kumar, 2019). So, these web-based technologies are useful tools for management (e.g., water and pesticides), disease prevention, and
detecting heavy metals in food samples with greater accuracy while quality enhancement (Popa et al., 2019). Similarly, a real-time remote
ensuring food safety. monitoring system has been developed. This system employs wireless
sensors to track temperature levels within the entire cold chain of a food
3.2. Pesticide residues processing system, encompassing food preparation and transportation
(Salunkhe & Nerkar, 2016).
Pesticide residues in food samples have emerged as a significant Chen and colleagues (Chen et al., 2014) proposed a smart cold chain
contemporary concern. In an era where food safety and environmental system to monitor food freshness. This system utilizes multiple sensors
sustainability are paramount, accumulating these chemical residues to record temperature, humidity, and pressure in cold chain products,
poses substantial risks. (Abubakar et al., 2020, pp. 29–42). The tradi­ along with a novel type of RFID. It ensures continuous monitoring of
tional methods for detecting pesticide residues in food samples can be food products using a mobile code, even when the backend system is
time-consuming, resource-intensive, and may have limitations in terms offline. This approach reduces the overall system load regarding storage,
of sensitivity and accuracy. As technology advances, integrating modern processing, and communication requirements since constant access to
tools and techniques like AI, IoT sensors, Big Data analytics, and ML has the backend system is unnecessary (Bibi, Guillaume, Gontard, & Sorli,
significantly improved the efficiency and effectiveness of pesticide 2017). Chakraborty et al. (2021) utilized the K-nearest neighbors (KNN)
detection in food samples. method to carry out classifications and forecasts regarding aflatoxin B1
A recent study presents a portable, highly sensitive, on-site system levels within maize kernels (Chakraborty et al., 2021).
for detecting organophosphorus compounds designed for IoT-enabled
food safety tracking. The proposed system is significant for food safety 4. Integration of web-based tools in food safety
tracking because it enables on-field detection of organophosphorus (OP)
compounds, which seriously threaten food safety and the environment. Food contamination can result from various safety risks occurring
The system is portable, has high sensitivity, low cost, and can be easily simultaneously. Knowing the presence of multiple food safety hazards
integrated into existing IoT-based food safety tracking systems (Bou­ can assist risk managers in prioritizing their efforts by addressing the
zembrak, Klüche, Gavai, & Marvin, 2019). Another research developed most significant hazards or targeting the primary contributors to these
an alternative pesticide residue detection system for agricultural prod­ occurrences (Van der Fels-Klerx et al., 2018). In existing research, food
ucts. Their approach integrated biosensors, wireless transmission, and a safety prediction is examined through two distinct approaches. One
dedicated detection apparatus driven by a single-chip microcomputer. center on forecasting general food safety hazards based on direct data,
The culmination of this system was a collaborative information-sharing such as information obtained from records related to food safety
platform that demonstrated its efficacy through successful tests on monitoring. The other approach involves predicting food safety in­
authentic samples (Yi Chen, Dou, Chang, & Fan, 2022). Scientists cidents using indirectly related data, such as consumer complaints and
combined deep learning with terahertz imaging methods to successfully reports, which may offer valuable insights into potential hazards (Jin
identify and visualize residues of Benzimidazole pesticides on Toona et al., 2020).
sinensis leaves. This breakthrough technique holds substantial promise
for enhancing the precision and efficiency of pesticide residue assess­ 4.1. Predicting food safety hazards
ment (Nie et al., 2021). Therefore, using advanced tools such as AI, ML,
and Big Data for pesticide residue detection is imperative in the current Bouzembrak and Marvin (2019) introduced a Bayesian Network
era due to their ability to revolutionize food safety practices. (BN) approach that utilizes agricultural, weather-related, and economic
elements as input parameters for forecasting the occurrence of food
safety risks in vegetables and fruits. The authors emphasized that this
methodology could assist risk managers in pinpointing the most

8
H.M.R. Abid et al. Food Control 162 (2024) 110471

influential factors and understanding the connections between the potential hazards (Chang et al., 2020).
relationship among these factors and the existence of food safety risks
(Bouzembrak & Marvin, 2019).Another research proposed a neural
4.2. Prediction of food fraud
network (NN) model that harnessed data from food safety inspections to
analyze and predict the emergence of food safety risks in sterilized milk,
The rise of food fraud, which involves deceptive practices like mixing
enabling early warning systems. The findings suggested that food in­
non-food-grade substances such as dioxin-containing oils and Sudan
spection data had the potential for predictive modeling, thereby
dyes into food products, is increasingly jeopardizing food safety and
furnishing risk managers with valuable tools for proactively ensuring
causing concerns among consumers (Wang, Wu, et al., 2022). ML ap­
food safety. This approach (Fig. 4) can establish the foundational data
plications in the context of food fraud primarily revolve around pre­
necessary for scientific guidance, thus fostering food safety and quality
dicting general categories of food fraud and identifying added materials
enhancements (Liu et al., 2022).
in food (Hong et al., 2017). ML-based BN model comprehensively ana­
Previous research endeavors involved the creation of a BN model
lyzes the key factors contributing to food fraud incidents. It provides
aiming to forecast the potential biological and harmful effects of both
valuable insights for risk managers to pinpoint the most influential
metal nanomaterials and metal oxide nanomaterials. The objective of
factors in food fraud, enhancing their ability to manage, mitigate, and
this model was to facilitate risk assessments related to human health by
prevent it and its associated risks (Soon, 2020).
classifying nanomaterials according to two critical attributes: hazard
Bouzembrak and Marvin (2016) constructed a BN model that relied
potential and biological effects. Additionally, it aimed to assess the
on data related to adulteration and fraud. This model can predict the
factors influencing these attributes (Bouzembrak & Marvin, 2019).
potential types of food fraud that may occur with specific imported
ML-based BN model has also been applied to forecast the types of food
products, considering variables such as the country of origin and prod­
safety hazards prevalent in herb and spice products imported into the
uct category. Such results can assist government agencies and the food
Netherlands. Additionally, their model determined which herb and spice
industry in direct enforcement efforts more effectively to combat food
products should receive priority for monitoring at various supply chain
fraud. Mithun et al. (2018) utilized deep learning, specifically NN, to
stages, spanning consumers, markets, border inspection points, and
classify bananas as naturally or artificially ripened based on spectral
suppliers (Wang, Bouzembrak, Lansink, & van der Fels-Klerx, 2022;
data. This innovative application demonstrates the versatility of ML
Yang, Wei, & Pei, 2019).
techniques in addressing various aspects of food quality and authenticity
Furthermore, the prediction and monitoring of food safety incidents
(Mithun et al., 2018). Chakraborty and his colleagues utilized the
were explored by analyzing information from websites, online media,
K-nearest neighbors (KNN) method to carry out classifications (Fig. 5)
emails, and other reports. ML algorithms then employed this data for
and forecasts regarding aflatoxin B1 levels within maize kernels
classification, identification, and forecasting of food safety events
(Chakraborty et al., 2021) (see Fig. 6).
(Magnus et al., 2021). Research indicates the utilization of text mining
Recent innovation introduced an innovative IoT-based system
and supervised ML methods to quickly scan online media to identify
tailored for detecting adulterants in food products, addressing a critical
reports on food safety hazards (Bernert et al., 2020). Some studies uti­
aspect of food fraud mitigation. Central to the system’s functionality was
lized text mining to analyze textual data extracted from daily reports
an array of sensors designed to monitor various parameters crucial for
and consumer complaints, using Naïve Bayes and SVM Classifiers to
identifying food adulterants. These sensors cover a spectrum of factors,
identify whether the Economic and Food Safety Authority was respon­
including temperature, oil content, humidity levels, salt concentration,
sible for the reported issues (Joung, Jung, Ko, & Kim, 2018). Chang et al.
metal presence, color consistency, pH levels, and viscosity. By
(2020) developed an automated alarm system for the safety of edible oil
leveraging these diverse sensor inputs, the system offers a comprehen­
by leveraging electronic receipts. These various approaches collectively
sive approach to detecting potential adulteration instances, thereby
contribute to advancing the field of food safety by harnessing data and
supporting food safety and integrity (Jiménez-Carvelo, Li, Erasmus,
advanced modeling techniques to predict, monitor, and mitigate
Wang, & van Ruth, 2022).

