The problem with traditional evaluation is structural rather than technical. Physicochemical analyses of pH, water-holding capacity and lipid oxidation, microbiological cultures that require days of incubation, and Warner–Bratzler shear force tests that destroy the very sample being judged are all accurate but slow, labor-intensive and fundamentally incompatible with a processing line moving thousands of carcasses per hour. An abnormal postmortem pH decline, for example, can produce pale, soft, exudative meat or dark, firm, dry meat, both of which damage appearance, processing yield and shelf life, yet conventional pH probes sample only tiny regions of a carcass. Water-holding capacity, which governs juiciness and drip loss, is measured destructively. Tenderness is assessed by cutting. By the time results arrive, the product has often already been shipped.

Nondestructive sensing technologies were the first step toward solving this, and the review catalogs them in detail. Near-infrared spectroscopy exploits the absorption of radiation by C–H, O–H and N–H molecular bonds, making it possible to estimate moisture, protein and fat content in seconds. Raman spectroscopy probes molecular vibrations to reveal protein secondary structure, lipid composition and water distribution in muscle, all of which correlate with tenderness and water-holding capacity. Hyperspectral imaging, the technology the authors treat as most promising, merges spectroscopy with photography: each pixel of an image carries hundreds of spectral bands, so a single scan can map pH, color, tenderness and intramuscular fat simultaneously across an entire cut. Electronic noses detect the volatile organic compounds emitted by growing bacteria and oxidizing lipids, while electronic tongues mimic taste perception to track flavor changes during storage.

What these sensors produce, however, is not an answer but a deluge. A hyperspectral image of a single beef steak can contain hundreds of thousands of spectral measurements, far too many for classical statistics to disentangle. This is where machine learning enters. Support vector machines construct optimal separating hyperplanes and, through kernel functions, capture nonlinear relationships in spectral data; they have been used to detect adulteration, differentiate meat species and predict microbial spoilage from electronic nose signals. Random forests build ensembles of decision trees and, crucially, can rank which variables matter most, a feature the review highlights for identifying influential quality markers in high-dimensional datasets. K-nearest neighbors offers a simpler instance-based approach widely applied to species identification and freshness classification, though it degrades as feature counts grow.

Deep learning pushes the capability further. Convolutional neural networks automatically extract spatial features from images using convolutional filters, and the review describes their use in automated carcass grading, marbling detection and authentication, including a study that identified duck meat adulteration in beef and lamb slices from digital photographs alone. Artificial neural networks, the older and simpler architecture, remain workhorses for predicting lipid oxidation and other physicochemical properties from spectral inputs. Hybrid strategies are increasingly favored: principal component analysis compresses hyperspectral data before modeling, and deep networks serve as feature extractors feeding into support vector machines or random forests. Notably, the review cautions that fancier is not always better, citing cases where partial least squares regression outperformed deep learning on hyperspectral data depending on how features were extracted.

The applications described read like a quality-control wish list being checked off in real time. Hyperspectral imaging paired with regression models has predicted tenderness in lamb and beef nondestructively, sidestepping the destructive shear force test, though the authors note that postmortem muscle transformation remains biochemically complex enough to keep prediction challenging. XGBoost models analyzing carcass images can estimate marbling scores and classify quality grades with accuracy rivaling trained inspectors, removing a major source of subjectivity from grading. Electronic noses coupled with support vector machines have classified freshness levels and estimated remaining shelf life from spoilage volatiles, and when such models incorporate temperature and humidity data through Internet of Things networks, shelf life becomes a dynamic, supply-chain-wide prediction rather than a static label date.

Food safety and authenticity emerge as perhaps the most consequential applications. Conventional microbiological testing requires incubation periods of hours to days, useless for rapid decisions on a processing line. Hyperspectral imaging combined with machine learning classifiers can detect the biochemical signatures of microbial growth, and electronic noses have identified spoilage-related volatile compounds produced by bacteria such as Pseudomonas and Brochothrix thermosphacta. On the fraud front, Fourier-transform near-infrared spectroscopy with chemometric analysis has accurately detected turkey adulteration in minced beef, and hyperspectral imaging has differentiated meat species, offering rapid alternatives to DNA-based methods. The review is careful to note the limits: microbial signals are hard to separate from the intrinsic variability of the meat itself, and authentication models are only as reliable as the reference datasets behind them.

The industrial vision extends beyond the laboratory into what the authors frame as Industry 4.0 for meat. Smart processing plants would integrate sensor networks, computer vision and machine learning to monitor quality continuously, adjusting chilling, aging and packaging parameters in real time. Digital twin technology, still nascent in the food sector, would create virtual replicas of physical processing systems fed by live sensor data, allowing a plant to simulate, for instance, how a temperature fluctuation during storage will accelerate microbial growth and intervene before quality deteriorates. The review also points toward generative artificial intelligence, which could simulate processing conditions or design novel product formulations with optimized sensory and nutritional profiles, including alternative protein systems, though it acknowledges this frontier remains in its infancy.

What keeps the review from being pure boosterism is its frank accounting of the obstacles. Meat is biologically heterogeneous, and most published models are trained on small laboratory datasets that fail to capture the noise of a real plant, so robustness suffers. High-dimensional sensor data demand intensive preprocessing, and mishandled preprocessing introduces bias. Deep learning’s black-box nature clashes with regulatory contexts where transparency about contamination or adulteration decisions is mandatory, which is why the authors call for explainable AI tools such as SHAP and LIME and for hybrid models that balance predictive power with statistical accountability. Infrastructure costs, sensor calibration, data interoperability and workforce training round out the practical barriers. The authors’ conclusion is measured but ambitious: machine learning is already transformative for predicting tenderness, marbling, shelf life and microbial safety, but its full promise depends on standardized datasets, interpretable models and the digital plumbing of next-generation smart factories. If those pieces fall into place, the steak on your plate may one day carry a quality verdict rendered not by a lab technician with a knife, but by an algorithm that saw it coming weeks earlier.

