Shelf life prediction method of fresh-cut jackfruit based on characteristic metabolites

Abstract

The present invention provides a method for predicting the shelf life of fresh-cut jackfruit based on characteristic metabolites, which relates to the field of food preservation technology and solves the problem that existing shelf life prediction methods rely on subjective evaluation or a single physical and chemical indicator, resulting in insufficient accuracy and inability to dynamically integrate environmental parameters. This method uses liquid nitrogen grinding combined with HS-SPME / GC-MS technology to detect the metabolite components of fruit buds of different freshness, and uses the OPLS-DA model to screen differential metabolites with VIP values ​​> 1 and fold changes ≥ 2 or ≤ 0.5. A BP neural network model is constructed to input metabolite concentrations and storage parameters to predict shelf life, and the model is verified and corrected by comparing the predicted values ​​with the actual values. This method can dynamically optimize warehousing and logistics strategies and reduce quality loss of fresh-cut jackfruit.

Description

Technical Field

[0001] The present invention relates to the technical field of food preservation, and more particularly to a method for predicting the shelf life of fresh-cut jackfruit based on characteristic metabolites. Background Art

[0002] Fresh-cut jackfruit is prone to quality deterioration during storage and distribution, and its shelf life is complexly influenced by changes in metabolites and environmental factors. Existing shelf-life prediction methods primarily rely on sensory evaluation or monitoring of single physicochemical indicators (such as hardness and pH), which have significant limitations. Sensory evaluation is highly subjective, significantly influenced by differences in human experience, and cannot achieve quantitative prediction. While physicochemical indicators can reflect some quality changes, they struggle to capture early signals of dynamic metabolite degradation, resulting in delayed warnings and an inability to effectively guide real-time adjustments in the supply chain.

[0003] Traditional metabolite detection methods have the problem of loss of volatile components during sample processing. Conventional grinding techniques are mostly carried out at room temperature or ordinary freezing conditions, which cannot completely inhibit enzyme activity and oxidation reactions, resulting in partial degradation or volatilization of volatile metabolites (such as esters and aldehydes) during the preparation process. In existing methods, the balance between sample fragmentation efficiency and metabolite stability is difficult to control: excessive grinding may cause local temperature rise, while insufficient fragmentation leads to insufficient release of cell contents, both of which will affect the representativeness of the detection data. In addition, the temperature fluctuations and exposure time during sample transfer are not strictly controlled, further exacerbating the risk of distortion of the metabolite spectrum.

[0004] In terms of differential metabolite screening, existing studies mostly rely on statistical significance thresholds (such as P values) or single fold change criteria, lacking multi-dimensional stability verification. For example, screening markers solely by inter-group difference folds may miss metabolites that have important predictive value for shelf life but have smaller changes; conversely, over-reliance on statistical significance is susceptible to experimental batch errors or instrument noise interference, resulting in the selection of false-positive markers. More importantly, most methods do not systematically evaluate the repeatability of metabolite detection (such as intra-group coefficient of variation) and instrument analysis stability (such as retention time drift), resulting in poor reproducibility of screening results when verified across platforms or batches, limiting the generalization ability of the prediction model.

[0005] Existing shelf-life prediction models have obvious deficiencies in their handling of environmental parameters. Most models use static input parameters and are unable to integrate temperature and humidity fluctuation data of the storage environment in real time. For example, the effect of temperature on the degradation rate of metabolites is usually simplified to a linear relationship, ignoring the exponential correlation between temperature and reaction rate in the Arrhenius equation, causing the predicted value to deviate from the actual shelf life when the temperature changes suddenly. In addition, the dual effects of humidity on microbial growth and fruit water loss have not been quantitatively modeled, making it difficult to accurately reflect the risk of mold in high-humidity environments or changes in water loss rates under low-humidity conditions. This static modeling approach makes the prediction results unable to adapt to dynamic environmental changes in warehousing and transportation, reducing the timeliness of supply chain management strategies.

[0006] Insufficient biological replication is another technical difficulty. Some studies have not set up sufficient biological replicates or have not strictly screened the consistency of data within the group, resulting in metabolite concentration data being significantly affected by individual differences, sampling location deviations, or operational errors. For example, factors such as the maturity gradient of fresh-cut jackfruit pods and differences in local microbial colonization may introduce non-systematic variation. If this is not controlled through sufficient replicates and coefficient of variation screening, it will directly affect the reliability of differential metabolite screening, thereby reducing the accuracy of the prediction model.

[0007] The root cause of these problems lies in the lack of systematic solutions in existing technologies: high-fidelity extraction of metabolites is required during sample preparation, multidimensional screening criteria need to be established to improve marker stability, and the predictive model must be able to dynamically integrate environmental parameters. However, achieving these goals faces multiple technical bottlenecks. For example, the equipment cost and operational complexity of maintaining a liquid nitrogen cryogenic environment restrict large-scale application; the heterogeneity of multidimensional data (metabolites, environment, quality indicators) increases the difficulty of model training; and the spatial resolution and real-time data transmission of environmental sensors are insufficient, making it difficult to construct a high-precision three-dimensional parameter grid. These challenges have slowed the practical application of fresh-cut jackfruit shelf life prediction technology, and breakthroughs at the methodological level are urgently needed. Summary of the Invention

[0008] One object of the present invention is to provide a shelf life prediction method for fresh-cut jackfruit based on characteristic metabolites to optimize supply chain management strategies, ensure that fresh-cut jackfruit maintains good quality during its shelf life, and reduce losses.

