Food and Bioprocess Technology最新文献

Postharvest losses of apples during storage and transportation pose a significant challenge to food security and economic benefits. This research proposed a framework for apple spoilage detection and prediction that combines multi-gas sensing, deep learning, and a cloud–edge device. Relevant environmental parameters were continuously collected from the storage environment and transmitted to a cloud platform for unified management and model training. Enhanced models LSTM-KAN, BiLSTM-KAN, and TCN-KAN were constructed by integrating the Kolmogorov–Arnold network (KAN) module into the neural network architecture. These models demonstrated higher prediction accuracy and robustness in apple spoilage prediction tasks and significantly improved training convergence speed. Based on multi-sensor array data capturing key physicochemical parameters, the trained models predicted the composite spoilage index (CSI), which quantitatively reflected the degree of apple spoilage, with the TCN-KAN model showing the best performance (({R}_{p}^{2}) = 0.968). Furthermore, a cloud–edge device collaboration framework was constructed to support centralized model updates, remote synchronization, and real-time inference. To ensure efficient deployment on resource-constrained devices, a structured pruning technique was developed to reduce model size and improve inference speed while minimizing the loss of prediction accuracy. The pruned TCN-KAN model achieved a 42.68% speed improvement. The proposed method provides an accurate, scalable, and cost-effective approach for real-time monitoring and prediction of spoilage in stored apples, highlighting the potential of deep learning and edge intelligence in smart agriculture applications.

Graphical Abstract

To overcome the long cooking time and poor gelatinization property of Coix seed for high-quality pregelatinized products, this study employed ultrasonic-assisted soaking pretreatment combined with far infrared drying (IRD) to optimize the processing parameters. Results showed that the optimal ultrasound pretreatment conditions were an ultrasonic power of 60 W, a frequency of 28 kHz, a treatment duration of 25 min, and a material-to-liquid ratio of 1:2 (g/mL). Ultrasound pretreatment significantly shortened the cooking time of Coix seeds while enhancing their water-holding capacity and swelling volume. The total starch content and color remained relatively stable; however, a slight decrease in protein content and an increase in fat content were observed. The logarithmic model exhibited the best fit for the drying kinetics of Coix seeds. At an IRD temperature of 60 ℃, the cooking time, contact angle, and hardness of Coix seeds reached their minimum values, while the rehydration ratio was maximized. The ultrasonic-assisted far infrared (US-IRD) process effectively enhanced and preserved the micro-porous surface structure of Coix seeds. Low-field nuclear magnetic resonance (LF-NMR) and magnetic resonance imaging (MRI) analyses revealed that during far infrared drying, moisture migrated progressively from the Coix seed core to the surface, resulting in uniform moisture distribution and efficient dehydration. These findings demonstrate the superiority of the ultrasonic-assisted far infrared process for enhancing the functional and structural properties of Coix seed, thereby establishing a theoretical foundation for the mechanism research of ultrasound in food processing.

The identification of genetically modified (GM) food is very important in the cross-border circulation of foods due to different regulatory policies between different countries. Current detection methods mainly rely on nucleic acid amplification techniques, lacking simple, stable, and low-cost approaches. In this work, we have proposed a novel strategy for amplification-free monitoring GM food (taking GM corn MON810 as an example) based on polyelectrolyte-induced self-assembly between the perylene derivative (PBTMA) and single-stranded DNA (MP). A supramolecular complex of PBTMA/MP was formed driven by the non-covalent interaction, resulting in fluorescence quenching. In the presence of MON810 target DNA, a double helix structure was formed between MP and the target DNA, weakening the interaction strength of PBTMA/MP and restoring fluorescence, thereby enabling detection of MON810. The detection linear range was 5–100 nM, with a limit of detection (LOD) of 1.17 nM. This strategy successfully distinguished and detected MON810 in actual samples and is expected to be extended to all GM food. To our knowledge, this work is the first to report the use of a polyelectrolyte-induced self-assembly strategy for enzyme-free and amplification-free detection of GM food.