Subject of Research: Machine learning applications for prediction and quality control of meat and meat products

Article Title: Machine learning–based prediction and quality control of meat and meat products: applications for intelligent quality management in the meat industry

Article References: Kim, Y.-J., Kwon, H. C., Cha, J. Y., Lee, S., Kim, T.-K., Kang, M.-C., Park, M. K., & Choi, Y.-S. (2026). Machine learning–based prediction and quality control of meat and meat products: applications for intelligent quality management in the meat industry. Food Science of Animal Resources, 46(1), Article 85. https://doi.org/10.1007/s44463-026-00085-6

Image Credits: AI Generated

DOI: 10.1007/s44463-026-00085-6

Keywords: machine learning, meat quality, hyperspectral imaging, food safety, spectroscopy, electronic nose, deep learning, shelf life prediction, food authenticity, digital twin, Industry 4.0, meat industry

The research team, led by Hee-Jin Kim and Hye-Seong Lee of Ewha Womans University in Seoul, took on a challenge that has long frustrated dairy scientists: commercial products differ not only in composition but also in raw material background, geographical origin, manufacturer, processing conditions, packaging, and storage history. Because these factors co-occur in real products rather than being independently manipulated in a laboratory, the team deliberately framed their work as an exploratory comparison of commercial product systems rather than a controlled experiment isolating the effect of any one ingredient. They selected eight market-available UHT milk products spanning skim, low-fat, and full-fat variants across three feeding-related product-system categories, identified as pasture-based/mixed-feeding, fermented-silage/mixed-feeding, and total mixed ration, or TMR, systems.

The analytical arsenal deployed was formidable. Physicochemical measurements captured basic composition with a MilkoScan FT2 instrument. Laser diffraction characterized the size distribution of fat droplets, which in milk, an oil-in-water emulsion, governs how light scatters and how the liquid flows. A spectrophotometer quantified color under standardized D65 illumination. A shear-rate-controlled rheometer measured apparent viscosity, and, in a particularly elegant touch, the same instrument was fitted with a ring-on-plate configuration lined with surgical tape to mimic the roughness of oral surfaces, allowing the researchers to measure friction coefficients at body temperature and generate Stribeck curves across boundary, mixed, and elastohydrodynamic lubrication regimes.

Complementing the physical measurements, electronic nose and electronic tongue systems generated instrumental fingerprints of aroma and taste variation. The electronic nose, equipped with fast gas chromatography columns, captured volatile profiles from the headspace of warmed samples, and a statistical screening procedure retained 24 peaks that were well represented by the major principal components. The electronic tongue, with its array of seven lipid-membrane and artificial-receptor sensors, distinguished the products with striking clarity: its first two principal components together explained nearly 94 percent of the variance in taste-related responses. The researchers were careful to note that these sensor systems do not directly measure human perception; they served as pattern-level evidence of product differentiation rather than as proxies for the tongue and nose of a drinker.

The human element came from two panels. Nine trained dairy panelists, each with more than three years of experience, rated the eight products on a battery of appearance, aroma, taste, flavor, mouthfeel, and aftertaste attributes using a nine-point intensity scale, following ISO 13299 guidelines with Williams Latin square presentation to balance order effects. Separately, 208 Korean adults aged 20 to 30, recruited at Ewha Womans University and screened for regular plain-milk consumption, rated overall liking on a nine-point hedonic scale under the same standardized room-temperature serving conditions, allowing direct alignment between expert perception and consumer preference.

The results painted a picture of coordinated, system-level differentiation. Within each product range, increasing fat content generally brought larger median particle diameters, greater lightness, higher apparent viscosity, lower friction in the mixed lubrication regime, and stronger perceived creaminess. Skim products had median droplet diameters around half a micrometer, while full-fat versions reached roughly 1.5 micrometers. Yet the pattern broke down in instructive ways. One full-fat product had a noticeably smaller particle size than its full-fat peers, and another showed lower friction despite comparable fat levels, evidence that lubrication behavior depends on particle-size distribution, interfacial composition, protein interactions, homogenization history, and heat treatment rather than fat alone. Color told a similar story: yellowness tracked the commercial product group more closely than fat level, with one product range appearing distinctly more yellow and another distinctly whiter.

Perhaps the most striking finding concerned creaminess itself. Among the three full-fat products, which differed significantly in particle size, viscosity, and friction, the trained panel detected no significant difference in perceived creaminess at all, with mean intensity scores clustering between 5.3 and 5.9. The authors interpret this as confirmation that creaminess is an integrated, emergent perception, one that the brain assembles from structural, rheological, tribological, and flavor cues rather than a direct readout of any single physical property. For product developers, this is a cautionary tale: engineering viscosity or droplet size in isolation may not deliver the mouthfeel consumers expect.