[0009] In order to achieve these objects and other advantages of the present invention, according to one aspect of the present invention, the present invention provides a method for predicting the shelf life of fresh-cut jackfruit based on characteristic metabolites, comprising the following steps:

[0010] Fresh-cut jackfruit bud samples of varying degrees of freshness were obtained. Each sample was ground with liquid nitrogen and placed in a headspace vial. Saturated NaCl solution and internal standard solution were added. Automated headspace solid-phase microextraction (HS-SPME) combined with gas chromatography-mass spectrometry (GC-MS) was used to detect metabolite components in the sample.

[0011] Based on the orthogonal partial least squares discriminant analysis OPLS-DA model, the metabolite components of fresh-cut jackfruit at different freshness levels were screened. The screening criteria were:

[0012] a) The variable importance projection VIP value is greater than 1;

[0013] b) fold changes of metabolites in fresh-cut jackfruit at different freshness levels ≥ 2 or ≤ 0.5;

[0014] The metabolite components that meet both conditions a) and b) are screened as differential metabolites;

[0015] Using the screened differential metabolites as markers, a back-propagation BP neural network prediction model was constructed. The concentration data of the differential metabolites and storage environment parameters were input, and the shelf life prediction value of fresh-cut jackfruit was output. By comparing the prediction result data of the prediction model with the actual shelf life data, the accuracy of the marker screening was verified, and the markers were further screened and the prediction model was corrected.

[0016] This study uses liquid nitrogen grinding combined with HS-SPME / GC-MS technology to effectively preserve the integrity of volatile metabolites in fresh-cut jackfruit, avoiding the interference of high-temperature degradation on the test results. Using the OPLS-DA model to screen for differential metabolites with VIP values ​​greater than 1 and significant fold changes, key markers related to shelf life can be accurately located. By integrating multi-source data through the construction of a BP neural network model, dynamic shelf life prediction can be achieved, providing a scientific basis for supply chain management and reducing quality fluctuations and losses caused by subjective judgment.

[0017] Preferably, after constructing a back-propagation BP neural network prediction model using the screened differential metabolites, the model is trained and optimized; after the training is completed, the real-time collected fresh-cut jackfruit metabolite data is input into the model, and the shelf life prediction results are dynamically adjusted in combination with the real-time monitored storage temperature and humidity parameters; based on the dynamically adjusted shelf life prediction results, the supply chain management strategy is optimized, including but not limited to adjusting logistics transportation time, storage conditions and sales rhythm, to ensure that the fresh-cut jackfruit maintains good quality during the shelf life and reduces losses.

[0018] The dynamic adjustment mechanism in this invention uses real-time temperature and humidity data and model feedback to automatically adjust shelf life predictions to environmental fluctuations, improving prediction accuracy. Integrating an incremental learning algorithm to regularly update the weight matrix enhances the model's adaptability to seasonal or regional environmental changes. This optimized supply chain strategy allows for targeted adjustments to warehousing and logistics plans, ensuring optimal quality for fresh-cut jackfruit throughout transportation and sales.

[0019] Preferably, the GC-MS detection conditions include:

[0020] Chromatographic conditions: DB-5MS capillary column, 30 m × 0.25 mm × 0.25 μm, carrier gas: high-purity helium, constant flow rate: 1.2 mL / min, inlet temperature: 250°C, splitless injection, solvent delay: 3.5 min. Temperature program: 40°C, hold for 3.5 min, increase to 100°C at 10°C / min, then to 180°C at 7°C / min, and finally to 280°C at 25°C / min, hold for 5 min.

[0021] Mass spectrometry conditions: electron impact ion source (EI), ion source temperature 230 °C, quadrupole temperature 150 °C, mass spectrometry interface temperature 280 °C, electron energy 70 eV, scanning mode selected ion detection mode SIM, qualitative and quantitative ion precise scanning.

[0022] The combination of the DB-5MS column and EI ion source optimizes metabolite separation efficiency and ensures accurate qualitative and quantitative analysis of volatile components. The programmed temperature gradient (40°C to 280°C) addresses the detection of both low- and high-boiling-point metabolites, minimizing peak overlap. The SIM scan mode precisely selects characteristic ions, minimizing the impact of background noise on detection results and improving data reliability.

[0023] Preferably, the differential metabolite screening further comprises:

[0024] The common differential metabolites screened out were plotted into metabolite abundance box plots to show the changing trends of differential metabolites under different freshness levels.

[0025] The metabolite abundance boxplots used in this study intuitively display marker trends at different freshness stages, facilitating rapid identification of key quality indicators. Visual analysis can verify the effectiveness of screening criteria, assist researchers in optimizing metabolite screening thresholds, and further enhance the robustness of model predictions.

[0026] Preferably, the input parameters of the BP neural network model include:

[0027] Differential metabolite concentration data determined by the OPLS-DA model and multiple screening criteria, including the concentration data of volatile and non-volatile metabolites with significant differences at different freshness levels;

[0028] Real-time monitoring of storage temperature parameters and storage time accumulation parameters, where the temperature parameter is accurate to 0.1 ° C and the time parameter is in hours, and the storage temperature and time parameters are recorded;

[0029] The core indicators characterizing fruit quality include soluble solids content, measured by refractometer with an accuracy of ±0.5° Brix; total acid content, measured as tartaric acid by acid-base titration; and pH value, with an accuracy of ±0.01.

[0030] The input parameters in this paper include differential metabolite concentrations, storage environment parameters, and core quality indicators. This multi-dimensional data coupling enhances the comprehensiveness of the model's predictions. The temperature parameter accuracy of ± 0.1°C and the detailed recording of time accumulation parameters strengthen the quantitative correlation between environmental factors and metabolite degradation rates, avoiding prediction bias caused by a single variable.

