Industrial Applications of Selected JFS Articles

The Journal of Food Science recently sent out a call for papers for a Special Issue on Artificial Intelligence in Food Science and Nutrition. This Special Issue should be compiled and posted sometime later this year. Artificial intelligence (AI) has many applications in the food industry, including food safety, product development, sensory evaluation, nutritional and health-related uses, and food security. In recent years, AI use in the food industry has attracted attention in many areas, including research into intelligence graphs, deep learning (DL), algorithms, language processing, development algorithms, machine learning (ML), and skilled systems. While AI has become a catch-all term in the media, in reality, there are differences between a Large Language Model (LLM) and Application-Specific AI (often called Narrow AI or Vertical AI), such as the ones for the food industry. The most obvious difference lies in what the AI is designed to do. LLMs (General Purpose) are trained on massive datasets covering almost every human topic. An LLM can write an essay or write Python code. They are flexible but can fail at application-specific tasks. Application-Specific AI (Narrow) is designed to excel at one specific task and nothing else. If an AI is controlling the temperature of a process, it uses deterministic logic or highly constrained statistical models that prioritize accuracy and repeatability over versatility. The main fields of AI in the food industry include robotics, expert systems, computer vision, DL, natural language processing, and ML. The development of AI in the food industry aims to save time, reduce labor, cut costs, increase production rates, improve the quality of final products, meet consumer demands, enhance nutritional value, create new food products, and boost overall food quantity. Enhancing food safety and traceability of food products in the food production industry is one of the major challenges. As a branch of AI, ML has been widely used to develop and refine algorithms and statistical models for food product development, sensory evaluation, and improving the quality and safety of final products. An example of AI in the food industry can be found in this month's highlights: “Enhancing Food Safety in the Cold Chain Through Internet of Things and Artificial Intelligence” by Wagner Augusto Müller, Sandro Binsfeld Ferreira, Suse Botelho da Silva (https://doi.org/10.1111/1750-3841.70871, the first article published for the special issue noted above) is a systematic review of 97 peer-reviewed studies (2010–2025) that highlight how the integration of Internet of Things (IoT) technologies and AI is reshaping food cold chain management. Traditional systems rely heavily on manual checks and fragmented temperature logging, often detecting problems only after product quality or safety has been compromised. In contrast, IoT-enabled sensors—tracking temperature, humidity, CO2, GPS location, and more—provide real-time visibility, while AI models convert those data streams into predictive insights that anticipate spoilage, contamination risk, and logistics disruptions. The review shows rapid growth in both IoT and AI adoption, driven by regulatory pressure, food waste concerns, and demand for traceability. Temperature sensors dominate current deployments, particularly in fruits, vegetables, meat, and seafood, but secondary sensors such as gas and humidity monitoring are increasingly used to capture early spoilage indicators. On the AI side, applications range from regression models for demand forecasting to advanced neural networks that predict product freshness, temperature deviations, or pathogen risk with high reported accuracy. Optimization algorithms are also being used to improve routing, reduce fuel use, and align delivery schedules with perishability profiles. The reviewers identify important gaps between laboratory performance and real-world deployment. Many AI models rely on limited or simulated datasets, raising concerns about generalizability under operational stress. Complex neural networks often lack interpretability, which complicates regulatory acceptance in food safety contexts where auditability is critical. IoT systems face their own challenges, including cybersecurity vulnerabilities, sensor drift, interoperability barriers, and the high upfront costs that can limit adoption among small and mid-sized operators. Looking ahead, the most promising direction is the development of closed-loop systems—where IoT sensors continuously feed data to AI models that automatically adjust refrigeration settings, reroute shipments, or trigger targeted interventions in real time. While most current implementations remain open-loop (monitoring only) or advisory (human action required), the review argues that integrating edge computing, explainable AI, and standardized data frameworks will be key to building resilient, transparent, and predictive cold chains. Additional highlights of studies published in the Journal of Food Science during the past month address many areas such as consumer acceptance, food security, food nutrition, food production sustainability, and food safety: Rice flour bread often suffers from low loaf volume and a firm, dense crumb because it lacks gluten's elastic network. In this study, researchers evaluated egg white (EW) as a functional improver in a rice flour–EW composite system designed to mimic breadmaking and determine why loaf quality changes. Thermal analysis showed that EW increased the system's thermal stability and helped suppress starch retrogradation—an important driver of staling and crumb firming in gluten-free breads. Rheology results indicated that EW also restrained starch swelling and shifted viscoelastic behavior: at 9% (w/w) EW, both storage (Gʹ) and loss (Gʺ) moduli reached their lowest values, while creep compliances (Jm and Jmax) peaked, reflecting a more fluid system with a weakened rigid structure. In practical terms, EW acted largely as an “inert filler” that reduced stiffness while changing how