Fig. 4. Image illustrating the use of cutting-edge food safety methods, including vision-based and chemical composition techniques, to ensure the highest standards
in food quality and safety.

9
H.M.R. Abid et al. Food Control 162 (2024) 110471

Fig. 5. Addressing the triad of Big Data challenges, AI information processing, and Blockchain data security in the context of food safety.

Fig. 6. Utilizing web-based tools such as Allerginome, CompTox, FDIRisk, and the EU Pesticide Database to enhance food safety.

Similarly, another study discussed an IoT-based solution specifically adulteration incidents. By focusing on a specific food product and
for detecting adulterants in milk, a commonly adulterated food product. leveraging advanced sensor technologies, the findings offer targeted
The system employed a range of sensors, including gas sensors for solutions for combating milk adulteration, thereby contributing to the
detecting volatile compounds, temperature sensors for monitoring broader efforts aimed at ensuring food authenticity and safety
thermal characteristics, viscosity sensors for assessing fluid consistency, throughout the supply chain (Jiménez-Carvelo et al., 2022).
salinity sensors for detecting salt content, and RFID readers for trace­
ability purposes. This multi-sensor approach enables real-time moni­
toring and analysis of milk composition, facilitating timely detection of

10
H.M.R. Abid et al. Food Control 162 (2024) 110471

5. Contribution to the food industries discerning pertinent data within each source and establishing connec­
tions between different data sets. This complexity underscores the need
The most prominent use of these technologies is via internet/ether­ for advanced data analysis techniques to extract meaningful insights and
net RFID and wireless sensor networks (WSN). Food industries employ facilitate informed decision-making processes in food safety
these technologies in various ways, e.g., tracking and tracing the prod­ management.
uct’s authenticity in the entire food supply chain, monitoring the shelf- The evolution from traditional food supply chains to interconnected
life of the products, and controlling and mitigating counterfeiting. Fathi, food supply networks presents both opportunities and obstacles in
Karmakar, Bhattacharya, and Bhattacharya (2020) proposed employing ensuring food safety. International and national organizations curate
RFID tags and readers to monitor products along the entire supply chain, online databases that provide critical data on hazards, encompassing
from farm to retail. Similarly, Karuppuswami, Mondal, Kumar, and monitoring data, chemical analyses, consumption patterns, and reports
Chahal (2020) implemented RFID tags in an IoT pilot project to track on animal and plant diseases. Among these databases, notable examples
food throughout the supply chain. Data transmission in such initiatives include the Global Environment Monitoring System (GEMS/food),
commonly employs WSN, WiFi, and Ethernet technologies. Yu et al. FDAirisk, Combase, Comptox, EU Pesticide database, The Allergen
(2022) developed an IoT-based traceability system leveraging fog and Database, etc. GEMS/food, for instance, contains millions of global
cloud platforms, RFID tags, and electronic product codes (EPC). monitoring entries, facilitating efficient data retrieval and analysis. Such
Meanwhile, Taneja, Jalodia, Byabazaire, Davy, and Olariu (2019) repositories offer indispensable resources for food safety research,
devised a traceability platform specifically for tilapia aquaculture, enabling investigations into chemical properties, microbial growth
integrating RFID and EPC technologies alongside Object Name Service conditions, and environmental factors that influence food safety out­
(ONS) and EPC Information Service (EPCIS) servers, as well as a data­ comes (Qin et al., 2023).
base server. Further, Yan, Zou, Xie, Gao, and Zheng (2020) demon­
strated an IoT system in their study focusing on pork anti-counterfeiting 6.1. FDA-iRISK
and traceability. This system incorporates EPC, RFID, NFC (near-field
communication) technologies, and ZigBee in the Network layer, The FDA-iRISK web interface allows users to customize their explo­
enabling effective control and mitigation of counterfeit activities. ration by selecting food and hazard types, adjusting input variables,
Recently, alongside other handheld devices, researchers have ach­ updating references, and managing data, information, and risk sce­
ieved success in detecting various contaminants and pathogens through narios. The tool has four main tabs: Home, Risk Model, Report, and
mobile phones with the help IoT based sensors. These include measuring Repositories. Of these tabs, under the Risk Models, the user define ele­
mercury contamination in water, detecting Ochratoxin A contamination ments for creating a risk scenario. While the "Reports" tab generates
in beer, identifying allergens in diverse food products, and detecting summary reports for their models and scenario rankings. The FDA and
microbial contamination such as Escherichia coli in both water and food FSIS developed a sophisticated mathematical model to assess risks
samples. These techniques allow food industries to enhance operational associated with various food hazards. The model combined data from 23
efficiency and adhere more closely to established protocols (Lu, Shi, & ready-to-eat food categories and used three dose-response models to
Liu, 2019). predict risk levels within these categories. It also streamlined data
Adopting decision support systems (DSS) fueled by big data analytics storage, allowed for in-depth analysis, and facilitated comparisons of
revolutionizes decision-making within the food industry. By exploring health implications from microbial and chemical hazards in food. Yuhan
the vast amounts of data generated at each stage of the food supply and his team reported studies on Listeria monocytogenes and Salmonella,
chain, DSS enhances operational efficiency, minimizes waste, and op­ demonstrating iRISK’s application in assessing risks and intervention
timizes resource allocation. This data-driven approach improves food effectiveness for managing microbial hazards (Chen et al., 2013).
safety and enhances overall productivity and profitability for food in­ Additionally, the architecture of iRISK is based on a risk-ranking
dustry stakeholders. For instance, SemaGrow, a big data collection and prototype that bifurcates into two primary domains: exposure assess­
analytics system, has emerged to support farmers in making informed ment and hazard characterization (BIOHAZ, 2012). After FDAiRISK
decisions. This system employs advanced algorithms and tools to query public release, several studies used this tool to identify potential risk
extensive datasets and disparate data sources efficiently. It is developed hazards in food ingredients. An example is a work by Bevilacqua et al.
specifically for the agricultural sector, addressing various use cases by (2023), which used it for risk assessments in three case studies: lettuce
integrating vast and diverse spatio-temporal datasets within the context (Listeria monocytogenes), chicken salad (Escherichia coli), and fresh egg
of climate adaptation efforts (Deeken, Wiemann, & Hertzberg, 2019). pasta. Furthermore, the study compared the risk of chicken salad
Likewise, food industries can implement big data analytics methodolo­ contaminated by different pathogens (E. coli O157:H7, Campylobacter
gies to optimize their operations systematically. For instance, Wal-Mart spp., and Salmonella sp.) and critically evaluated the potential risk.
Stores Inc. has implemented the Sustainable Paperless Auditing and Moreover, Qin et al. (2023) predicted the aflatoxin risk in multiple
Record Keeping (SPARK) system, which automates data uploading, such consumable foods, and another study estimated the risk of C. perfringens
as food temperature, to a web-based record-keeping platform. This from consuming beef jerky, showcasing the tool’s robust features (Nam,
accumulation of data enables rapid identification of undercooked Kim, & Yoon, 2018). The FDA-iRISK model categorizes process types
chicken, demonstrating the efficacy of such analytics techniques in affecting hazard transformation in the food chain, enabling compre­
facilitating decision-making related to food safety protocols within the hensive food-hazard scenarios for risk assessment.
industry (Pham, 2018). Overall, the research’s contributions to the food This statistical output of -iRISK is achieved by advanced algorithms,
industry pave the way for a more resilient, efficient, and i.e., the Monte Carlo Simulation algorithm or Bayesian Benchmark Dose
consumer-centric food system that prioritizes safety, transparency, and Modelling, which provides a report with a food safety risk scenario. For
sustainability. instance, iRisk predicted an annual decline in disability adjustable life
years (DALYs) from 11.7 to 0.128 when storage intervention was
6. Quantitative approaches for predicting food safety by web- applied on soft-ripened cheese for consumer’s health by decreasing
based resources temperature from 13.0 ◦ C to 10.6 ◦ C thereby lessening the
L. monocytogenes from 5.79 to 3.42 log CFU. In another scenario, DALYs
The utilization of diverse data sources holds promise for enhancing further reduced to 0.00187 when a temperature variant of 8.2 ◦ C was
food safety protocols, with avenues such as online databases, sensor applied and substantially minimized bacterial count to 1.64 log CFU
data, genomics data, and social media data offering valuable insights. compared to 13.0 ◦ C. Such scenario generation helps food researchers
However, this wealth of information also presents challenges, notably in adjust product storage times to extend shelf life. Despite its role in