[0031] Preferably, when screening differential metabolites based on the OPLS-DA model, the number of biological replicates is set to ≥ 3;

[0032] In univariate analysis, P value or FDR value was used as the indicator for determining statistical significance, where P value had to satisfy P < 0.05, or FDR value was controlled at ≤ 0.05, and the number of biological replicates corresponding to univariate analysis was ≥ 2.

[0033] The present invention uses a biological replicate count of ≥ 3, combined with P-value / FDR screening, to significantly enhance the statistical confidence of differential metabolite screening. Multiple validation mechanisms (P < 0.05 or FDR ≤ 0.05) in univariate analysis effectively eliminate random error interference and ensure the biological significance of screening results is consistent with actual shelf life.

[0034] Preferably, the step of screening for variable importance projection VIP value > 1 comprises:

[0035] Based on ≥ 3 independent biological replicate samples, each containing 5 or more fresh-cut fruit buds, the coefficient of variation (CV) of the peak area of ​​metabolites within the group was calculated, and only metabolites with CV ≤ 15% were retained for OPLS-DA modeling;

[0036] For volatile metabolites detected by GC-MS, retention time (RT) drift was calculated using quality control samples. For metabolites with a retention time (RT) deviation ≤ 0.15 min and a relative standard deviation (RSD) < 2%, the variable importance projection (VIP) value was multiplied by a stability factor of 1.2.

[0037] For metabolites with fold changes ≥ 2 or ≤ 0.5 in fresh-cut jackfruit at different freshness levels, the variable importance projection VIP value was superimposed with a difference enhancement factor of 0.3;

[0038] The standard deviation (SD) of the variable importance projection VIP values ​​of three independent experiments was calculated. Only metabolites with a standard deviation (SD) < 0.1 were retained, and candidate markers with poor reproducibility were eliminated.

[0039] For metabolites with variable importance projection VIP>1, the peak symmetry factor As is required to be between 0.9 and 1.1, and the signal-to-noise ratio S / N>30 to exclude instrument noise interference;

[0040] 200 permutation tests were introduced to ensure that metabolites with variable importance projection VIP>1 had a high predictive ability Q in the permutation model. 2 The intercept is <0.05 to avoid false positives caused by overfitting.

[0041] In this study, multi-level quality control, including CV ≤ 15%, RT drift correction, and signal-to-noise ratio screening, significantly reduced the impact of experimental fluctuations and instrument noise on VIP values. The application of differential enhancement factors and permutation tests enhanced the ability to identify significantly differentially expressed metabolites while mitigating the risk of model overfitting, ensuring robustness and reproducibility in biomarker screening.

[0042] Preferably, each sample is ground with liquid nitrogen and placed in a headspace bottle, and the process of adding saturated NaCl solution and internal standard solution specifically includes:

[0043] Fresh-cut fruit buds were ground with liquid nitrogen using a cryo-grinder at a frequency of 18-20 Hz. The grinding time was controlled within 30-45 s per grinding. Liquid nitrogen was continuously replenished during the grinding process to maintain the temperature ≤ -196°C, ensuring a cell disruption rate of ≥ 95% while preventing degradation of volatile metabolites caused by high temperatures.

[0044] After grinding, the sample was immediately transferred to a 50 mL headspace vial precooled to -80 °C, with a transfer time of <15 s;

[0045] The solution was added at a ratio of sample mass g: saturated NaCl solution volume mL = 1:4, and the salting-out effect was used with an ionic strength of ≥ 6 mol / L to promote the transfer of volatile metabolites from the aqueous phase to the headspace phase;

[0046] Ultrasonic pretreatment was performed at 40 kHz, 200 W power, and oscillation for 5 min;

[0047] Add 20 μL of a 10 μg / mL deuterated internal standard solution whose retention time differs from that of the target metabolite by ≥ 2 min;

[0048] 0.1% v / v ascorbic acid was added to the internal standard solution.

[0049] In this method, liquid nitrogen grinding (temperature ≤ -196°C) combined with salting-out (ionic strength ≥ 6 mol / L) maximizes the extraction of volatile metabolites and inhibits oxidative degradation. Ultrasonic pretreatment (40 kHz) promotes metabolite release, and the addition of deuterated internal standards and ascorbic acid effectively corrects instrument drift, ensuring data accuracy and batch-to-batch consistency.

[0050] Preferably, the HS-SPME detection conditions include:

[0051] The extraction was performed using an SPME Arrow extraction head. The extraction conditions were as follows: constant temperature oscillation at 60°C for 5 min and an oscillation rate of 200 rpm; the extraction head was kept 5 mm away from the sample liquid surface during headspace extraction for 15 min; and desorption was performed at 250°C for 5 min.

[0052] The use of the SPME Arrow tip (120 μm coating) in this study improves the enrichment efficiency of volatile metabolites by optimizing the adsorption phase and extraction time (15 minutes). Precise control of the constant temperature oscillation (60°C) and desorption temperature (250°C) balances the extraction rate with the risk of thermal decomposition, ensuring detection sensitivity and stability.

[0053] Preferably, the method for dynamically adjusting the shelf life prediction result in combination with the real-time monitored storage temperature and humidity parameters includes:

[0054] A temperature and humidity sensor array is deployed in the storage environment with a temperature accuracy of ± 0.1°C and a humidity accuracy of ± 2% RH. Real-time data is collected every 5 minutes and synchronized to the data processing terminal via a LoRa wireless transmission module. A three-dimensional environmental parameter grid covering the storage space is constructed with a spatial resolution of ≤ 0.5 m. 3 ;

[0055] Based on the Arrhenius equation, the degradation rate of characteristic metabolites at different temperatures was fitted. When the real-time temperature deviated from the preset storage temperature by ± 2°C, the metabolite concentration attenuation coefficient in the BP neural network was automatically adjusted with an adjustment step size of ≤ 0.05 min.