Consumer liking added another layer of nuance. The TMR full-fat product earned the highest liking score at 6.23, but a full-fat product from the pasture-based range remained among the least liked, scoring in the range of 3.59 to 4.14 for that product group. Fat content, in other words, did not uniformly buy favorability. A partial least squares regression model identified the sensory attributes driving preference: milky flavor, white color, and sweet taste were positively associated with liking, whereas hay-like flavor, yellow color, astringency, and salty taste pushed liking in the opposite direction. A follow-up model linking these liking-relevant sensory attributes to 34 instrumental predictors showed coordinated covariation across measurement domains, with whiteness and sweetness aligned with lightness, particle size, and viscosity, and hay-like, astringent, and salty notes oriented toward yellowness and selected electronic-sensor responses.

The statistical centerpiece of the study was a multiple factor analysis that integrated ten active data blocks comprising 59 variables, from sensory ratings and particle-size measures to color, rheology, tribology, and the electronic nose and tongue fingerprints. The first two dimensions of this integrated quality space accounted for over 82 percent of total variance and revealed two overlapping directions of differentiation: one a coordinated structural and mouthfeel axis linking droplet size, lightness, viscosity, creaminess, and oral coating, and a second flavor-appearance axis pitting hay-like, yellow, salty, and astringent notes against milky, white, and sweet ones. Because many variables loaded on both dimensions, the authors stress these represent overlapping system-level patterns, not independent causal pathways.

The team is candid about the limitations. Fat content, feeding system, origin, manufacturer, and processing history were confounded by design, so no single factor can be credited with the observed differences. Fatty acid analysis, which showed one pasture-range full-fat product with the highest omega-3 content and lowest omega-6 to omega-3 ratio, covered only the three full-fat products. The consumer panel was limited to younger Korean adults with an unequal sex distribution, and all samples were served at 24 degrees Celsius rather than refrigerator temperatures. Even so, the message for the dairy industry is clear and actionable: benchmarking and developing UHT milk demands integrated management of flavor, appearance, and oral-processing properties as a single quality system. In a global market where shelf-stable milk is a staple, the difference between a product consumers love and one they tolerate may hinge not on fat percentage printed on the label, but on the harmonious choreography of droplets, light, friction, and flavor reaching the mouth with every sip.

Subject of Research: Sensory, physical, and consumer-response quality differentiation of commercial ultra-high-temperature milk

Article Title: Integrated quality differentiation of commercial ultra-high-temperature milk: relationships among physical, sensory, and consumer-response characteristics

Article References: Integrated quality differentiation of commercial ultra-high-temperature milk: relationships among physical, sensory, and consumer-response characteristics. (n.d.). https://doi.org/10.1007/s10068-026-02311-9

Image Credits: AI Generated

DOI: 10.1007/s10068-026-02311-9

Keywords: UHT milk, sensory analysis, creaminess, tribology, rheology, consumer liking, electronic nose, electronic tongue, multiple factor analysis, dairy science, food emulsions, mouthfeel

Researchers led by Shanshan Han and Xiaoyu Wang set out to test whether Astragalus polysaccharides, bioactive macromolecules extracted from the medicinal herb Astragalus membranaceus, could reshape the quality profile of late lactation goat milk in ways that differ from conventional probiotic supplementation. Astragalus polysaccharides, commonly abbreviated APS, have attracted scientific attention for their stability, immunomodulatory properties, and potential metabolic effects. Previous work in heat-stressed dairy cows linked APS intervention to shifts in serum metabolites connected with glucose metabolism, amino acid metabolism, glutathione metabolism, and prolactin signaling, while separate studies reported improved oxidative stability in Cashmere goats receiving Astragalus supplementation. What remained unclear was whether APS could produce coordinated changes in the actual composition of goat milk, spanning everything from bulk nutritional traits to volatile fingerprints and molecular signatures, and whether those changes would look different from what probiotics achieve.

To find out, the team recruited thirty clinically healthy Saanen dairy goats from a commercial farm in Xi’an, China, all in late lactation at roughly 260 days in milk, with balanced parity and baseline milk yields of about 1.2 kilograms per day. Animals with a history of mastitis or recent antibiotic treatment were excluded. After a seven-day acclimatization period, the goats were randomly assigned to two groups of fifteen. One group received a basal total mixed ration top-dressed with five grams per head per day of a composite microbial agent containing Lactobacillus plantarum, Bacillus subtilis, and Saccharomyces cerevisiae. The other received the same ration supplemented with five grams per head per day of feed-grade Astragalus polysaccharide powder with 55 percent polysaccharide purity. The trial ran for thirty days, with milk samples collected at the start and end of the intervention under strict hygienic protocols, snap-frozen in liquid nitrogen, and coded so that analytical personnel remained blind to treatment assignments. All animal procedures followed ARRIVE guidelines and were approved by the Animal Ethics Committee of the Bio-Agriculture Institute of Shaanxi.

The analytical arsenal deployed on these samples was formidable. Conventional composition, including fat, protein, lactose, dry matter, milk urea nitrogen, and somatic cell count, was measured with a CombiFoss FT+ analyzer. Volatile-related sensory profiles were digitized using a PEN3-Plus electronic nose equipped with ten metal oxide semiconductor sensors covering sensitivities to aromatic compounds, nitrogen oxides, sulfur-containing compounds, alkanes, alcohols, and more. The proteome was quantified using data-independent acquisition mass spectrometry on an Orbitrap Astral instrument, with identification controlled at a one percent false discovery rate against the goat protein database. Untargeted metabolomics was performed on an Orbitrap Exploris 480 in both positive and negative ionization modes, while lipidomics employed a Q Exactive mass spectrometer with methyl tert-butyl ether extraction to capture the full lipid repertoire. Supervised multivariate models were rigorously validated with seven-fold cross-validation and two hundred permutation tests to guard against overfitting.