Graphical Abstract

Listeria monocytogenes is a major foodborne pathogen capable of surviving under adverse conditions, and combining high hydrostatic pressure (HHP) with nisin offers a promising strategy for its effective control. However, knowledge remains limited regarding their combined efficacy across strains with different pressure tolerances and in varied model systems. This study systematically assessed the antimicrobial efficacy and synergistic effect of HHP (200-350 MPa, 10 min) combined with nisin (100–500 IU/mL) against five L. monocytogenes strains with varying pressure tolerance in BHI broth and ACES buffer. Nisin MICs ranged from 300 to 500 IU/mL, with 10430S being the most susceptible. Combined HHP-nisin treatments resulted in a significant synergistic effect across most tested conditions. At 300 MPa + 500 IU/mL nisin, combined inactivation reached 5.2-5.3 log₁₀ reduction in BHI and 2.7-4.0 log₁₀ in ACES for the most sensitive strains (LO28, FBR13, 10403S). The synergistic effect was mostly observed at higher pressures (300-350 MPa) and nisin concentrations (500 IU/mL), reaching up to 2.9 log₁₀ reduction in ACES buffer and 1.5 log₁₀ reduction in BHI. The antimicrobial efficacy and synergistic effect were strongly influenced by strain, pressure intensity, and medium. The highest synergistic effect was observed in FBR13 under most conditions (p < 0.05) despite not being the most pressure-sensitive strain. In BHI, synergism was generally lower than in ACES buffer across most treatment conditions, except at 500 IU/mL. At 300 MPa + 200 IU/mL nisin, the synergistic effect ranged from 0.6 to 1.2 log₁₀ reduction in ACES and 0.1 to 0.6 log₁₀ in BHI. Nisin alone resulted in significantly greater reduction in ACES (1.1 to 1.9 log₁₀ reduction) than in BHI (0.3 to 1.1 log₁₀ reduction) across all concentrations. This study highlights the strong potential of the HHP-nisin combination as a multi-hurdle approach for L. monocytogenes control, offering valuable insights to optimize treatment parameters.

There is strong commercial interest in the successful fractionation of whey protein into pure products. However, the effect of environmental conditions on membrane fractionation is not fully understood. In the current study filtration pressures, membranes of various pore sizes and types, and pH manipulation, were employed to investigate the effect of operating conditions on the enrichment of α-lactalbumin from whey protein. All of these factors were shown to affect the protein profile of the filtrate produced. Industrially relevant polyethersulfone microfiltration and ultrafiltration membranes, on a small-scale membrane filtration rig, were both able to filter out lactoferrin and lactoperoxidase, producing a filtrate that primarily contained β-lactoglobulin and α-lactalbumin. When scaled-up to a large cross-flow system, only the microfiltration membrane demonstrated a filtrate fractionation with an α-lactalbumin-enriched protein stream. When operated at 3 bar with a 1 in 3 dilution of a whey protein feedstock, a 140% increase in α-lactalbumin was achieved compared with a commercial whey protein liquid feed. This study shows the importance of the optimisation of all conditions during membrane fractionation and demonstrates that this technique can be used to produce an α-lactalbumin enriched product of industrial relevance. 