[0056] When humidity is >85% RH, a mold growth warning is triggered, and a shelf life reduction factor of 0.8-0.9 is superimposed on the model output layer. When humidity is <60% RH, a water loss rate compensation parameter of 10%-15% is added to correct the quality prediction deviation caused by water loss.

[0057] A sliding window incremental learning algorithm was used. Every 24 hours, the 50 most recently collected sets of temperature and humidity data and the corresponding differential metabolite concentrations were input into the BP neural network. The weight matrix related to environmental parameters was locally updated to ensure the model's adaptability to seasonal environmental fluctuations. The prediction error dynamically converged to ≤ 5%.

[0058] The three-dimensional temperature and humidity grid (resolution ≤ 0.5 m) in the present invention 3 ) combined with the Arrhenius equation enables spatially refined modeling of metabolite degradation rates. Humidity-triggered reduction coefficients and compensation parameters dynamically correct prediction deviations. An incremental learning algorithm (updating 50 sets of data every 24 hours) ensures the model continuously adapts to environmental changes and maintains a convergence state with a prediction error of ≤ 5%. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 This is a flowchart of an embodiment of the present application;

[0060] Figure 2 This is a schematic diagram of a fresh-cut fruit bract in a rigid state according to one embodiment of the present application;

[0061] Figure 3 This is a schematic diagram of a fresh-cut fruit bud in an immature state according to an embodiment of the present application;

[0062] Figure 4 This is a schematic diagram of a fresh-cut fruit bud in a mature state according to an embodiment of the present application;

[0063] Figure 5 This is a schematic diagram of a fresh-cut fruit bud in an over-ripe state according to an embodiment of the present application;

[0064] Figure 6 This is a box plot of the differential metabolite (E)-2-methylbutyric acid-3,7-dimethyl-2,6-octadienyl ester described in one embodiment of the present application;

[0065] Figure 7 This is a box plot of the differential metabolite 1H-pyrrole-2-carbonitrile described in one embodiment of the present application;

[0066] Figure 8 This is a box plot of the differential metabolite 5-ethyl-2-heptanol described in one embodiment of the present application;

[0067] Figure 9 This is a box plot of the differential metabolite isoamyl acetoacetate described in one embodiment of the present application;

[0068] Figure 10 This is a box plot of the differential metabolite 3-methylbutyric acid-2-phenylethyl ester described in one embodiment of the present application;

[0069] Figure 11This is a box plot of the differential metabolite ethyl isonicotinate described in one embodiment of the present application;

[0070] Figure 12 This is a box plot of the differential metabolite quinoxaline described in one embodiment of the present application;

[0071] Figure 13 This is a box plot of the differential metabolite 4-methyl-1-(1-methylethyl)bicyclo(3.1.0)-3-hexen-2-one described in one embodiment of the present application;

[0072] Figure 14 This is a box plot of the differential metabolite 5-butyldihydro-2(3H)-furanone described in one embodiment of the present application;

[0073] Figure 15 This is a box plot of the differential metabolite (Z)-1-(1-methoxyethoxy)-3-hexene described in one embodiment of the present application;

[0074] Figure 16 This is a box plot of the differential metabolite 1H-pyrrole-3-carbonitrile described in one embodiment of the present application;

[0075] Figure 17 This is a box plot of the differential metabolite 2-methylbutyric acid-2-phenylethyl ester described in one embodiment of the present application. DETAILED DESCRIPTION

[0076] The present invention will be described in further detail below in conjunction with the accompanying drawings so that those skilled in the art can implement the invention with reference to the description.

[0077] like Figure 1 As shown, the present invention provides a method for predicting the shelf life of fresh-cut jackfruit based on characteristic metabolites, comprising the following steps:

[0078] S1. Fresh-cut jackfruit bud samples of varying degrees of freshness were obtained. Each sample was ground with liquid nitrogen and placed in a headspace vial. Saturated NaCl solution and internal standard solution were added. Automated headspace solid-phase microextraction (HS-SPME) combined with gas chromatography-mass spectrometry (GC-MS) was used to detect metabolite components in the sample.

[0079] Specifically, the process of obtaining fresh-cut jackfruit bud samples of different freshness levels includes:

[0080] Fresh, eight-tenths-ripe "Thailand No. 8" jackfruits were harvested from Sandao Town, Baoting Li and Miao Autonomous County, Hainan Province, China, and transported to the National Institute of South China Seed Production and Production, Chinese Academy of Agricultural Sciences, Yazhou District, Sanya City, Hainan Province. Fruits were selected based on uniform color, size, surface mechanical damage, and freedom from pests and diseases. They were then grouped and used for subsequent experiments. After being stored at simulated room temperature (22°C), the jackfruits were cut open along the main axis using a sterilized stainless steel knife in a relatively clean room (26°C) on days 0, 3, 6, and 9. The stems were removed and intact pods were removed. The fruit was then packaged, weighed, and placed in covered fresh-keeping containers. The packaged jackfruits were then stored at different temperatures (10°C and room temperature), and random measurements and samples were taken at 0, 1, and every other day for subsequent experiments.

[0081] The following four points at different states were selected based on the physiological and biochemical indicators of the fruit bud to measure its volatile substance composition:

[0082] 1. Lignification: The fruit of the jackfruit stored for 0 days is relatively mature. On the 7th day of storage at low temperature, the fruit pods did not rot but began to turn white. The fruit pods of the whole fruit stored for 0 days and on the 9th day of low temperature shelf life were selected as the sample points for the lignification phenomenon. Figure 2 shown.

