Journal:Food informatics: Review of the current state-of-the-art, revised definition, and classification into the research landscape

Full article title	Food informatics: Review of the current state-of-the-art, revised definition, and classification into the research landscape
Journal	Foods
Author(s)	Krupitzer, Christian; Stein, Anthony
Author affiliation(s)	University of Hohenheim
Primary contact	christian dot krupitzer at uni-hohenheim dot de
Editors	Jaiswal, Amit K.
Year published	2021
Volume and issue	10(11)
Article #	2889
DOI	10.3390/foods10112889
ISSN	2304-8158
Distribution license	Creative Commons Attribution 4.0 International
Website	https://www.mdpi.com/2304-8158/10/11/2889/htm
Download	https://www.mdpi.com/2304-8158/10/11/2889/pdf (PDF)

Abstract

Background: The increasing population of humans and their changing food consumption behavior, as well as the recent developments in the awareness for food sustainability, lead to new challenges for the production of food. Advances in the internet of things (IoT) and artificial intelligence (AI) technology, including machine learning and data analytics, might help to account for these challenges.

Scope and approach: Several research perspectives—among them precision agriculture, industrial IoT, internet of food, and smart health—already provide new opportunities through digitalization. In this paper, we review the current state-of-the-art of the mentioned concepts. An additional concept to address is food informatics, which so far is mostly recognized as a mainly data-driven approach to support the production of food. In this review paper, we propose and discuss a new perspective for the concept of food informatics as a supportive discipline that subsumes the incorporation of information technology, mainly IoT and AI, in order to support the variety of aspects tangential to the food production process and delineate it from other existing research streams in the domain.

Key Findings and conclusions: Many different concepts related to digitalization in food science overlap. Further, food informatics is vaguely defined. In this paper, we provide a clear definition of food informatics and delineate it from related concepts. We corroborate our new perspective on food informatics by presenting several case studies about how it can support food production (as well as the intermediate steps until its consumption), and further describe its integration with related concepts.

Keywords: food informatics, internet of things, precision agriculture, smart agriculture, internet of food, food computing, food supply chain management

Introduction

Scientists have been alerting the world about climate change for a very long time; examples include the World Scientists’ Warning to Humanity from 1992 and the more recent World Scientists’ Warning to Humanity: A Second Notice in 2017. However, it required Greta Thunberg and Fridays for Future to raise the awareness about climate change and the necessity to protect the environment and society.

One aspect that, on the one hand, impacts climate change but on the other hand is also highly influenced by it, is the production of food. Roughly 11% of the Earth’s population was unable to meet their dietary energy requirements between 2014 and 2016, representing approximately 795 million people. [1] On the contrary, food production for the population of industrial nations in particular highly contributes to climate change due to a meat-focused diet, with the expectation of seasonal fruits throughout the entire year, as well as a high waste of food. [2] Both situations will become more complex in the next decades as the global population is predicted to grow to 10 billion by 2050 according to the United Nations. [1] This might increase the number of people with insufficiently satisfied dietary energy requirements. The increasing welfare in emerging countries will likely lead to more people adopting the resource-demanding nutrition of the industrial nations.

Traditional food production approaches will not be able to deal with those issues sufficiently; hence, novel approaches are required. The integration of current research advances in the internet of things (IoT) seems especially promising in supporting various aspects of food production, including farming, supply chain management, monitoring, processing, or demand estimation. Whereas a commonly accepted definition of IoT is not present in the literature, it is agreed that IoT refers to connected computational resources and sensors which often supplement everyday objects. The sensors support the collection of data which can be analyzed for identifying changes in the environment, and the IoT system can react to accommodating those changes. Procedures from artificial intelligence (AI)—the idea that machines should be able to carry out tasks in a smart way—and machine learning (ML)—techniques for machines to learn from data—can complement the analysis and system controlling process in IoT systems. The actions of analyzing and controlling IoT systems are also named as a reason for adaptation. [3] The purposeful application of those methods can complement and optimize many existing processes. The research in this field is distributed across several domains, such as precision agriculture, smart farming, internet of food, food supply chain management, food authentication, industrial IoT (IIoT, or Industry 4.0) for food production, food safety and authentication, "food computing," or smart/pervasive health. Often, those concepts overlap and are not completely distinguished.