11
H.M.R. Abid et al. Food Control 162 (2024) 110471

determining disease-associated risk levels in various foods, the tool has in egg". Based on pH, NaCl, temperature, and time, this predictive model
limitations, including data complexity and predictive algorithms only provides predictions for uncured and cured meat in which pattern
focusing on specific hazards like L. monocytogenes and Salmonella raise LogCFU/g increases.
(Allende, Bover-Cid, & Fernández, 2022; Ji et al., 2024 ). The literature review highlights ComBase’s pivotal role in food
microbiology research. For example, Gaspar et al. (2021) used ComBase
6.2. EU pesticide database databases to illustrate microbial interactions and their food safety im­
plications. Researchers also studied the growth of L. monocytogenes on
The E.U. Pesticide Database is a comprehensive repository of active various fruits and vegetables (Boleratz & Oscar, 2022). Walker, Sun, and
and basic pesticide substances, highlighting low-risk and substitutable Thippareddi (2023) developed a ComBase-derived database for micro­
substances and their EU approval status by linking food products to bial growth dynamics. ComBase, while robust, may have geographical
active substances’ Maximum Residue Levels (MRLs) (Authority, 2015; biases, limiting its global microbial dynamics capture. Despite 62
Sanco, 2014). The platform allows users to specify pesticide residues, models, challenges persist in fully encompassing microbial behaviors,
distinguishing present and past Maximum Residue Level (MRL)(mg or especially with emerging food technologies. Conclusively, computations
μg/kg) applications to certain foodstuffs. The EU pesticide database harnessed in statistical analyses provide a strong pillar for quantitative
search for almonds reveals a 0.05 mg/kg limit for permethrin (including food safety analyses (Boleratz & Oscar, 2022; Gaspar et al., 2021; Kumar
isomer residues) and provides information on previous and current et al., 2024).
residue regulations. Research studies compare pesticide residue analyses
with EU database MRLs to assess risk thresholds. According to a study, 6.4. CompTox
45% of samples of commonly consumed vegetables were within or
below the MRL range, while 55% exceeded the limits. This linkage helps CompTox, which stands for Computational Toxicology, is a collec­
agriculture researchers and farmers identify active substances for plant tion of tools and integrated databases developed by the US Environ­
protection, safety, and environmental impacts (Almutiriy et al., 2024). mental Protection Agency (EPA). It is a comprehensive resource with
The EU Pesticides Database is a valuable resource for food biotech­ various types of available, open-source data, including high-throughput
nology, providing comprehensive data on plant protection products, screening, rapid exposure and dose, animal toxicity, chemistry, and
pesticide residues, and emergency authorizations in the EU Member virtual tissues. The updated compTox platform is now known as
States. The user interface, the "Active Substances" option, provides in­ "Computational Toxicology and Exposure Online Resources," which is
formation on low-risk or substitution candidates, basic substances, and hosting ChemExpo Knowledgebase, Cheminformatics Modules, Comp­
EU endorsement. Users can search for specific food products, "oranges," Tox Chemicals Dashboard, GenRA Tool, ECOTOX Knowledgebase,
which gave 514 pesticide residues with their MRL like 1,1-dichloro-2,2- ToxRefDB, and SeqAPASS Tool. These tools offer a wealth of chemistry,
bis(4-ethylphenyl) ethane (F) with 0.01* MRL (mg/kg), Azoxystrobin toxicity, and exposure data that is publicly accessible. Additionally,
with 15.0* MRL (mg/kg), Bromide ion with 30.0 MRL (mg/kg) indi­ more than 20 external sources like DSSTox, PubChem, and ChemSpider
cating the significance of the EU Pesticide Database in research en­ were included in the CompTox Chemicals Dashboard to provide a user-
deavors. Additionally, several studies have been conducted to identify friendly interface. The Toxicity Reference Database (ToxRefDB), con­
pesticides in tomatoes, cabbage, and lettuce, and also some studies to tains in vivo study data from over 5900 guideline or guideline-like
design nanoparticles to compare the pesticides effect on agricultural soil studies for over 1100 chemicals.
(Kolani, Mawussi, & Sanda, 2016; Musa, Buwono, Iman, Ayuning, & Similarly, the Aggregated Computational Toxicology Resource
Lusiana, 2019). Thus, the EU Pesticide Database is essential for re­ (ACToR) has >1000 worldwide public sources of environmental chem­
searchers and farmers alike, enabling them to make informed decisions ical data. Overall, the CompTox resource contains more than 266,804
about pesticide use and its impact on food safety and the environment. entries with food safety, 346,310 for Organic Food, 212,499 for Irradi­
ated Food, 245,017 for Quality Assurance, and 277,434 for toxic com­
6.3. ComBase pounds (with 122,053 Pesticides Product Label). The additional feature
of CompTox, is the addition of genetic and phenomics data associated
ComBase is a comprehensive database for food microbiology, offer­ with PFOS (Perfluorooctanesulfonic acid), a synthetic compound and a
ing the ComBase database and models to illustrate and estimate mi­ persistent organic pollutant linked to health and environmental issues
crobial behavior under various food conditions, including over 60,000 (Kim, Kim, Choi, & Park, 2023). This highlights the database’s capacity
records on environmental factors and food preservation from pathogens. to assist in uncovering complex biological pathways and their relevance
The database works on predictive modeling tools encompassing math­ in addressing environmental and health-related concerns (Grulke, Wil­
ematical equations that provide a base for statistical inferences liams, Thillanadarajah, & Richard, 2019; Panter et al., 2023). This
involving numerous tools, including logistic models, regression models, feature enhances research efficiency and raises a deeper understanding
and Gompertz and Baranyi models. Validation of these models is critical of chemical interactions, which is vital for environmental and pharma­
as this improves existing models’ worth and widens their scope in food ceutical studies (Martin et al., 2022). CompTox’s contribution is show­
matrices (Boleratz & Oscar, 2022; Gaspar, Alves, & Pinto, 2021; Kumar, cased by its utility in predicting potential health hazards and prioritizing
Ahire, & Taneja, 2024). chemical safety evaluations. Thus, it is an invaluable resource for re­
The ComBase interface has a user login and profile with multiple tabs searchers across multiple disciplines in advancing scientific knowledge
for different information. For example, the "Browser" tab offers several and addressing critical questions in toxicology, pharmacology, and
search options, such as organism (e.g., Bacillus cereus), food category environmental science.
(beef, bread, cheese, milk, juice, grain, poultry, water, sausage, etc.),
food name (apple, potatoes, etc.), conditions (anaerobic environment, 6.5. The Allergen Database
argon in the environment, CO2, acetic acid, sugar in the environment,
etc.), properties (acid tolerance response, induced inoculum, duration of Initially, the Biotechnology Information for Food Safety (BIFS)
indicated history, spore inoculation, temperature before time 0 of this database was established in 1998 to comprehensively list allergens from
record, etc.), temperature (− 76 to 120 ◦ C), Aw/NaCl (0.07–1 mM), pH diverse sources, which was then followed by databases like Allergen
(range 0.0–11.0), and author records. Online (http://www.allergenonline.org), The Structural Database of
The "Broth Model" tab includes three categories: growth, thermal Allergen Proteins (SDAP), The Allergen Database for Food Safety (ADFS)
inactivation, and non-thermal survival of microbes. The "Food Model" and The Allergome providing extensive insights into allergens, their
tab has two additional categories: "Perfringens Predictor" and "Salmonella origins, and interactions. (Brusic et al., 2003; Gendel, 2009; Mari, Rasi,