[0083] 2. Unripe: The fresh-cut jackfruit buds stored for 3 days at low temperature can maintain the change of bud hardness and inhibit the increase rate of conductivity and total phenol content. Fresh-cut jackfruits with soluble solid content below 11% and total acid content below 20 g / kg are in an unripe state and have not reached the range suitable for consumption. Therefore, the fresh-cut jackfruits stored for 3 days at low temperature are still in an unripe state on the first day of the shelf life. Figure 3 shown.

[0084] 3. Perfectly-ripe: The hardness of fresh-cut jackfruit in the mature state is about 10 N-12 N, the soluble solids are above 15%, and the total acid content is 23g / kg. The fresh-cut fruit buds stored for 6 days at a low temperature shelf life of 3 days have bright and vivid color on the surface of the jackfruit, a stronger smell, the soluble solids are at their highest point, and the total acid is in a steadily increasing stage. Therefore, this point is selected as the fully mature sample point (Perfectly-ripe). Figure 4 shown.

[0085] 4. Overripe (Decay): The whole fruit of jackfruit stored for 9 days is too mature. Although its mature aroma is very strong and the fruit is sweet, its shelf life is too short. When the whole fruit is stored for 9 days and the low-temperature shelf life is 4 days, the fruit bud is intact but softened, and the total acid content and soluble solids content are high. Therefore, this point is selected as the sample point of overripe state (Decay). Figure 5 shown.

[0086] Specifically, each sample was ground with liquid nitrogen and placed in a headspace bottle. The process of adding saturated NaCl solution and internal standard solution included the following:

[0087] Fresh-cut fruit buds were ground with liquid nitrogen using a cryo-grinder at 18 Hz. The grinding time was controlled within 40 seconds. Liquid nitrogen was continuously replenished during the grinding process to maintain a temperature of ≤ -196°C, ensuring a cell disruption rate of ≥ 95% while preventing degradation of volatile metabolites caused by high temperatures.

[0088] After grinding, approximately 500 mg of the sample was immediately transferred to a 50 mL headspace vial precooled to -80 °C, with a transfer time of <15 s.

[0089] The solution was added at a ratio of sample mass g: saturated NaCl solution volume mL = 1:4, and the salting-out effect was used with an ionic strength of ≥ 6 mol / L to promote the transfer of volatile metabolites from the aqueous phase to the headspace phase;

[0090] Ultrasonic pretreatment was performed at 40 kHz, 200 W power, and oscillation for 5 min;

[0091] Add 20 μL of a 10 μg / mL deuterated internal standard solution (deuterated ethyl acetate, deuterated methyl benzoate can also be used) whose retention time differs from that of the target metabolite by ≥ 2 min;

[0092] 0.1% v / v ascorbic acid was added to the internal standard solution.

[0093] Specifically, the detection conditions of fully automated headspace solid phase microextraction HS-SPME include:

[0094] The extraction was performed using an SPME Arrow tip (120 µm, DVB / CWR / PDMS) under the following conditions: constant temperature shaking at 60°C for 5 min, shaking rate at 200 rpm; headspace extraction was performed for 15 min with the tip kept 5 mm from the sample surface; and desorption was performed at 250°C for 5 min.

[0095] Before sampling, the extraction tip was aged in a Fiber Conditioning Station at 250°C for 5 min. Note: New extraction tips were aged in a Fiber Conditioning Station for 2 h before extraction.

[0096] Specifically, the detection conditions of gas chromatography-mass spectrometry (GC-MS) include:

[0097] Chromatographic conditions: DB-5MS capillary column, 30 m × 0.25 mm × 0.25 μm, carrier gas: high-purity helium, constant flow rate: 1.2 mL / min, inlet temperature: 250°C, splitless injection, solvent delay: 3.5 min. Temperature program: 40°C, hold for 3.5 min, increase to 100°C at 10°C / min, then to 180°C at 7°C / min, and finally to 280°C at 25°C / min, hold for 5 min.

[0098] Mass spectrometry conditions: electron impact ion source (EI), ion source temperature 230 °C, quadrupole temperature 150 °C, mass spectrometry interface temperature 280 °C, electron energy 70 eV, scanning mode selected ion detection mode SIM, qualitative and quantitative ion precise scanning.

[0099] S2. Based on the orthogonal partial least squares discriminant analysis (OPLS-DA) model, the metabolite components of fresh-cut jackfruit at different freshness levels were screened. The screening criteria were:

[0100] a) The variable importance projection VIP value is greater than 1;

[0101] b) fold changes of metabolites in fresh-cut jackfruit at different freshness levels ≥ 2 or ≤ 0.5;

[0102] The metabolite components that meet both conditions a) and b) are screened as differential metabolites;

[0103] Specifically, the first step is to screen differential metabolites by pairwise comparison. Based on the variable importance projection (VIP) obtained by the OPLS-DA model (biological replicates ≥ 3), the differential metabolites between samples of different freshness can be preliminarily screened.

[0104] At the same time, the P-value / FDR (biological replicates ≥ 2) or FC value of univariate analysis can be combined to further screen out differential metabolites. The screening criteria for differential metabolites in this application are:

[0105] 1. Select metabolites with VIP>1. The VIP value indicates the influence of the inter-group difference of the corresponding metabolite in the classification and discrimination of each group of samples in the model. It is generally believed that metabolites with VIP>1 have significant differences.

[0106] 2. Select metabolites with a fold change ≥ 2 and a fold change ≤ 0.5. That is, if the difference between the control and experimental groups is more than 2 times or less than 0.5, the difference is considered significant.