the dough deforms under stress. Microscopic interpretation suggested EW proteins occupy interstitial spaces between starch granules, improving dough continuity and, crucially, boosting gas retention during proofing/baking. That translated into better finished bread performance, including higher specific volume, lower hardness, and a more uniform crumb structure—suggesting EW as a targeted strategy for improving gluten-free rice bread quality. https://doi.org/10.1111/1750-3841.70666 These researchers affiliated with Kansas State University in Manhattan, Kansas, USA, examined how ethnocentrism influences Indians’ willingness to buy imported foods—and what price premiums shoppers assign based on country of origin. Using an online survey of 630 consumers, researchers tested five commonly purchased items in India (bananas, milk, lentils, potatoes, and namkeen) sourced from 13 countries, including India, the United States, Australia, New Zealand, China, and others. Respondents reported both willingness to buy and the price they were willing to pay for each product-country pairing. Ethnocentric attitudes were measured using the generalized ethnocentrism scale, then analyzed via exploratory factor analysis and segmented using hierarchical and k-means clustering into high, moderate, and low ethnocentrism groups. Even with these differing levels, a clear pattern emerged: consumers collectively preferred domestically produced foods and were willing to pay the highest premium for “Made in India” options across categories. Imported products from developed countries—particularly Australia, the United States, and New Zealand—were viewed more favorably than most other foreign sources but still ranked behind domestic foods. China stood out as the least favorable origin among the 13 countries evaluated, indicating that country-of-origin perceptions can significantly shape acceptance even before quality or brand cues enter the picture. Demographics helped explain who holds stronger ethnocentric beliefs. Income had the largest influence, followed by education level, gender, and age. For food marketers, the findings point to a practical product segmentation playbook. Messaging should align with local-production cues for ethnocentric segments, while positioning select imported origins as “trusted, premium alternatives” for less ethnocentric or more globally oriented consumers. For policymakers, the results reinforce the strategic value of strengthening domestic supply and leveraging “Made in India” initiatives to match consumer preference with national production goals. https://doi.org/10.1111/1750-3841.70919 A new survey and lab analysis of industrial fruit-processing byproducts in Sri Lanka highlights a sizable—and largely underused—stream of nutrient- and bioactive-rich materials. Across four major processors, pineapple, mango, papaya, and passion fruit were identified as the most heavily processed fruits, generating high waste yields of roughly half to more than two-thirds of incoming fruit mass. These residues were primarily peels, seeds, and bagasse, underscoring just how much potential raw material is available for upcycling into value-added food and nutraceutical ingredients. When the wastes were dehydrated and milled into powders, proximate composition varied widely by fruit fraction. Protein ranged from as low as ∼0.5% to as high as ∼25.9% (highest in papaya seed), fat spanned ∼0.36% to ∼28.2% (notably high in some seeds), and crude fiber reached ∼13%–43%, indicating strong potential for fiber enrichment applications. Mineral screening showed meaningful levels of essential elements such as potassium, calcium, magnesium, zinc, and copper across multiple waste types, supporting a positioning as nutrition-forward ingredients when properly formulated. The bioactive profile was equally notable. Total phenolics ranged from about 1.85 to 9.47 mg GAE/g, while flavonoids ranged from ∼1.38 to 21.82 mg RE/g. Antioxidant measures also varied substantially, with total antioxidant capacity spanning ∼1.85–25.54 mg AAE/g and DPPH scavenging performance (IC50) indicating particularly strong activity in mango and papaya peels and mango seed kernels—consistent with their higher phenolic and flavonoid levels. In practical terms, these fractions look especially promising as natural antioxidant sources for functional foods, extracts, or ingredient blends. From an industry standpoint, the work suggests prioritizing high-volume streams (pineapple peel/bagasse; mango peel/seed) and targeting applications that fit their strengths—fiber enrichment, antioxidant fortification, and nutraceutical ingredient development. The study also flags a key commercialization consideration. While toxic metals were mostly below permissible limits, lead exceeded recommended thresholds in some samples, reinforcing the need for robust sourcing, monitoring, and mitigation protocols before scaling food-grade uses. https://doi.org/10.1111/1750-3841.70938 Whey permeate powders (WPPs) are widely used dairy coproduct ingredients, but their tendency to cake, brown, and lose flowability during storage remains a persistent industry challenge. In this comprehensive study, researchers analyzed three commercial WPPs to establish their key “water relationships”—linking composition, structure, water activity, and glass transition behavior—and to explain why these powders are inherently unstable under typical storage conditions. Microscopy and x-ray diffraction supported a two-phase structural model: WPP particles contain crystalline α-lactose monohydrate and a distinct amorphous phase composed of residual lactose, organic acids (notably lactic acid), peptides, and minerals. Two dominant morphologies were observed—(1) small lactose crystals embedded in a continuous amorphous matrix and (2) large lactose crystals coated with a thin amorphous layer. Thermal analysis revealed that the amorphous phase exhibits a very low glass transition temperature (Tg), even at low water activities, and is strongly plasticized by moisture. Critically, the powders as received (aw ≈ 0.18–0.24) were