12
H.M.R. Abid et al. Food Control 162 (2024) 110471

Palazzo, & Scala, 2009). The databases mainly provide information on draft, Visualization, Formal analysis. Athar Hussain: Writing – original
allergenic molecules (allergens) causing IgE-mediated diseases draft, Visualization, Methodology, Formal analysis. Zainab Bintay
(anaphylaxis, asthma, atopic dermatitis, conjunctivitis, rhinitis, urti­ Anis: Writing – original draft, Visualization, Methodology. Muhammad
caria) selected from scientific journals and web-based resources. It in­ Nadeem: Writing – original draft, Data curation, Conceptualization.
cludes data on allergenic sources, regardless of the source (animals or Nauman Khalid: Writing – review & editing, Writing – original draft,
plants), tissues (dander, fruit, pollen, seed, spores, venoms, whole Visualization, Validation, Conceptualization.
bodies, etc.), or routes of exposure (contact, ingestion, inhalation, in­
jection, etc.) (Radauer & immunology, 2017). Each allergenic molecule
Declaration of competing interest
has a dedicated monograph (BLUE monographs), and the proliferation
of allergen-related food has emerged as a prominent concern (Poulsen,
The authors declare that they have no known competing financial
2004; Soon et al., 2020; Weir & Mykhalovskiy, 2010).
interests or personal relationships that could have appeared to influence
Different allergen databases contain varying types and amounts of
the work reported in this paper.
data. For example, Version 9.0 of Allergen Online includes 1386 se­
quences of known and putative allergenic proteins from 236 species. It is
Data availability
"well-curated," containing only protein sequences proven to be allergens
or from genetic source models associated with allergens (Radauer &
Data will be made available on request.
immunology, 2017). Similarly, the SDAP encompasses 737 allergens and
isoallergens, 829 protein sequences, and 22 IgE and IgG epitopes.
References
Furthermore, the ADFS currently contains 1340 sequences, 76 epitopes,
and 57 molecular three-dimensional structures derived from food, Abubakar, Y., Tijjani, H., Egbuna, C., Adetunji, C. O., Kala, S., Kryeziu, T. L., et al.
airborne, pathogen-related, and venom/salivary allergens. The Aller­ (2020). Pesticides, history, and classification Natural remedies for pest, disease and weed
gome database contains over 4400 molecules from bacteria to mammals, control. Elsevier.
Al Rasyid, M. U. H., Mubarrok, M. H., & Hasim, J. A. N. (2020). Implementation of
including parasite structures, isoforms, and B epitopes (Futamura et al.,
environmental monitoring based on KAA IoT platform. Bulletin of Electrical
2008). Engineering and Informatics, 9(6), 2578–2587.
For user convenience, Allergen Prediction Tools FAO/WHO (WHO Allende, A., Bover-Cid, S., & Fernández, P. S. (2022). Challenges and opportunities
related to the use of innovative modelling approaches and tools for microbiological
guideline-based similarity and identity to allergen), Motif-Based AI-
food safety management. Current Opinion in Food Science, 45, Article 100839.
derived search, Proteins (NCBI-BLAST based), and Epitopes (NCBI- Almutiriy, R. S., Alnajeebi, A. M., Elhalwagy, M. E., Thabet, O. A., Alenzi, F. K., &
BLAST based) are also hosted at ADFS. The motif-based method predicts Aljadani, M. M. (2024). Investigation of pesticide residues level on commonly
a query protein as allergenic if: 1) it matches an allergen motif with a consumed leafy vegetables picked from the central market in Jeddah, Saudi Arabia.
Environmental Science and Pollution Research, 31(4), 6232–6242.
normalized score (by pftool) greater than 8.5, or 2) it aligns to the 142 Authority, E. F. S. (2015). Combined review of the existing maximum residue levels
allergen sequences with high similarity by the BLAST algorithm (default (MRLs) for the active substances metalaxyl and metalaxyl-M. EFSA Journal, 13(4),
E-value, <10− 3) (Bianco, Ventura, Calvano, Losito, & Cataldi, 2022). 4076.
Ayyildiz, M., & Erdal, G. (2021). The relationship between carbon dioxide emission and
However, allergomics is still in its early stages, and a comprehensive crop and livestock production indexes: A dynamic common correlated effects
allergome for any allergenic source remains elusive. The design and approach. Environmental Science and Pollution Research, 28(1), 597–610.
maintenance of allergy databases pose financial and operational chal­ Bendre, S., Shinde, K., Kale, N., & Gilda, S. (2022). Artificial intelligence in food industry:
A current panorama. Asian Journal of Pharmacy and Technology, 12(3), 242–250.
lenges. Additionally, existing databases often present data inconsis­ Bernert, R. A., Hilberg, A. M., Melia, R., Kim, J. P., Shah, N. H., & Abnousi, F. (2020).