The conventional composition results immediately revealed two divergent response profiles. In the probiotic group, milk fat percentage increased significantly, accompanied by a parallel rise in total solids, suggesting the probiotic effect was largely driven by the fat fraction. Milk urea nitrogen, a widely used indicator of nitrogen utilization efficiency, rose markedly in the probiotic group, a pattern generally associated with less efficient nitrogen use. The APS group told a different story. Milk fat percentage fell significantly relative to baseline, total solids did not increase, and milk urea nitrogen dropped by 6.20 milligrams per deciliter, a reduction the authors interpret as a more favorable nitrogen-use profile. Lactose remained stable in both groups, indicating that the aqueous secretory characteristics of milk were preserved, and protein percentage showed a non-significant upward tendency of 0.42 percent under APS. Somatic cell counts, an indicator of udder health, did not change in either group, providing a stable background against which to interpret the other shifts.

The electronic nose data added a sensory dimension to this divergence. Radar plots of the ten sensor channels showed that the probiotic group’s volatile fingerprint remained broadly similar to its baseline, with only modest expansion. The APS group, by contrast, displayed a pronounced expansion of its sensor response polygon, particularly in the W5S, W2W, and W1S channels, which respond broadly to nitrogen oxide-related compounds, sulfur-containing and aromatic compounds, and methane-related or broad-range volatile classes. Principal component analysis confirmed that APS endpoint samples shifted clearly away from their baseline distribution and formed a distinct cluster, while probiotic endpoint samples stayed closer to the center. Loading analysis identified W5S as the dominant contributor to the separation, and the authors note the intriguing coincidence that lower milk urea nitrogen and a strong W5S response, both nitrogen-related indicators, changed in parallel under APS supplementation.

Proteomic profiling deepened the picture. APS endpoint samples separated clearly from their baselines in principal component space, whereas probiotic samples did not, indicating a stronger proteomic shift under APS. Functional enrichment analysis showed that proteins increased after APS supplementation were annotated mainly to starch and sucrose metabolism, aminoacyl-tRNA biosynthesis, glycolysis and gluconeogenesis, nucleotide metabolism, and sphingolipid signaling pathways, while decreased proteins mapped to cholesterol metabolism and various glycan biosynthesis and degradation pathways. A heatmap of the top fifty differentially abundant proteins revealed two opposing protein modules: a probiotic-associated module containing fibrinogen gamma chain, complement component 9, lipoprotein lipase, and other extracellular matrix and immune-related proteins, and an APS-associated module enriched in heat shock protein 90 beta, calreticulin, protein disulfide isomerases, calnexin, asparagine synthetase, and glycolytic enzymes such as lactate dehydrogenase A and glyceraldehyde-3-phosphate dehydrogenase.

Targeted inspection of lipid metabolism proteins sharpened the APS signature further. Compared with baseline, the APS group showed elevated abundance of acetyl-CoA carboxylase alpha, fatty acid synthase, and ATP citrate lyase, with fold changes of 1.49, 1.70, and 2.52 respectively, enzymes central to acetyl-CoA supply, malonyl-CoA formation, and fatty acid biosynthesis. NADPH-generating enzymes, including isocitrate dehydrogenase 1, glucose-6-phosphate dehydrogenase, and 6-phosphogluconate dehydrogenase, also rose, with the latter two increasing more than two-fold, providing the reducing equivalents that lipid biosynthesis demands. Meanwhile lipoprotein lipase abundance fell dramatically to 0.26 of baseline, suggesting a redistribution of lipid-processing capacity rather than a blanket increase. The probiotic group showed only mild changes in the same protein set, reinforcing the distinctness of the APS response.

The metabolomic and lipidomic layers completed the molecular portrait. Untargeted metabolomics identified 904 differential metabolites in the APS group versus 587 in the probiotic group, with lipid-related metabolites and organic acids, including tricarboxylic acid cycle intermediates such as citrate and malate, among the prominent responsive categories. Lipidomic profiling detected 2,553 lipid molecules spanning 42 subclasses, with triacylglycerols dominating at more than 88 percent of total lipid signal as expected for mammalian milk. Within that architecture, APS supplementation was associated with a 1.12-fold increase in total triacylglycerol signal, enrichment of medium-chain triacylglycerol species containing C6:0, C8:0, and C10:0 acyl groups, which are characteristic contributors to goat milk flavor after lipolysis, and a coordinated remodeling of polar lipids. Phosphatidylethanolamine and phosphatidylcholine, major components of the milk fat globule membrane, increased approximately 1.4-fold and 1.5-fold respectively, while ceramide rose nearly five-fold and sphingomyelin declined, an inverse pattern the authors interpret as sphingolipid compositional remodeling within the milk matrix.

Finally, an integrated correlation network linking proteins, lipids, metabolites, and electronic nose responses tied the layers together. Glucose-6-phosphate dehydrogenase and fatty acid synthase correlated strongly and positively with phosphatidylethanolamine- and ceramide-related lipid features, with correlation coefficients exceeding 0.85, while lipoprotein lipase was strongly negatively correlated with a specific ceramide species at r equal to minus 0.96. Strikingly, the ether-linked phosphatidylcholine species PC(18:0e/22:4) and an oxidized ceramide were positively correlated with W5S- and W2W-related sensor responses at r equal to 0.83, connecting milk polar lipid chemistry directly to volatile-related sensor signals. The authors are careful to frame these findings as association-based and hypothesis-generating rather than causal proof, noting the absence of an untreated control group, the use of milk rather than mammary tissue, and the sensor-level rather than compound-level nature of electronic nose data. Even so, the coherent multi-omics signature, spanning reduced milk urea nitrogen, a distinct volatile fingerprint, enriched lipid-synthesis proteins, and selective remodeling of medium-chain triacylglycerols, membrane phospholipids, and sphingolipids, positions Astragalus polysaccharide as a promising plant-derived, non-microbial nutritional strategy for stabilizing goat milk quality precisely when the industry needs it most.