Another research stream can be recognized under the notion of food informatics, which is often understood as data-centric research for supporting food production and consumption. [4,5,6,7] However, research alone does not provide a clear concept of food informatics. In this review paper, we want to distinguish the various research streams related to the topics of food production and consumption. Further, we motivate our perspective on food informatics as a supportive research stream that can contribute to the wide field of applying IoT and AI/ML to optimize food production and, hence, can be seen as an underlying technological basement for the other information and communication technology (ICT)-related research streams that target aspects of the food supply chain. Additionally, we present several case studies related to the production of food, discuss how food informatics contributes to those applications, and highlight the relation to the other presented research streams. However, we do not aim at providing a full-fledged survey as this would be not possible for a broad coverage of topics. Accordingly, we target providing a systematic mapping [8] approach to offer a cross section of the research landscape.

In summary, our contributions are threefold:

• Delineation of concepts: We provide a delineation of various concepts related to the digitalization of food science production.
• Definition of "food informatics": We review the state-of-the-art in food informatics and motivate a new understanding of food informatics as a supportive discipline for food production and an underlying technical basement for digitalization.
• Application: We discuss the potential of IoT and AI/ML to support the process of food production and supply—in our understanding, the central role of food informatics—with regard to the socio-technical perspective of the various stakeholders.

The remainder of this paper is structured as follows: The next section compares research streams related to the production and consumption of food, and the subsequent section presents a new definition of food informatics. We then present several research perspectives, as well as research challenges, when applying ICT in the food production domain. The penultimate section discusses possible threats to the validity of our claims. Finally, we close with related work and our conclusions.

Delineation of concepts

The production of food is a highly complex process. On the one hand, there is a high diversity in the combination of ingredients and intermediaries with many dependencies, for example, in the order of processing. Further, byproducts, side-products, or co-products might arise, such as butter milk when producing butter. On the other hand, food has hygienic, olfactory, sensory, or preservation requirements. In general, the food production process can be divided into several phases:

• Agriculture: Production of ingredients/food
• Logistics': Transportation of food while obeying hygienic constraints
• Processing: Processing of ingredients into food products in an industrial process, while obeying hygienic constraints
• Retail: Selling of food
• Consumer use: Consumption by humans or animals
• Food waste handling: Handling and disposal of food waste in an intelligent and sustainable way (not in the scope of this paper)

In this paper, we see this process as a sequential process. However, in practice, a circular economy might be favorable from a sustainability viewpoint. Further, the mentioned byproducts, side-products, or co-products create a value-added network rather than a traditional value chain. However, in this paper we focus on how to support the different steps through informatics or ICT. Consequently, a sequential view on food production will not limit the validity of our arguments.

Take for example the different processing phases for the production of Spätzle, a German pasta (see Figure 1). Production starts with the planting and harvesting of wheat (crop cultivation) as well as the production of eggs (livestock production). Both ingredients are transported to the production facility, where Spätzle is produced by adding water and salt. Subsequently, the product is delivered to wholesale trades, food retail markets, or directly to the consumer/restaurants, where the product is eventually consumed. In all phases, IoT devices can be integrated to either support data collection or actively control the processes through adaptation. (Adaptation is essentially an adjustment of the production process to handle machine faults, use traffic forecasts to re-calculate routes, or adjust production plans to most any delay. Additionally, technology known from smart health research, such as wearables, can help observe the consumption behavior of consumers. Data collection and analysis is supported by cloud and edge technology. With cloud resources, we refer to flexible server resources that can be used to complement the often limited computational resources of production machines. Those can be in-house resources, shared by multiple factories, or external resources like Google Cloud, Amazon Web Services, or Microsoft Azure. Edge devices act as additional computational resources within a factory that extend the computational resources of production machines.

Figure 1. Overview on the different activities in the food supply chain using the example of Spätzle production.

Several concepts apply methods and technology from computer science, mainly from IoT and AI/ML, in order to support the food production process. Those concepts often address only one phase of the production process. In the following subsections, we discuss and compare the different concepts. The purpose is to delineate the different research streams rather than provide a detailed review of each of them.

Precision agriculture

Clearly, the first step in the food supply chain is comprised by the cultivation of crops, husbandry of livestock, and the overall management of farmland. Besides the actual operations and business aspects, which are usually summarized by the term "farming," the more general notion of "agriculture" refers to all the tangent scientific and technological aspirations around it. We therefore use the notion of agriculture as an umbrella term in this article.