Microencapsulated and freeze-dried phenolic olive pomace extracts (M-OPE and F-OPE, respectively) were added to “cavatelli” fresh pasta samples, with the aim of evaluating  their effects on the physicochemical and microbiological quality of fresh pasta during refrigerated storage for 120 days. The results showed a reduction in moisture, a_w, total phenolic compounds and antioxidant activity throughout storage. However, at the end of storage, pasta samples had moisture and a_w higher than 24% and 0.92, respectively. Notably, experimental samples with M-OPE showed lower water content, and higher TPC, compared to the corresponding samples with F-OPE. Both M-OPE and F-OPE significantly reduced pH values, compared to control pasta (CP). L* values decreased as the extracts increased, showing an uptrend from the time of production, until 120 days of storage. Samples with M-OPE had higher a* values, compared to the corresponding samples with F-OPE. On the other hand, yellow index (b*) showed a decreasing trend until end of storage time, more pronounced for samples with F-OPE. Just after production, CP and experimental samples showed low concentrations of volatile organic compounds. However, the oxidation process during storage affected volatile profile, particularly for CP and for the sample with the highest F-OPE content. The addition of the extracts, and particularly M-OPE, reduced the growth of spoilage microorganisms; experimental samples exhibited significantly lower counts for total aerobic mesophilic bacteria, moulds and yeast microbial groups, compared to CP, showing OPE potential to extend the shelf-life of fresh pasta.

Food such as concentrated maple sap is prone to microbial contamination during collection, handling and transport, requiring effective processing to ensure safety. Conventional thermal treatments can compromise nutritional quality, sensory and physicochemical properties. Ohmic heating (OH) offers a promising alternative green process with valuable benefits while preserving food quality, reducing energy consumption and greenhouse gas emissions. This study evaluates Clostridium sporogenes PA 3679 destruction rates as a surrogate for low-acid (pH 7.9) concentrated maple sap (10, 20 and 28.5°Brix) at different temperature (95–121 °C) under direct OH at a voltage gradient (V_g) of 90 (V/cm). A nonlinear microbial death rate was observed with a sharp spore inactivation during come-up time (CUT) due to high applied V_g, followed by a tailed lag during holding time at a lower V_g. This profile likely caused a change in spore protein structure that resulted in a change of spore inactivation rate due to the synergistic effect of the thermal and electrical field. The survival data were fitted to various primary kinetic models (Weibull, log-linear model, Gompertz and sigmoid). The Weibull model provided the best fit (δ: 0.010–0.446; β: 0.238–0.408; adj-R²: 0.891–0.934), yielding equivalent D-values ranging from 0.21 to 3.3 min across different temperatures (95 to 121 °C) and media tested. Imaging using transmission electron microscopy and scanning electron microscopy revealed distinct structural destruction changes of C. sporogenes spores under rapid CUT as a result of electrical-thermal shock during OH. These findings offer a scientific basis for designing safe pasteurization or sterilization processes for food products such as concentrated maple sap, fruit juices and milk using a green emerging technology such as OH.

Novel nanoparticles composed of zein/choline chloride were effectively synthesized by combining various mass ratios of 3:1, 2:1, 1:1, 1:2, and 1:3 (ratio of zein to choline chloride w/w). The zein/choline chloride composite nanoparticles with ratios of 3:1 and 2:1 exhibited contact angles nearing 90° (94.72° and 87.68°, respectively), demonstrating appropriate wettability for the formulation of oil-in-water Pickering emulsions. The successful interaction between zein and choline chloride nanoparticles was validated through FTIR analysis. The size distribution of the zein/choline chloride composite nanoparticles, as assessed by DLS tests and TEM imaging, showed good correlation. The novel formulated composite nanoparticles (at ratios of 3:1 and 2:1) were utilized as colloidal particles to create a Pickering emulsion containing curcumin, and their application (5, 10, and 15% v/v) was examined within the structure of electrospun zein fibers. FESEM images of the electrospun zein fibers indicated that the incorporation of a Pickering emulsion containing curcumin at levels exceeding 10% resulted in bead formation within the electrospun fibers. The tensile strength, Young’s modulus, and elongation at break were improved due to the incorporation of Pickering emulsions containing curcumin. A sustained release of curcumin was noted over 60 days in various food simulant media. Fickian diffusion was identified as the primary mechanism governing the release of curcumin in these simulated conditions. Furthermore, the Peppas-Sahlin model was determined to be the most suitable model for characterizing the release profile of curcumin.