Subject of Research: Multi-omics analysis of milk quality responses to Astragalus polysaccharide supplementation in late lactation dairy goats

Article Title: Milk-based multi-omics reveals distinct quality signatures associated with Astragalus polysaccharide supplementation in late lactation goats

Article References: Han, S., Wang, P., Hu, Y., Zhang, Q., Wan, K., & Wang, X. (2026). Milk-based multi-omics reveals distinct quality signatures associated with Astragalus polysaccharide supplementation in late lactation goats. Food Chemistry: X, 39, Article 104395. https://doi.org/10.1016/j.fochx.2026.104395

Image Credits: AI Generated

DOI: Not provided

Keywords: goat milk, Astragalus polysaccharide, late lactation, multi-omics, proteomics, metabolomics, lipidomics, electronic nose, milk urea nitrogen, milk fat globule membrane, probiotics, dairy quality

The team focused on three iconic Korean fermented soybean products: cheonggukjang, doenjang, and meju. Although all three begin with boiled soybeans, they diverge dramatically in their fermentation conditions, microbial communities, and resulting chemistry. Cheonggukjang is fermented briefly and with little salt, allowing Bacillus species to dominate. Doenjang undergoes prolonged maturation in a high-salt environment that reshapes its microbial ecology. Meju, the fermented soybean brick that serves as the foundation for both doenjang and soy sauce, is dominated by fungi such as Aspergillus species and is rich in the enzymes and metabolites those microbes generate. Because these pastes are known to contain bioactive compounds with antioxidant, anti-inflammatory, and immunomodulatory properties, the researchers reasoned that their extracts might act as functional starter ingredients rather than mere flavorings.

To test the idea, the scientists prepared extract solutions from commercially purchased cheonggukjang, doenjang, and meju by diluting each paste 1:100 in distilled water, then centrifuging and filtering the mixture. Each extract was inoculated at 0.5 percent by volume into 400 milliliters of milk cream. A control butter was fermented with a commercial starter containing Streptococcus thermophilus and Lactobacillus delbrueckii subsp. bulgaricus. All creams were fermented at 37 degrees Celsius for 24 hours, aged at 4 degrees Celsius for 12 hours, and then churned at 280 revolutions per minute for 15 minutes. The resulting butters were subjected to a battery of analyses covering pH, color, viscosity, moisture and fat content, microbial counts, electronic nose and electronic tongue profiling, and a sensory evaluation by a sixteen-member trained panel.

The pH results immediately revealed a fundamental biochemical divide. Butters fermented with the soy paste extracts had significantly higher pH values than the control, with the cheonggukjang sample highest of all. The explanation lies in the contrasting metabolisms of the microbial communities involved. Commercial lactic acid starter cultures flood the cream with lactic acid early in fermentation, driving pH down. The mixed communities drawn from traditional pastes, by contrast, include fungi and Bacillus species that decompose proteins and deaminate amino acids, releasing ammonia and other alkaline metabolites that push pH upward. The authors suggest this milder acidity may actually benefit the product, reducing sourness while allowing flavor-producing microbes to remain metabolically active, potentially enhancing both flavor quality and the delivery of probiotic organisms.

Physical properties told a reassuring story for manufacturers. Viscosity showed no significant differences among any of the butters, treated or control. Butter’s high-fat water-in-oil matrix provides substantial emulsification stability, and the small quantities of microbial metabolites generated during fermentation were simply not enough to alter its flow behavior. This means soy paste extracts can be incorporated without compromising the texture consumers expect. Color, however, did change: the meju and cheonggukjang butters were significantly more yellow than the control and the doenjang butter. Bacillus species abundant in these pastes produce peptides, free amino acids, and Maillard reaction products during fermentation, while molds and yeasts can promote browning reactions between reducing sugars and amino acids. The yellowness matters commercially, because previous research has shown that consumers associate a deeper golden hue in butter with higher purchase intention.

Microbial counts exposed the most striking differences. Total plate counts were significantly higher in all three extract-fermented butters than in the control, with cheonggukjang butter highest, reflecting its Bacillus-rich, low-salt, short-fermentation origin. Lactic acid bacteria counts were also elevated in all treated samples, but here meju butter took the lead. The researchers attribute this to Aspergillus oryzae, the fungus central to meju fermentation, whose powerful enzymes break proteins and carbohydrates into low-molecular-weight compounds that effectively feed lactic acid bacteria. Meju-derived communities also showed greater tolerance and adaptability to environmental stress than freeze-dried commercial strains. The doenjang-derived bacteria, adapted to high-salt conditions, grew more slowly in butter’s low-salt, high-fat environment, while cheonggukjang organisms, adapted to high water activity, were similarly constrained in the low-moisture product.