[0107] Using the screening results of the pairwise comparison, we screened out the differential metabolites in each comparison and made a box plot to show the trend of differential metabolites under different freshness levels. Figure 6-17 As shown:

[0108] The differential metabolites include: (E)-2-methylbutyric acid-3,7-dimethyl-2,6-octadienyl ester, 1H-pyrrole-2-carbonitrile, 5-ethyl-2-heptanol, isoamyl acetoacetate, 3-methylbutyric acid-2-phenylethyl ester, ethyl isonicotinate, quinoxaline, 4-methyl-1-(1-methylethyl)bicyclo(3.1.0)-3-hexen-2-one, 5-butyldihydro-2(3H)-furanone, (Z)-1-(1-methoxyethoxy)-3-hexene, 1H-pyrrole-3-carbonitrile, and 2-methylbutyric acid-2-phenylethyl ester.

[0109] S3. Use the screened differential metabolites as markers to construct a back-propagation BP neural network prediction model, input the concentration data of the differential metabolites and storage environment parameters, and output the shelf life prediction value of fresh-cut jackfruit; by comparing the prediction result data of the prediction model with the actual shelf life data, verify the accuracy of the marker screening, thereby further screening the markers and correcting the prediction model.

[0110] Specifically, the input parameters of the BP neural network model include:

[0111] Differential metabolite concentration data determined by the OPLS-DA model and multiple screening criteria, including the concentration data of volatile and non-volatile metabolites with significant differences at different freshness levels;

[0112] Real-time monitoring of storage temperature parameters and storage time accumulation parameters, where the temperature parameter is accurate to 0.1 ° C and the time parameter is in hours, and the storage temperature and time parameters are recorded. For example: temperature sensors are placed at the top, middle and bottom of the storage environment to form a three-dimensional monitoring grid with a spatial resolution of ≤ 0.5 m 3The data logging module is connected to the sensor via a shielded cable, fixed to the outer wall of the storage box. The pH meter probe is inserted directly into the fruit pod sample, avoiding exposure to air. During operation, temperature data is transmitted to the data processing terminal via the LoRa module.

[0113] The core indicators characterizing fruit quality are included, including soluble solids content, measured by refractometer with an accuracy of ±0.5°Brix; total acid content, calculated as tartaric acid, determined by acid-base titration; and pH value, with an accuracy of ±0.01.

[0114] Specifically, the model was trained using the backpropagation algorithm, with three hidden layers, 16, 8, and 4 nodes, respectively, and the ReLU activation function. The training dataset contained 200 sets of historical data, the validation set consisted of 50 sets, the learning rate was set to 0.001, and the number of iterations was 500.

[0115] In terms of equipment selection, model training can be run based on the TensorFlow or PyTorch frameworks and deployed on an embedded industrial computer with a CPU frequency ≥ 2.4 GHz and ≥ 8 GB of memory. The humidity sensor uses a capacitive sensor with a range of 0 to 100% RH and an accuracy of ± 2%. Training data is stored on a vibration-resistant SSD hard drive suitable for storage environments. The industrial computer is installed in the storage room control cabinet and connected to the sensor array via Ethernet. Humidity sensors are placed at the four corners of the fruit hull storage area.

[0116] During operation, a sliding window incremental learning algorithm is used every 24 hours, feeding the model with the latest 50 data sets to update the weight matrix associated with environmental parameters. After training, the model outputs a shelf life prediction, which is then validated using the mean squared error (MSE ≤ 0.1) compared to the actual shelf life data. If the error exceeds a threshold, differential metabolites are re-screened and the model structure optimized.

[0117] During the verification phase, the predicted results are compared with the actual shelf life data and the coefficient of determination R is calculated. 2 ≥ 0.9. The correction method included eliminating metabolites with a standard deviation of VIP values ​​> 0.1 and increasing the number of biological replicates to 5. The sample size for a single experiment was 30 fruit buds, divided into 3 groups, with 10 in each group. The statistical method used was the t-test ( P <0.05) or FDR-corrected (Q < 0.05).

[0118] Regarding equipment selection, data comparison can be performed using the MATLAB or Python SciPy libraries, running on the same industrial computer. Statistical analysis was performed using analysis of variance (ANOVA), with a significance level of α = 0.05. Regarding materials, the experimental samples were "Thailand No. 8" jackfruit buds, harvested from Baoting County, Hainan Province, and stored at 10°C and room temperature. Regarding assembly location, verification data was stored in a local database on the industrial computer, partitioned separately from the predictive model output data.

[0119] During the process, actual shelf life data is obtained through a combination of manual sensory evaluation and instrumental testing, including fruit firmness (tested using a texture analyzer), color change (tested using a colorimeter), and mold rate. If the prediction error consistently exceeds 5%, a model correction process is triggered: re-sample collection, optimization of the OPLS-DA screening conditions, and updating of the input layer node number of the BP neural network.

[0120] Furthermore, after using the screened differential metabolites to construct a back-propagation BP neural network prediction model, the model was trained and optimized; after the training was completed, the real-time collected fresh-cut jackfruit metabolite data was input into the model, and combined with the real-time monitored storage temperature and humidity parameters, the shelf life prediction results were dynamically adjusted; based on the dynamically adjusted shelf life prediction results, the supply chain management strategy was optimized, including but not limited to adjusting logistics transportation time, storage conditions and sales rhythm, to ensure that the fresh-cut jackfruit maintains good quality during the shelf life and reduces losses.