tently, emphasizing the need for standardization to facilitate Artificial intelligence and suicide prevention: A systematic review of machine
information exchange and accurate interpretation. learning investigations. International Journal of Environmental Research and Public
Health, 17(16), 5929.
Bevilacqua, A., De Santis, A., Sollazzo, G., Speranza, B., Racioppo, A., Sinigaglia, M.,
7. Conclusion et al. (2023). Microbiological risk assessment in foods: Background and tools, with a
focus on risk ranger. Foods, 12(7), 1483.
Bianco, M., Ventura, G., Calvano, C. D., Losito, I., & Cataldi, T. R. (2022). A new
Food safety issues are a matter of concern for everyone, so their
paradigm to search for allergenic proteins in novel foods by integrating proteomics
proper tracing and eradication are inevitable. Environmental toxins, analysis and in silico sequence homology prediction: Focus on spirulina and chlorella
pesticides, mycotoxins, and microbes seriously threaten consumers. In microalgae. Talanta, 240, Article 123188.
the current era, technology has overcome the issues and pinpoints of risk Bibi, F., Guillaume, C., Gontard, N., & Sorli, B. (2017). A review: RFID technology having
sensing aptitudes for food industry and their contribution to tracking and monitoring
before time and keeps records in real-time scenarios. Various food of food products. Trends in Food Science & Technology, 62, 91–103.
processing instruments and tools can be paired with software or the Boleratz, B. L., & Oscar, T. P. J. (2022). Use of ComBase data to develop an artificial
Internet to have real-time data processing track. Another helpful neural network model for nonthermal inactivation of Campylobacter jejuni in milk
and beef and evaluation of model performance and data completeness using the
approach is big data, an immense data stream that can tackle food safety acceptable prediction zones method. Journal of Food Safety, 42(4), Article e12983.
issues using machine learning or artificial intelligence networks. For Bouzembrak, Y., Klüche, M., Gavai, A., & Marvin, H. J. (2019). Internet of Things in food
detection purposes, convolutional neural networks and machine safety: Literature review and a bibliometric analysis. Trends in Food Science &
Technology, 94, 54–64.
learning models are at their peak, which can help to sort and grade food Bouzembrak, Y., & Marvin, H. J. (2019). Impact of drivers of change, including climatic
commodities under set standards. Various online web tools like Com­ factors, on the occurrence of chemical food safety hazards in fruits and vegetables: A
Base, Allerginome, and others are on the front line to work on food bayesian network approach. Food Control, 97, 67–76.
Brusic, V., Millot, M., Petrovsky, N., Gendel, S. M., Gigonzac, O., & Stelman, S. J. (2003).
safety. These engines can use various predictive qualitative and quan­ Allergen databases. Allergy, 58(11), 1093–1100.
titative analytics. Like ComBase, it has been compiled using multiple Camatari, F. O.d. S., Santana, L. C. L.d. A., Carnelossi, M. A. G., Alexandre, A. P. S.,
research and securing data from published platforms, making the actual Nunes, M. L., Goulart, M. O. F., et al. (2017). Impact of edible coatings based on
cassava starch and chitosan on the post-harvest shelf life of mango (Mangifera
prediction of food microbiological issues feasible. In the future, as
indica) ’Tommy Atkins’ fruits. Food Science and Technology, 38, 86–95.
computer processing-related fields are uprising, this can be a good Casino, F., Kanakaris, V., Dasaklis, T. K., Moschuris, S., Stachtiaris, S., Pagoni, M., et al.
approach, giving us a bilateral glance at the past and future of particular (2021). Blockchain-based food supply chain traceability: A case study in the dairy
prospective and retrospective data. sector. International Journal of Production Research, 59(19), 5758–5770.
Casson, A., Giovenzana, V., Tugnolo, A., Fiorindo, I., Beghi, R., & Guidetti, R. (2020).
Environmental impact of a new concept of food service: A case study for the re-use of
CRediT authorship contribution statement naval shipping containers. Journal of Cleaner Production, 274, Article 122912.
Chakraborty, S. K., Mahanti, N. K., Mansuri, S. M., Tripathi, M. K., Kotwaliwale, N., &
Jayas, D. S. (2021). Non-destructive classification and prediction of aflatoxin-B1
Hafiz Muhammad Rizwan Abid: Writing – original draft, Visuali­ concentration in maize kernels using Vis–NIR (400–1000 nm) hyperspectral
zation, Formal analysis, Data curation. Nimrah Khan: Writing – original imaging. Journal of Food Science and Technology, 58, 437–450.