already in the rubbery state at room temperature (25°C), explaining their high susceptibility to caking and browning during transport and storage. Modified Guggenheim–Anderson–de Boer equation (GAB) and Gordon–Taylor models, accounting for crystal-bound water in lactose monohydrate, showed that simply drying WPPs further offers limited stabilization potential. Even fully dried powders remain close to or below practical Tg thresholds because of the inherently low Tg of the amorphous fraction. In conclusion, the researchers state that stability is dictated less by total moisture and more by the composition and Tg of the amorphous phase. Increasing the Tg of this phase—rather than pursuing more aggressive drying—emerges as a more effective strategy to mitigate caking and color formation. By establishing water vapor sorption behavior and Tg as functions of both water content and water activity, this work provides a practical framework for defining safe storage conditions and guiding future formulation and processing interventions to improve WPP functionality. https://doi.org/10.1111/1750-3841.70883 Shrinkage—defined as post-hardening volume loss—is a persistent defect in high-protein, high-overrun frozen desserts, particularly under temperature-cycled storage. In this study, researchers evaluated formulations made with 6% milk protein concentrate (MPC), sodium caseinate (NaCN), or whey protein isolate (WPI), with and without 0.15% mono- and diglycerides (MDGs), at 100% and 150% overrun. By tracking ice crystal size, air cell size, fat destabilization, melting behavior, and volume loss over 6 weeks of temperature cycling, the researchers identified how microstructure and interfacial chemistry govern foam collapse and shrinkage. Across formulations, temperature cycling promoted ice recrystallization and air cell coarsening, but the magnitude depended strongly on protein source and MDG addition. MDGs significantly reduced ice crystal growth and, in MPC and WPI systems, increased fat destabilization by displacing proteins from the fat interface—creating a destabilized fat network that improved structural integrity. WPI with 0.15% MDGs and 150% overrun maintained the smallest ice crystals after 6 weeks and did not exceed the 50 µm sensory coarseness threshold. In contrast, NaCN systems showed minimal fat destabilization and different interfacial behavior, influencing both air stability and storage performance. Melting data indicated that fat destabilization was the strongest predictor of slower drip-through rates and better shape retention. Products with greater destabilized fat networks collapsed more slowly at room temperature, while those with weaker interfacial structures exhibited faster foam breakdown. However, shrinkage during frozen storage was not strongly predicted by ice crystal size alone. Instead, air phase destabilization—particularly gas diffusion, channel formation, and interfacial collapse—was the critical driver. Overall, the study demonstrates that shrinkage in high-protein frozen desserts is best understood as frozen foam collapse. Optimizing protein–emulsifier interactions, promoting controlled fat destabilization, and maintaining a structured serum phase are key formulation steps for manufacturers seeking to deliver high-protein, high-air products that resist recrystallization, meltdown, and volume loss during distribution. https://doi.org/10.1111/1750-3841.70895 Researchers evaluated a novel approach to fruit beer that combines Amazonian cupuassu (Theobroma grandiflorum) pulp with a commercial prolyl endopeptidase from Aspergillus niger (AN-PEP) to influence aroma chemistry and key beer quality parameters. Using a British golden ale base, they produced three treatments: AN-PEP added at the start of fermentation (A1), AN-PEP added at the start of maturation (A2), and a control without enzyme (A3), with cupuassu pulp added for all beers. Protein changes were screened by SDS-PAGE, volatiles were profiled by HS-SPME–GC–MS, and physicochemical metrics followed American Society of Brewing Chemists methods. Flavor volatile data pointed to predominantly floral and sweet aromatic notes, with prominent signals for compounds such as linalool and 2-phenylethanol, alongside higher alcohols like 3-methyl-1-butanol that can contribute whiskey/malty notes. In two of the three formulations, elevated levels of these key aroma-active compounds were observed, supporting the idea that proteolysis may increase precursor availability and/or promote release of bound volatiles, complementing cupuassu's naturally intense, acidic-aromatic character. From a commercialization standpoint, the beers landed squarely in the “sessionable” space, at roughly 2.8%–3.4% alcohol by volume (ABV), with an acidic profile driven by the fruit addition. For craft brewers, the study suggests a viable pathway to build differentiated fruit beers that pair regional ingredients with enzyme-assisted process steps to provide aroma and potentially reduce high–molecular weight proteins, which add value in both “innovation flavor” and “better-for-you” product marketing. https://doi.org/10.1111/1750-3841.70916 This study evaluated grape leaf powder (GLP) as a sustainable, plant-based fortifier for wheat bread, testing substitution levels from 0% to 6%. As GLP increased, bread color shifted noticeably: both and crumb and more reflecting the of the leaf These changes at higher rates, which be a for positioning but a potential for GLP levels produced with reduced volume and specific volume, alongside changes in dough and with higher while dough In the bread, crumb increased during when GLP exceeded a less crumb over time. Microscopy supported these performance a more and matrix as GLP with and in network development. the time, nutritional and functional and fat levels while total and antioxidant activity increased as GLP reinforcing potential as a natural functional For and product the practical is GLP use meaningful in antioxidant capacity and content without loaf quality. That GLP as a that can and “better-for-you” formulation and volume in the