[0121] Specifically, the method for dynamically adjusting the shelf life prediction results by combining real-time monitoring of storage temperature and humidity parameters includes:

[0122] A temperature and humidity sensor array is deployed in the storage environment with a temperature accuracy of ± 0.1°C and a humidity accuracy of ± 2% RH. Real-time data is collected every 5 minutes and synchronized to the data processing terminal via a LoRa wireless transmission module. A three-dimensional environmental parameter grid covering the storage space is constructed with a spatial resolution of ≤ 0.5 m. 3In terms of equipment selection, temperature sensors can be thermocouples or platinum resistance sensors, humidity sensors can be capacitive or resistive sensors, and the wireless transmission module can support LoRa or NB-IoT communication protocols. Regarding materials, the sensor housing can be made of ABS engineering plastic, and the sealing ring can be made of silicone rubber. Regarding installation location, the temperature sensor can be installed on the top, middle, and bottom shelf beams of the storage room, while the humidity sensor can be placed in the four corners and center of the fruit hull storage area. The LoRa wireless transmission module can be fixed in the storage room control cabinet and connected to the sensor via a shielded cable. During operation, the sensor collects data every 5 minutes, filters it, and uploads it via the wireless module to construct a real-time, updated three-dimensional temperature and humidity distribution map. In the parameter setting method, the coordinate mapping of the three-dimensional grid can be achieved based on UWB positioning technology. Data verification uses the CRC-16 algorithm, with a packet loss rate of ≤0.1%. The sensor calibration cycle can be set to every 30 days, using a standard temperature and humidity source for calibration. During the experimental verification, a sensor array was deployed in a 10°C constant temperature warehouse, and data was collected continuously for 72 hours. The error of the calculated spatial temperature uniformity was ≤0.3°C.

[0123] Based on the Arrhenius equation, the degradation rate of characteristic metabolites at different temperatures is fitted. When the real-time temperature deviates from the preset storage temperature by ± 2°C, the metabolite concentration attenuation coefficient in the BP neural network is automatically adjusted with an adjustment step of ≤ 0.05 / min. When the real-time temperature deviates from the preset value by ± 2°C, the metabolite concentration attenuation coefficient adjustment step can be set to ≤ 0.05 / min. The activation energy parameter of the Arrhenius equation can be set to 35-45kJ / mol, and the frequency factor can be set to 1×10 12 - 5×10 13 min -1 . In terms of equipment selection, the data processing terminal can be equipped with an embedded industrial computer, the CPU main frequency can be ≥ 2.4 GHz, and the memory can be ≥ 8 GB. In terms of materials, the calculation data can be stored in an industrial-grade SSD hard drive, and the heat sink can be made of aluminum alloy. In terms of assembly location, the industrial computer can be installed in the storage room control cabinet and connected to the wireless transmission module via Ethernet. The dynamic parameter adjustment module can be integrated into the algorithm script written in Python or MATLAB. During the operation, the temperature data is input into the Arrhenius equation to calculate the real-time degradation rate, and the output result is transmitted to the BP neural network through the API interface to update the weight parameters of the hidden layer. The experimental method includes testing the metabolite degradation curve at three temperature gradients of 15 ℃, 25 ℃, and 35 ℃, and fitting the equation to determine the coefficient R 2≥ 0.95. During parameter correction, if the temperature fluctuation lasts for more than 10 minutes, the full model weight update is triggered. Statistical verification uses linear regression analysis, and the residual sum of squares ≤ 0.08;

[0124] When humidity exceeds 85% RH, a mold growth warning is triggered, and a shelf life reduction factor of 0.8 to 0.9 is added to the model output layer. When humidity is below 60% RH, a moisture loss rate compensation parameter of 10% to 15% is added to correct for quality prediction deviations caused by moisture loss. Regarding equipment selection, the warning module can be implemented using a buzzer or LED indicator, and the compensation algorithm can be embedded in the TensorFlow Lite framework. Regarding materials, the humidity probe can be made of a polyimide substrate, with an epoxy resin coating for moisture-proof encapsulation. Regarding assembly location, the warning indicator can be installed on the control panel at the storage room entrance, and the buzzer can be integrated into the temperature and humidity sensor housing. During operation, when humidity exceeds the threshold, the model output layer automatically adds the reduction factor and sends an alert signal to the logistics management system via the CAN bus. After the compensation parameters are updated, the shelf life prediction is recalculated, with an error convergence time of ≤ 1 hour. Functional testing includes simulating 85% and 55% RH environments in a humidity chamber to verify that the response delay of the reduction factor and compensation parameters is ≤ 2 minutes. In the structural design, the compensation algorithm is implemented using a table lookup method with a query interval of 1 minute.

[0125] A sliding window incremental learning algorithm was used. Every 24 hours, the 50 most recently collected sets of temperature and humidity data and the corresponding differential metabolite concentrations were input into the BP neural network. The weight matrix related to environmental parameters was locally updated to ensure the model's adaptability to seasonal environmental fluctuations. The prediction error dynamically converged to ≤ 5%.

[0126] The above-described embodiment utilizes high-precision environmental monitoring, dynamic parameter adjustment, and humidity compensation mechanisms to optimize the shelf life prediction model in real time. The system adapts to temperature and humidity fluctuations in the storage environment, reduces prediction errors caused by environmental deviations, improves supply chain management response speed, effectively maintains the quality stability of fresh-cut jackfruit, and reduces the risk of loss.

[0127] Furthermore, the step of screening the variable importance projection VIP value>1 includes:

[0128] Based on ≥ 3 independent biological replicate samples, each containing 5 or more fresh-cut fruit buds, the coefficient of variation (CV) of the peak area of ​​metabolites within the group was calculated, and only metabolites with CV ≤ 15% were retained for OPLS-DA modeling;

[0129] For volatile metabolites detected by GC-MS, retention time (RT) drift was calculated using quality control samples. For metabolites with a retention time (RT) deviation ≤ 0.15 min and a relative standard deviation (RSD) < 2%, the variable importance projection (VIP) value was multiplied by a stability factor of 1.2.