13
H.M.R. Abid et al. Food Control 162 (2024) 110471

Chander, B. (2020). Deep learning network: Deep neural networks neural Networks for Kleter, G. A., & Marvin, H. J. (2009). Indicators of emerging hazards and risks to food
natural language processing. IGI Global, 1–30. safety. Food and Chemical Toxicology, 47(5), 1022–1039.
Chang, W.-T., Yeh, Y.-P., Wu, H.-Y., Lin, Y.-F., Dinh, T. S., & Lian, I.-b. (2020). An Knorr, D., & Augustin, M. (2021). Food processing needs, advantages and
automated alarm system for food safety by using electronic invoices. PLoS One, 15 misconceptions. Trends in Food Science & Technology, 108, 103–110.
(1), Article e0228035. Köhler, S., & Pizzol, M. (2020). Technology assessment of blockchain-based technologies
Chapman, B., & Gunter, C. (2018). Local food systems food safety concerns. Preharvest in the food supply chain. Journal of Cleaner Production, 269, Article 122193.
Food Safety, 249–260. Kolani, L., Mawussi, G., & Sanda, K. (2016). Assessment of organochlorine pesticide
Chen, Y., Dennis, S. B., Hartnett, E., Paoli, G., Pouillot, R., Ruthman, T., et al. (2013). residues in vegetable samples from some agricultural areas in Togo. American Journal
FDA-iRISK—a comparative risk assessment system for evaluating and ranking food- of Analytical Chemistry, 7(4), 332–341.
hazard pairs: Case studies on microbial hazards. Journal of Food Protection, 76(3), Kumar, V., Ahire, J. J., & Taneja, N. K. J. T. M. (2024). Advancing microbial food safety and
376–385. hazard analysis through predictive mathematical modeling, Article 100049.
Chen, Y., Dou, H., Chang, Q., & Fan, C. (2022). Prias: An intelligent analysis system for Kumar, I., Rawat, J., Mohd, N., & Husain, S. (2021). Opportunities of artificial
pesticide residue detection data and its application in food safety supervision. Foods, intelligence and machine learning in the food industry. Journal of Food Quality, 2021,
11(6), 780. 1–10.
Damalas, C. A., & Khan, M. (2017). Retracted: Pesticide use in vegetable crops in Pakistan: Liu, N., Bouzembrak, Y., Van den Bulk, L. M., Gavai, A., van den Heuvel, L. J., &
Insights through an ordered probit model. Elsevier. Marvin, H. J. (2022). Automated food safety early warning system in the dairy
D’Amore, G., Di Vaio, A., Balsalobre-Lorente, D., & Boccia, F. (2022). Artificial supply chain using machine learning. Food Control, 136, Article 108872.
intelligence in the water–energy–food model: A holistic approach towards Liu, G., Li, G., Yang, R., & Guo, L. (2018). Improving food safety in supply chain based on
sustainable development goals. Sustainability, 14(2), 867. big data. In Paper presented at the E3S web of conferences.
Datta, B. N., & Sahoo, B. (2021). Machine learning, regression and optimization. Data Liu, Z., & Chen, T. (2018). Automated binocular vision measurement of food dimensions
Science and SDGs: Challenges, Opportunities and Realities, 177–197. and volume for dietary evaluation. Computing in Science & Engineering. https://doi.
Deeken, H., Wiemann, T., & Hertzberg, J. (2019). A spatio-semantic approach to org/10.1109/MCSE.2018.243113429
reasoning about agricultural processes. Applied Intelligence, 49(11), 3821–3833. Lu, Y., Shi, Z., & Liu, Q. (2019). Smartphone-based biosensors for portable food
Desai, P. V. (2018). A survey on big data applications and challenges. In Paper presented evaluation. Current Opinion in Food Science, 28, 74–81.
at the 2018 second international conference on inventive communication and Magnus, I., Virte, M., Thienpont, H., & Smeesters, L. (2021). Combining optical
computational technologies (ICICCT). spectroscopy and machine learning to improve food classification. Food Control, 130,
Dobchev, D., & Karelson, M. (2016). Have artificial neural networks met expectations in Article 108342.
drug discovery as implemented in QSAR framework? Expert Opinion on Drug Mari, A., Rasi, C., Palazzo, P., & Scala, E. (2009). Allergen databases: Current status and
Discovery, 11(7), 627–639. perspectives. Current Allergy and Asthma Reports, 9(5), 376–383.
Dong, J., Wu, H., Zhou, D., Li, K., Zhang, Y., Ji, H., et al. (2021). Application of big data Martin, M. M., Baker, N. C., Boyes, W. K., Carstens, K. E., Culbreth, M. E., Gilbert, M. E.,
and artificial intelligence in COVID-19 prevention, diagnosis, treatment and et al. (2022). An expert-driven literature review of "negative" chemicals for
management decisions in China. Journal of Medical Systems, 45(9), 84. developmental neurotoxicity (DNT) in vitro assay evaluation. Neurotoxicology and
EFSA Panel on Biological Hazards (BIOHAZ). (2012). Scientific Opinion on the Teratology, Article 107117.
development of a risk ranking framework on biological hazards. EFSA Journal, 10(6), Mezgec, S., Eftimov, T., Bucher, T., & Seljak, B. K. (2019). Mixed deep learning and
2724. natural language processing method for fake-food image recognition and
Fathi, P., Karmakar, N. C., Bhattacharya, M., & Bhattacharya, S. (2020). Potential standardization to help automated dietary assessment. Public Health Nutrition, 22(7),
chipless RFID sensors for food packaging applications: A review. IEEE Sensors 1193–1202.
Journal, 20(17), 9618–9636. Mithun, B., Shinde, S., Bhavsar, K., Chowdhury, A., Mukhopadhyay, S., Gupta, K., …
Futamura, N., Totoki, Y., Toyoda, A., Igasaki, T., Nanjo, T., Seki, M., et al. (2008). Kimbahune, S. (2018). Non-destructive method to detect artificially ripened banana
Characterization of expressed sequence tags from a full-length enriched cDNA library of using hyperspectral sensing and RGB imaging. In Paper presented at the Sensing for
Cryptomeria japonica male strobili (Vol. 9, pp. 1–14). agriculture and food quality and safety X.
Gaspar, P. D., Alves, J., & Pinto, P. J. C. (2021). Simplified approach to predict food Miyazawa, T., Hiratsuka, Y., Toda, M., Hatakeyama, N., Ozawa, H., Abe, C., et al. (2022).
safety through the maximum specific bacterial growth rate as function of extrinsic Artificial intelligence in food science and nutrition: A narrative review. Nutrition