The presence of variability and uncertainty inherent in many facets of agriculture has been recognized for many decades. [9] With this increasing awareness and a focus on the “field” (in the sense of farmland)—that is, recognizing that, for instance, soil and crops might exhibit varying conditions—combined with technological innovations such as global positioning systems (GPS), computational and information management systems, as well as the advent of autonomous systems and robotics into agricultural machinery, a subarea of agricultural sciences—namely precision agriculture—can be defined. With the focus on the cultivation land in mind, Gebbers and Adamchuk provide a concise definition of the term precision agriculture as "a way to apply the right treatment in the right place at the right time."[10]

They further specify and summarize the goals of precision agriculture as three-fold: (1) The optimization of required resources (e.g., the utilized amount of seeds and fertilizers) for obtaining at least the same amount and quality of crops in a more sustainable manner; (2) the alleviation of negative environmental impacts; and (3) improvements regarding the work environments and social aspects of farming in general. An alternative, intuitive definition is provided by Sundmaeker et al., who describe the field of precision agriculture as "the very precise monitoring, control and treatment of animals, crops or m² of land in order to manage spatial and temporal variability of soil, crop and animal factors." [11]

Smart agriculture

Advances over the last several decades in ICT—such as smart devices, cloud and edge computing, and near field communication (NFC)—as well as the resulting technological possibilities in nearly any branch of society and industry (summarized by the term "IoT"; see next subsection) naturally also opens up a wide variety of adoption scenarios for agriculture. Smart agriculture appears as the most common notion in that respect.

Wolfert et al. review the application of big data in the context of smart farming. Their survey further provides another concise definition of the term, as "a development that emphasizes the use of information and communication technology in the cyber-physical farm management cycle." [12]

We note that a new term is introduced in the above definition: "cyber-physical farm." As is often the case when new technologies are emerging, a variety of terms referring to essentially the same thing appear in the literature. Another terms that also shows up occasionally is "digital farming." For the sake of completeness, we want to highlight that the notion of digital farming or agriculture is sometimes also conveyed to mean the integrated and combined utilization of both precision and smart agriculture concepts. The interested reader is referred to a recent position paper of the Deutsche Landwirtschafts Gesellschaft (DLG) (the German Agricultural Society). [13] Since this article focuses on the notion of food informatics and not solely smart agriculture, we proceed without a further differentiation; throughout this work, we only differentiate between precision agriculture and smart agriculture, for the sake of simplicity. (However, note our differentiation of “E-Farming” or "Farming 4.0," with the German term being “Landwirtschaft 4.0,” the latter intended to relate to the German-coined notion of Industry 4.0.)

Industry 4.0 or industrial IoT

The vision of Industry 4.0 is to integrate cyberspace and the physical world through the digitization of production facilities and industrial products. [14] This synchronizes the physical world and a digital model of it, the so called "digital twin." The "industrial internet," also known as industrial IoT (IIoT), enables a flexible process control of an entire plant. [15] The current interpretation of the term appeared with the rise of cloud technologies. The central elements of both concepts—digital twin aside—are the smart factory, cyber-physical production systems, and an intelligent and fast communication infrastructure.

Food production may benefit from Industry 4.0 approaches. Predictive maintenance can lead to improved production efficiencies, especially as machine defects in the context of food production have a more serious impact due to the perishability of ingredients, in contrast to tangible product elements in the production area. Further, the flexibility of Industry 4.0 approaches can help to facilitate the production of individual, customized food articles. Luque et al. review the state-of-the-art of applying Industry 4.0 technology for the food sector and propose a framework for implementing Industry 4.0 for food production centered around the activities of the supply chain. [16]

Internet of food

The term "internet of food" was first used by Kouma and Liu in 2011. [17] They proposed to equip food items with IP address-like identifiers for continuous monitoring using technology known from the IoT. Hence, it is a combination of identifiers, hardware, and software to monitor food and allow an observation of the consumers for optimizing nutrition. Somewhat contrarily, other authors describe the use of IoT for food-related purposes rather than the identification aspect as the internet of food, an example being smart refrigerators. [18] Holden et al. [19] review current developments in the area of the internet of food with a focus on the support of sustainability.

Food computing

Min et al. [20] present a definition of the term "food computing" in combination with a review of the current state-of-the-art. According to them, food computing is concerned with the acquisition and analysis of food-related data from various sources, focusing on the perception, recognition, retrieval, recommendation, and monitoring of food. Hence, food computing is a consumer-focused research stream, with the objective being to support the consumer with respect to optimal nutrition. Data sources can include pictures taken with smartphones, and data from web sites or social media data. Accordingly, the research integrates approaches from information retrieval, picture recognition, recommendation systems, and prediction tools. For further information on the relevant approaches, the interested reader is referred to overviews on the current state-of-the-art. [20,21,22,23]

Smart health or pervasive health

According to Varshney [24], pervasive healthcare can be defined as "healthcare to anyone, anytime, and anywhere by removing locational, time, and other restraints while increasing both the coverage and the quality of healthcare."