The flavor chemistry was mapped with an electronic nose, which identified elevated levels of volatile compounds including trimethylamine, ethanethiol, ethyl acetate, 2-methylbutanal, 3-methyl-1-butanol, and propyl acetate in the extract-fermented butters. Ethyl acetate, associated with buttery and fermented notes, was markedly higher in the meju butter. Principal component analysis of the aroma data achieved a discrimination index of 88, with the first principal component alone explaining 97.4 percent of the variance, cleanly separating the control from the treated samples and distinguishing the treated samples from one another. Notably, the sulfurous, rancid, and beany flavors often associated with fermented soybean products were not detected in the finished butter, suggesting the extracts can contribute desirable fermented aromas without importing off-flavors.

The electronic tongue added a taste dimension. Extract-fermented butters scored higher than the control in sourness, saltiness, and umami. The elevated sourness tracked with lactic acid bacteria counts, since more bacteria meant more organic acids. Saltiness was highest in the doenjang butter, consistent with residual salts carried over from its brine-fermented origin. Umami, measured against a monosodium glutamate reference, was highest in the meju butter, reflecting its greater content of glutamic acid, small peptides, and nucleotides released by the proteolytic activity of Aspergillus and Bacillus enzymes. Interestingly, the taste-based principal component analysis showed minimal differences among the three treated butters, indicating that despite their different origins, the pastes share overlapping microbial communities and metabolite profiles that converge in the finished dairy matrix.

The sensory panel delivered the verdict that matters most to consumers. The meju butter scored highest in flavor, taste, absence of off-flavor, and overall acceptability, and together with the cheonggukjang butter earned the top appearance scores, mirroring the color measurements. Panelists rated all extract-fermented butters higher than the control for texture attributes, likely because metabolites such as melanoidins, polyphenols, and peptides disperse within the fat matrix and moderate greasiness. The authors caution that their extracts contained both microorganisms and metabolites, so the observed effects reflect their combined action, and further microbiological work will be needed to separate the two. Still, the conclusion is clear: meju extract in particular can meaningfully improve the quality, microbial profile, and sensory appeal of fermented butter. As demand grows for dairy products that offer health benefits beyond basic nutrition, this study suggests that centuries-old Korean fermentation wisdom may have a place on the modern breakfast table, one golden, umami-rich pat at a time.

Subject of Research: Use of Korean traditional fermented soy paste extracts as starter ingredients to improve the quality and sensory properties of fermented butter

Article Title: Quality properties of fermented butter using extract solution from Korean traditional fermented pastes

Article References: Jeong, Y.-S., Yong, H. I., & Park, S.-Y. (2026). Quality properties of fermented butter using extract solution from Korean traditional fermented pastes. Food Science of Animal Resources, 46(1), Article 92. https://doi.org/10.1007/s44463-026-00089-2

Image Credits: AI Generated

DOI: 10.1007/s44463-026-00089-2

Keywords: fermented butter, Korean fermented soy paste, meju, doenjang, cheonggukjang, lactic acid bacteria, electronic nose, electronic tongue, sensory evaluation, food fermentation, dairy quality, umami

The research team, led by Fan Wu and Jiaolong Li of the Jiangsu Academy of Agricultural Sciences, set out to address a gap that has grown alongside China’s booming market for prepared dishes. Pork meatballs are among the most widely consumed pre-cooked meat products in the country, yet the effects of common reheating methods on their flavor quality remained poorly characterized. Because flavor is the primary driver of consumer acceptance and repeat purchase in the prepared-food sector, the researchers argued that understanding how reheating alters flavor chemistry is essential for optimizing industrial processes and household practice alike.

To do this, the team subjected pre-cooked pork meatballs to four reheating treatments: boiling, steaming, microwave heating, and a combined steam-microwave method. They then analyzed the resulting samples using a battery of instrumental techniques that together capture both aroma and taste. An electronic nose and an electronic tongue provided rapid, sensor-like fingerprints of overall flavor, while gas chromatography-mass spectrometry, or GC-MS, allowed the researchers to identify and quantify individual volatile compounds. Free amino acid analysis completed the picture by tracking the molecules most responsible for savory, umami taste.

The analytical effort paid off with an unusually comprehensive chemical inventory. Across the samples, the researchers identified a total of 105 volatile compounds, a roster that included aldehydes, esters, alcohols, ketones, and other classes of molecules that collectively define the smell of cooked meat. Critically, the act of reheating itself significantly increased the abundance of these volatile compounds compared with the un-reheated controls, confirming that the second heating pass is chemically active rather than merely a warming exercise. Among the compound classes, aldehydes and esters emerged as the predominant contributors to aroma formation, shaping the characteristic meaty and fruity-tinged notes that consumers associate with freshly reheated pork dishes.

Not all reheating methods pushed the chemistry in the same direction, however. Steaming produced the richest volatile profile of the four treatments, generating the most diverse and abundant array of aroma compounds. The researchers attribute this to the gentle, moisture-rich heat of steam, which promotes the formation of desirable aroma molecules without driving them off or degrading them through excessive thermal stress. Boiling, by contrast, involves direct immersion in hot water, which can leach water-soluble flavor precursors out of the meatball matrix and dilute the aromatic payload that reaches the nose.

Microwave reheating told a different story. While microwaves are prized for speed and convenience, the study found that this method promoted the formation of lipid oxidation-related compounds, most notably hexanal. Hexanal is a well-established marker of fat degradation in cooked meats and is closely associated with warmed-over flavor, the stale, cardboard-like off-note that develops when pre-cooked meat is stored and reheated. The rapid, uneven heating characteristic of microwave energy appears to accelerate oxidative reactions in the meatball’s fat fraction, generating compounds that consumers perceive as a loss of freshness even when the food is technically safe and hot.