and intrinsic parameters. 5(2), 22. Reviews, 80(12), 2288–2300.
Gemechu, T., & Aliyo, A. (2024). Enteric bacterial infections, antimicrobial susceptibility Mounika, B., Anusha, P., Narayana, V. L., & Lakshmi, G. V. (2020). Use of blockchain
pattern, intestinal parasites, and associated factors among food handlers in yabelo technology in providing security during data sharing. Journal of Critical Reviews, 7
town, borena zone, Southern Ethiopia. Microbiology Insights, 17, Article (6), 338–343.
11786361231221717. Musa, M., Buwono, N. R., Iman, M. N., Ayuning, S. W., & Lusiana, E. D. (2019). Pesticides
Gendel, S. M. (2009). Allergen databases and allergen semantics. Regulatory Toxicology in kalisat river: Water and sediment assessment. Aquaculture, Aquarium, Conservation
and Pharmacology, 54(3), S7–S10. & Legislation, 12(5), 1806–1813.
Gomez, C., Chessa, S., Fleury, A., Roussos, G., & Preuveneers, D. (2019). Internet of Nam, G. W., Kim, S. J., & Yoon, K. S. (2018). Quantitative microbial risk assessment of
things for enabling smart environments: A technology-centric perspective. Journal of Clostridium perfringens in beef jerky. Korean Journal of Food Science and Technology,
Ambient Intelligence and Smart Environments, 11(1), 23–43. 50(6), 621–628.
Gong, Y., Xiao, Z., Tan, X., Sui, H., Xu, C., Duan, H., et al. (2019). Context-aware Ngabirano, H., & Birungi, G. (2022). Pesticide residues in vegetables produced in rural
convolutional neural network for object detection in VHR remote sensing imagery. south-western Uganda. Food Chemistry, 370, Article 130972.
IEEE Transactions on Geoscience and Remote Sensing, 58(1), 34–44. Nie, P., Qu, F., Lin, L., He, Y., Feng, X., Yang, L., et al. (2021). Trace identification and
Grulke, C. M., Williams, A. J., Thillanadarajah, I., & Richard, A. M. (2019). EPA’s DSSTox visualization of multiple benzimidazole pesticide residues on toona sinensis leaves
database: History of development of a curated chemistry resource supporting using terahertz imaging combined with deep learning. International Journal of
computational toxicology research. Computational Toxicology, 12, Article 100096. Molecular Sciences, 22(7), 3425.
Gurtu, A., & Johny, J. (2019). Potential of blockchain technology in supply chain Olaniyi, E. O., Adekunle, A. A., Odekuoye, T., & Khashman, A. (2017). Automatic system
management: A literature review. International Journal of Physical Distribution & for grading banana using GLCM texture feature extraction and neural network
Logistics Management, 49(9), 881–900. arbitrations. Journal of Food Process Engineering, 40(6), Article e12575.
Hong, E., Lee, S. Y., Jeong, J. Y., Park, J. M., Kim, B. H., Kwon, K., et al. (2017). Modern Pal, A., & Kant, K. (2018). IoT-based sensing and communications infrastructure for the
analytical methods for the detection of food fraud and adulteration by food category. fresh food supply chain. Computer, 51(2), 76–80.
Journal of the Science of Food and Agriculture, 97(12), 3877–3896. Panter, G. H., Brown, R. J., Jones, A., Körner, O., Lagadic, L., & Weltje, L. (2023).
Ji, X., Zhou, Y., Xiao, Y., Lyu, W., Wang, W., Shao, K., et al. (2024). A tiered approach of Detection of anti-androgenic activity of chemicals in fish studies: A data review.
hazard-prioritization and risk-ranking for chemical hazards in food commodities: Critical Reviews in Toxicology, 53(5), 326–338.
Application for selected mycotoxins (Vol. 178), Article 113946. Pariente, E. S., Cancilla, J. C., Wierzchos, K., & Torrecilla, J. S. (2018). On-site images
Jiménez-Carvelo, A. M., Li, P., Erasmus, S. W., Wang, H., & van Ruth, S. M. (2022). taken and processed to classify olives according to quality–The foundation of a high-
Spatial-temporal event analysis as a prospective approach for signalling emerging grade olive oil. Postharvest Biology and Technology, 140, 60–66.
food fraud-related anomalies in supply chains. Foods, 12(1), 61. Pham, H. (2018). The impact of blockchain technology on the improvement of food supply
Jin, C., Bouzembrak, Y., Zhou, J., Liang, Q., Van Den Bulk, L. M., Gavai, A., et al. (2020). chain management: Transparency and traceability: A case study of walmart and atria.
Big Data in food safety-A review. Current Opinion in Food Science, 36, 24–32. Piccialli, F., Casolla, G., Cuomo, S., Giampaolo, F., & Di Cola, V. S. (2019). Decision
Joung, J., Jung, K., Ko, S., & Kim, K. (2018). Customer complaints analysis using text making in IoT environment through unsupervised learning. IEEE Intelligent Systems,
mining and outcome-driven innovation method for market-oriented product 35(1), 27–35.
development. Sustainability, 11(1), 40. Popa, A., Hnatiuc, M., Paun, M., Geman, O., Hemanth, D. J., Dorcea, D., et al. (2019). An
Kalinowska, K., Wojnowski, W., & Tobiszewski, M. (2021). Smartphones as tools for intelligent IoT-based food quality monitoring approach using low-cost sensors.
equitable food quality assessment. Trends in Food Science & Technology, 111, Symmetry, 11(3), 374.
271–279. Poulsen, L. K. (2004). Allergy assessment of foods or ingredients derived from
Karuppuswami, S., Mondal, S., Kumar, D., & Chahal, P. (2020). RFID coupled passive biotechnology, gene-modified organisms, or novel foods. Molecular Nutrition & Food
digital ammonia sensor for quality control of packaged food. IEEE Sensors Journal, 20 Research, 48(6), 413–423.
(9), 4679–4687. Qian, C., Murphy, S., Orsi, R., & Wiedmann, M. (2023). How can A.I. help improve food
Kim, M., Kim, S. H., Choi, J. Y., & Park, Y. J. (2023). Investigating fatty liver disease- safety? Annual Review of Food Science and Technology, 14, 517–538.
associated adverse outcome pathways of perfluorooctane sulfonate using a systems Qin, M., Cheng, L., Li, Y., Tang, X., Gan, Y., Zhao, J., et al. (2023). Disease burden
toxicology approach. Food and Chemical Toxicology, Article 113781. contributed by dietary exposure to aflatoxins in a mountainous city in Southwest
China. Frontiers in Microbiology, 14.