In a similar fashion, authors define the research for smart health or mobile health. [25] Applications in those areas include health monitoring, intelligent emergency management systems, smart data access and analysis, and ubiquitous mobile telemedicine. Often, those applications rely on wearables—that is, small devices with sensors attached to the body of users—for data collection and signaling of critical health conditions. This requires efficient communication technology, smart IoT devices, and intelligent data analytics. Nutrition monitoring might be a relevant aspect in such health monitoring, as well as telemedicine. Vice versa, smart health apps might influence the consumption of food. [26] Additionally, somewhat related to this area, are newer works that target the field of (personalized) nutrition (e.g., smart food choices that support the choice for food of a consumer [27]), as well as nutrition informatics, which “describes approaches to understand the interactions between an organism and its nutritional environment via bioinformatics-based integration of nutrition study data sets.” [28]

Food supply and logistics

Supply chain management describes the optimization of the intra- and extralogistics. In the case of food production, this includes the transportation of ingredients to the production facility, the moving of ingredients and products in the facility, and final transportation to retailers or customers. In contrast to other tangible goods, food has specific requirements concerning temperature, hygiene, and storage, for example, avoiding pressure on the products. In the following, we focus on the extralogistics of food, its transportation outside of a production facility. Current approaches try to integrate IoT technology for monitoring of logistics, especially the monitoring of temperature and air quality. The application of RFID improves the tracking of food and furthers information management. [29] Currently, approaches propose to integrate blockchain technology into the food supply chain to guarantee traceability [30,31], i.e., food provenance. Introini et al. [32] provide an overview on traceability in the food supply chain.

Food safety and authentication

According to a recent overview by Danezis et al., "food authentication is the process that verifies that a food is in compliance with its label description." [33] Food authentication is one element of food safety, which comprises the monitoring and control of food to guarantee its quality throughout the value chain. Some authors present works that integrate IoT technology, mainly based on sensors for monitoring to achieve food safety. [34,35] Recent approaches propose integrating blockchain technologies to achieve a high reliability and availability of information. [30,31] This might help to increase the security of the stored information. However, one common issue for data-related analysis is the “garbage in, garbage out” (GIGO) principle, which says that the quality of the output of an analysis is determined by the quality of the input. GIGO is not solved by blockchain technology as it just acts as secure data storage.

Summary

The presented concepts share some similarities. First, they can be grouped along the mentioned phases of the food production process: agriculture, logistics, production, and consumption. For retail, we focus on the logistics part. Hence, we did not explicitly discuss the specifics of retail activities. Precision and smart agriculture is mainly concerned with the operational (and scientific) aspects of crop and livestock production as well as farmland husbandry and management. IIoT and internet of food approaches concentrate on supporting the production of food. The consumer-centered research domains of smart health and food computing target the optimization of food consumption behavior. The logistics aspects of food supply links the different phases of the process. Food authentication spans the whole process chain, as it provides a continual monitoring of food; however, it is limited to the activity of monitoring the process to guarantee the authenticity of the ingredients and products. Accordingly, those concepts provide customized mechanisms for specific tasks, though they are not generically applicable or reusable in several phases of the food production process.

Second, the presented research streams rely on advances in IoT (mainly on sensors for data collection) and AI (mostly autonomous robotics and ML). However, researchers mostly try to integrate or customize existing technology instead of developing new methodologies optimized for the requirements specific to food production. Furthermore, often the suggested technology is customized to specific purposes instead of providing more generic and flexible frameworks that can, with only minor adjustments, be used in several phases of the entire food production process.

Third, some research streams are related. Smart agriculture and precision agriculture both address agricultural processing and can be integrated to maximize their benefits. The internet of food research stream overlaps with food supply, as it addresses the monitoring of food. Further, as monitoring of food is an inevitable element for food authentication, internet of food is also related to food authentication and food safety. Lastly, food computing and smart health overlap in their purpose as well as some other methods, for example, data extraction from pictures captured with smartphones.

Consequently, we propose the development of generic approaches relying on IoT and AI that can support various process steps. This seems especially beneficial for data analytics procedures that analyze sensor data or forecast future system states, as those implement generic ML mechanisms. In the next section, we present how food informatics could step into the breach by means of proposing a new definition, which comprises our notion of the term.

Notes

This presentation is faithful to the original, with only a few minor changes to presentation and updates to spelling and grammar. In some cases important information was missing from the references, and that information was added. The authors did not provide citations for the "World Scientists'" papers; citations were added for this version.