The taste side of the analysis revealed equally meaningful differences. When the researchers measured free amino acids, the combined steam-microwave treatment stood out for maintaining higher levels of umami amino acids, the building blocks of savory taste that include glutamic acid and its relatives. The electronic tongue corroborated this finding, registering stronger umami and richness responses in the samples reheated by the combined method. This suggests that the gentler steam phase helps retain taste-active molecules that the aggressive, rapid heating of a microwave alone might degrade or drive off, while the microwave phase then brings the product quickly to serving temperature.

Taken together, the results position the steam-microwave combination as the most favorable reheating strategy among those tested. By enhancing desirable aroma compounds while simultaneously preserving favorable taste characteristics, the hybrid method produced what the researchers describe as a more balanced flavor profile. In practical terms, a consumer who steams meatballs briefly and then finishes them in the microwave gets the best of both worlds: the aromatic richness of steam heating and the speed and convenience of microwave energy, without the oxidative penalty that pure microwave reheating imposes on the fat in the meat.

The findings carry implications well beyond the home kitchen. The prepared-dish industry, which depends on cold-chain logistics and consumer reheating to complete the cooking process, now has instrumental evidence that reheating protocol should be treated as a formal part of product design rather than an afterthought. Manufacturers could specify recommended reheating methods on packaging to protect flavor quality, and product developers could reformulate meatball fat content or antioxidant systems to mitigate hexanal formation in microwave-dominant consumption scenarios. The study also adds to a growing body of literature showing that thermal processing method, not just ingredient quality, determines the final sensory outcome of meat products.

For the science of flavor, the study is a reminder that the last ninety seconds of a meal’s journey to the plate can matter as much as the recipe itself. With 105 volatile compounds shifting in abundance depending on nothing more than how heat was delivered, the humble meatball becomes a case study in how physical energy transfer shapes chemistry, and chemistry shapes pleasure. As pre-cooked foods continue to expand globally, the steam-then-microwave approach may prove to be one of the simplest, most actionable flavor-preserving interventions available to both industry and consumers.

Subject of Research: Effects of different reheating methods on volatile compound formation and flavor quality in pre-cooked pork meatballs

Article Title: Effect of different reheating methods on the formation of volatile compounds in pre-cooked meatballs

Article References: Wu, F., Li, N., Zhang, M., Li, P., Sun, C., Xu, W., Wang, D., & Li, J. (2026). Effect of different reheating methods on the formation of volatile compounds in pre-cooked meatballs. Food Science and Biotechnology. https://doi.org/10.1007/s10068-026-02247-0

Image Credits: AI Generated

DOI: 10.1007/s10068-026-02247-0

Keywords: pre-cooked meatballs, reheating methods, volatile compounds, steaming, microwave heating, lipid oxidation, hexanal, umami, free amino acids, electronic nose, GC-MS, food flavor chemistry

The research team, led by Si-Jia Lin and corresponding author Xu-Hui Huang of Dalian Polytechnic University in China, set out to identify the volatile compounds responsible for both the beloved and the objectionable notes in wakame, and then to track how those compounds respond to two commercial levers: progressive pickling and packaging material. China produced 206,100 tons of wakame in 2022, and global output has climbed steadily since 2012, so even small improvements in flavor retention carry substantial commercial weight. The team’s central question was whether the off-odors that plague the industry could be tamed through processing rather than through additives or breeding.

To do this, the researchers combined classical sensory evaluation with an arsenal of instrumental techniques. Gas chromatography-mass spectrometry revealed 55 volatile compounds in the samples, including 19 aldehydes, 10 alcohols, 6 ketones, 5 acids, and 4 esters. Aldehydes, with their characteristically low odor thresholds, emerged as the dominant contributors to wakame’s smell, accounting for 35.2 percent of the volatile profile, followed by alcohols at 16.6 percent and ketones at 13.1 percent. Using odor activity values, a metric that weighs a compound’s concentration against the concentration at which humans can detect it, the team narrowed the field to 18 decisive odorants. Among these were hexanal, (E)-2-nonenal, (E)-2-decenal, beta-cyclocitral, caryophyllene, and 1-octen-3-one, each imparting everything from grassy and oily notes to floral, fruity, earthy, and mushroom-like character.

The experimental design was rigorous. Ready-to-eat wakame supplied by a commercial producer in Liaoning was washed, desalted, blanched at 90 degrees Celsius for two minutes to protect color, and then subjected to two sequential seasoning steps: a first pickling at 10 degrees Celsius for two hours in a 4 percent salinity solution, and a second pickling for half an hour at 0.5 percent salinity. Samples were then sealed either in aluminum foil bags or in polyethylene terephthalate, or PET, casings, sterilized, and stored at 4 degrees Celsius. Six sample categories captured the full matrix of two packaging types across three processing stages. Thirty trained panelists scored smell, taste, color, and texture, while an electronic nose fitted with ten metal oxide semiconductor sensors and an electronic tongue capable of distinguishing 25 taste scales provided objective corroboration.

The sensory and colorimetric data told a nuanced story about salt. Proper pickling turned wakame greener, a effect the researchers attribute to chlorophyll forming more stable complexes with metal ions such as sodium from sodium chloride. Elevated salt also suppresses enzymatic activity and microbial growth by lowering water activity, slowing the oxidation reactions that generate stale aromas. But there is a tipping point. The second pickling round damaged pigment, darkened color, and softened texture, with hardness dropping significantly after the first seasoning and gumminess falling significantly after the second. The lesson, the authors suggest, is that moderate salinity preserves quality while excessive salinity destroys it, a balance producers must strike carefully.