14
H.M.R. Abid et al. Food Control 162 (2024) 110471

Radauer, C. J. I., immunology (2017). Navigating through the jungle of allergens: Van der Fels-Klerx, H., Van Asselt, E., Raley, M., Poulsen, M., Korsgaard, H.,
Features and applications of allergen databases. 173(1), 1–11. Bredsdorff, L., et al. (2018). Critical review of methods for risk ranking of food-
Rong, D., Xie, L., & Ying, Y. (2019). Computer vision detection of foreign objects in related hazards, based on risks for human health. Critical Reviews in Food Science and
walnuts using deep learning. Computers and Electronics in Agriculture, 162, Nutrition, 58(2), 178–193.
1001–1010. Vithu, P., & Moses, J. (2016). Machine vision system for food grain quality evaluation: A
Salunkhe, P., & Nerkar, R. (2016). IoT driven smart system for best cold chain review. Trends in Food Science & Technology, 56, 13–20.
application. In Paper presented at the 2016 international conference on global trends in Walker, L., Sun, S., & Thippareddi, H. (2023). Growth comparison and model validation
signal processing, information computing and communication (ICGTSPICC). for growth of Shiga toxin-producing Escherichia coli (STEC) in ground beef. LWT,
Sanco, D. (2014). E.U. pesticides database. MRLs updated on, 28(1), 2014. 182, Article 114823.
Septiarini, A., Sunyoto, A., Hamdani, H., Kasim, A. A., Utaminingrum, F., & Hatta, H. R. Wang, X., Bouzembrak, Y., Lansink, A. O., & van der Fels-Klerx, H. (2022). Application of
(2021). Machine vision for the maturity classification of oil palm fresh fruit bunches machine learning to the monitoring and prediction of food safety: A review.
based on color and texture features. Scientia Horticulturae, 286, Article 110245. Comprehensive Reviews in Food Science and Food Safety, 21(1), 416–434.
Sharma, D., & Sawant, S. D. (2017). Grain quality detection by using image processing Wang, Z., Wu, Z., Zou, M., Wen, X., Wang, Z., Li, Y., et al. (2022). A voting-based
for public distribution. In Paper presented at the 2017 international conference on ensemble deep learning method focused on multi-step prediction of food safety risk
intelligent computing and control systems (ICICCS). levels: Applications in hazard analysis of heavy metals in grain processing products.
Sihag, P., Keshavarzi, A., & Kumar, V. (2019). Comparison of different approaches for Foods, 11(6), 823.
modeling of heavy metal estimations. SN Applied Sciences, 1, 1–11. Weir, L., & Mykhalovskiy, E. (2010). Global public health vigilance: Creating a world on alert
Smith, M. J. (2018). Getting value from artificial intelligence in agriculture. Animal (Vol. 10). Routledge.
Production Science, 60(1), 46–54. Witjaksono, G., Saeed Rabih, A. A., Yahya, N.b., & Alva, S. (2018). IOT for agriculture:
Song, W., Jiang, N., Wang, H., & Vincent, J. (2020). Use of smartphone videos and Food quality and safety. In Paper presented at the IOP conference series: Materials
pattern recognition for food authentication. Sensors and Actuators B: Chemical, 304, science and engineering.
Article 127247. Xu, Y., Li, X., Zeng, X., Cao, J., & Jiang, W. (2022). Application of blockchain technology
Sood, S., & Singh, H. (2021). Computer vision and machine learning based approaches in food safety control： current trends and future prospects. Critical Reviews in Food
for food security: A review. Multimedia Tools and Applications, 80(18), 27973–27999. Science and Nutrition, 62(10), 2800–2819.
Soon, J. M. (2020). Application of bayesian network modelling to predict food fraud Yan, Y., Zou, Z., Xie, H., Gao, Y., & Zheng, L. (2020). An IoT-based anti-counterfeiting
products from China. Food Control, 114, Article 107232. system using visual features on Q.R. code. IEEE Internet of Things Journal, 8(8),
Soon, J. M., Brazier, A. K., & Wallace, C. A. (2020). Determining common contributory 6789–6799.
factors in food safety incidents–A review of global outbreaks and recalls 2008–2018. Yang, Y., Wei, L., & Pei, J. (2019). Application of Bayesian modelling to assess food
Trends in Food Science & Technology, 97, 76–87. quality & safety status and identify risky food in China market. Food Control, 100,
Stracener, C., Samelson, Q., Mackie, J., & Ihaza, M. (2019). The Internet of Things grows 111–116.
artificial intelligence and data sciences. I.T. Professional, 21(3), 55–62. Yu, Z., Jung, D., Park, S., Hu, Y., Huang, K., Rasco, B. A., et al. (2022). Smart traceability
Talari, G., Cummins, E., McNamara, C., & O’Brien, J. (2022). State of the art review of for food safety. Critical Reviews in Food Science and Nutrition, 62(4), 905–916.
Big Data and web-based Decision Support Systems (DSS) for food safety risk Zhang, D. (2020). The innovation research of contract farming financing mode under the
assessment with respect to climate change. Trends in Food Science & Technology, 126, block chain technology. Journal of Cleaner Production, 270, Article 122194.
192–204. Zhong, C., Reibman, A. R., Mina, H. A., & Deering, A. J. (2021). Designing a computer-
Taneja, M., Jalodia, N., Byabazaire, J., Davy, A., & Olariu, C. (2019). SmartHerd vision application: A case study for hand-hygiene assessment in an open-room
management: A microservices-based fog computing–assisted IoT platform towards environment. Journal of Imaging, 7(9), 170.
data-driven smart dairy farming. Software: Practice and Experience, 49(7), Zhou, Q., Zhang, H., & Wang, S. (2021). Artificial intelligence, big data, and blockchain
1055–1078. in food safety. International Journal of Food Engineering, 18(1), 1–14.

15