On taste, the electronic tongue revealed that progressive pickling reduced bitterness, a trait traced to mannitol, algin, bitter amino acids, and bitter peptides in the seaweed. As salt levels rose, bitter amino acid content fell, and panelists’ umami, sweet, and salty scores climbed. The electronic nose, meanwhile, recorded rising responses from sensors sensitive to nitrogen oxides and sulfur compounds, reflecting the way salt disrupts cell structures and releases nitrogenous material. Panelists noted declining fishy, rancid, and overripe aromas as pickling progressed, and principal component analysis separated the samples almost perfectly, with cumulative contribution rates of 99.72 percent for odor and 99.66 percent for taste.

The quantitative volatile data were even more striking. After progressive pickling, the content of beta-cyclocitral, a signature algae-derived odorant, surged from 2.52 to 485.42 nanograms per gram, while caryophyllene, a pleasantly floral and fruity sesquiterpene, rose from 1.35 to 578.05 nanograms per gram. These increases enriched wakame’s characteristic marine aroma and helped mask off-notes. Simultaneously, the levels of undesirable compounds declined: 1-octen-3-one, an earthy, metallic ketone formed by oxidation of unsaturated fatty acids, dropped from 641.4 to 371.5 nanograms per gram in aluminum foil packages and from 725.66 to 459.08 nanograms per gram in PET, while fishy (E)-2-nonenal in foil-packaged wakame fell from 11 to 3.21 nanograms per gram.

Packaging proved to be the quiet hero of the study. Aluminum foil, with its low permeability to gases and water vapor and its complete opacity to light, preserved more of the volatile compounds that make wakame appealing while blocking the migration of oxidation-promoting substances into the food. Foil-packaged wakame retained higher hardness, stickiness, and resilience, showed greener color, and contained compounds such as (E,E)-2,4-heptadienal, 1-heptanol, and (E)-3-hexen-1-ol that were undetectable in the PET samples. Beta-cyclocitral levels in foil-packaged wakame reached 485.42 nanograms per gram compared with just 128.67 nanograms per gram in PET. In total, foil-packaged samples after double seasoning contained 16 key odorants against 14 in their PET counterparts, and panelists judged the foil samples more aromatic and less fishy.

To verify that these 18 compounds truly drive wakame’s aroma, the team built six aroma recombination models by adding the key odorants at their measured concentrations to an odorless wakame matrix, and then performed omission tests in which single compounds were removed and 15 panelists ran triangle tests to detect the difference. In freshly processed samples, removing hexanal, (E)-2-nonenal, (E)-2-decenal, or 1-octen-3-one produced significant to highly significant perception changes, confirming their importance. After pickling, beta-cyclocitral became the standout, while compound interactions grew so complex that individual omissions became harder to detect, a phenomenon the authors attribute to synergies among the expanding roster of odorants.

The practical implications are immediate for a global seaweed industry seeking to convert health-conscious consumers into repeat buyers. Choose packaging that blocks oxygen, moisture, and light; calibrate salt levels to stabilize pigment and suppress microbes without wrecking texture; and accept that progressive seasoning, done judiciously, can flip wakame’s chemistry from off-odor generator to flavor enhancer. As seaweed moves from niche health food to mainstream sustainable protein source, studies like this one show that flavor is not an accident of nature but a controllable outcome of engineering, one aluminum foil bag at a time.

Beyond the headline findings, the study’s methodology offers a window into how modern flavor science increasingly operates. By pairing trained human panels with electronic nose and electronic tongue instruments, the researchers followed a growing trend in food analysis sometimes called sensory omics, in which machine-based readings are calibrated against human perception to produce reproducible, quantifiable flavor fingerprints. This dual approach helps offset the subjectivity and fatigue inherent in panel work, since each taster in the study fasted for three hours beforehand and evaluated samples in isolated compartments under controlled lighting and temperature.

The findings also sit within a broader body of research on seaweed preservation. Earlier work on kelp showed that higher salt concentrations and lower temperatures reduce alginate lyase activity, an enzyme that degrades cell wall polysaccharides and accelerates quality loss. Comparable strategies, including the combination of natural antioxidants with fermentation to remove fishiness from fresh kelp, have been reported by other Chinese research groups, suggesting that salt-mediated enzyme inhibition is a recurring theme across brown algae processing.

Priorities for future work follow naturally from these results. Because plastic packaging can permit oxygen and moisture migration that drives oxidation, comparative studies of barrier materials, including glass, multilayer films, and foil laminates, may refine packaging recommendations further. Extending shelf-life trials beyond laboratory storage to real distribution conditions, and testing whether the same pickling thresholds hold for other commercially farmed seaweeds, would help translate this molecular map into industry-wide standards for one of the world’s fastest-growing aquaculture sectors.

Subject of Research: How pickling methods and packaging materials affect the flavor quality of ready-to-eat wakame

Article Title: Effect of pickling and packaging difference on characteristic flavor quality of ready-to-eat wakame

Article References: Lin, S.-J., Zhang, T.-T., Guo, Y., Zhang, K., Qin, L., & Huang, X.-H. (2026). Effect of pickling and packaging difference on characteristic flavor quality of ready-to-eat wakame. Blue Biotechnology, 3(1), Article 8. https://doi.org/10.1186/s44315-026-00059-9

Image Credits: AI Generated

DOI: 10.1186/s44315-026-00059-9

Keywords: wakame, pickling, flavor, volatile compounds, aluminum foil, PET packaging, seaweed, beta-cyclocitral, caryophyllene, off-odor, electronic nose, food quality