[0130] For metabolites with fold changes ≥ 2 or ≤ 0.5 in fresh-cut jackfruit at different freshness levels, the variable importance projection VIP value was superimposed with a difference enhancement factor of 0.3;

[0131] The standard deviation (SD) of the variable importance projection VIP values ​​of three independent experiments was calculated. Only metabolites with a standard deviation (SD) < 0.1 were retained, and candidate markers with poor reproducibility were eliminated.

[0132] For metabolites with variable importance projection VIP>1, the peak symmetry factor As is required to be between 0.9 and 1.1, and the signal-to-noise ratio S / N>30 to exclude instrument noise interference;

[0133] 200 permutation tests were introduced to ensure that metabolites with variable importance projection VIP>1 had a high predictive ability Q in the permutation model. 2 The intercept is <0.05 to avoid false positives caused by overfitting.

[0134] The above-described embodiment improves the accuracy of differential metabolite screening through multi-stage stability screening, dynamic correction, and rigorous statistical validation. The system effectively eliminates interference from instrument noise and experimental fluctuations, reduces the selection of false-positive markers, and ensures the reliability and generalizability of the prediction model, providing a stable data foundation for shelf life prediction of fresh-cut jackfruit.

[0135] Although the embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the description and implementation methods. They can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily implemented. Therefore, without departing from the general concept defined by the claims and the scope of equivalents, the present invention is not limited to the specific details and illustrations shown and described herein.

Claims

1. A method for predicting the shelf life of fresh-cut jackfruit based on characteristic metabolites, characterized in that: The following steps are involved: Fresh-cut jackfruit bud samples of varying degrees of freshness were obtained. Each sample was ground with liquid nitrogen and placed in a headspace vial. Saturated NaCl solution and deuterated internal standard solution were added. Automated headspace solid-phase microextraction (HS-SPME) coupled with gas chromatography-mass spectrometry (GC-MS) was used to detect metabolite components in the sample. Based on the orthogonal partial least squares discriminant analysis OPLS-DA model, the metabolite components of fresh-cut jackfruit at different freshness levels were screened. The screening criteria were: a) Variable importance projection VIP value > 1; b) fold changes of metabolites in fresh-cut jackfruit at different freshness levels ≥ 2 or ≤ 0.5; The metabolite components that meet both conditions a) and b) are screened as differential metabolites; Using the screened differential metabolites as markers, a back-propagation BP neural network prediction model was constructed. The differential metabolite concentration data and storage environment parameters were input, and the shelf life prediction value of fresh-cut jackfruit was output. The accuracy of the marker screening was verified by comparing the prediction results of the prediction model with the actual shelf life data. This allowed for further screening of the markers and correction of the prediction model. GC-MS detection conditions include: Chromatographic conditions: DB-5MS capillary column, 30 m × 0.25 mm × 0.25 μm, carrier gas: high-purity helium, constant flow rate: 1.2 mL / min, inlet temperature: 250 °C, splitless injection, solvent delay: 3.5 min, temperature program: 40 °C, hold for 3.5 min, increase to 100 °C at 10 °C / min, then increase to 180 °C at 7 °C / min, and finally increase to 280 °C at 25 °C / min and hold for 5 min. Mass spectrometry conditions: electron impact ion source (EI), ion source temperature 230°C, quadrupole temperature 150°C, mass spectrometry interface temperature 280°C, electron energy 70 eV, scanning mode selected ion detection (SIM), precise qualitative and quantitative ion scanning; The input parameters of the BP neural network prediction model include: Differential metabolite concentration data determined by the OPLS-DA model and multiple screening criteria, including the concentration data of volatile and non-volatile metabolites with significant differences at different freshness levels; Real-time monitoring of storage temperature parameters and storage time accumulation parameters, where the temperature parameter is accurate to 0.1°C and the time parameter is in hours, and the storage temperature and time parameters are recorded; The core indicators for characterizing fruit quality include soluble solids content, measured by refractometer with an accuracy of ±0.5°Brix; total acid content, measured as tartaric acid by acid-base titration; and pH value, with an accuracy of ±0.

01.

2. The method for predicting the shelf life of fresh-cut jackfruit based on characteristic metabolites according to claim 1, wherein After using the screened differential metabolites to construct a back-propagation BP neural network prediction model, the model was trained and optimized. After the training was completed, the real-time collected metabolite data of fresh-cut jackfruit was input into the model, and the shelf life prediction results were dynamically adjusted in combination with the real-time monitored storage temperature and humidity parameters. Based on the dynamically adjusted shelf life prediction results, the supply chain management strategy was optimized, including adjusting logistics transportation time, storage conditions and sales rhythm, to ensure that the fresh-cut jackfruit maintains good quality during the shelf life and reduce losses.

3. The method for predicting the shelf life of fresh-cut jackfruit based on characteristic metabolites according to claim 1, wherein Differential metabolite screening further includes: The common differential metabolites screened out were plotted into metabolite abundance box plots to show the changing trends of differential metabolites under different freshness levels.

4. The method for predicting the shelf life of fresh-cut jackfruit based on characteristic metabolites according to claim 1, wherein When screening differential metabolites based on the OPLS-DA model, the number of biological replicates was set to ≥3; In univariate analysis, P value or FDR value was used as the indicator for determining statistical significance, where P value was required to satisfy P < 0.05, or FDR value was controlled at ≤ 0.05, and the number of biological replicates corresponding to univariate analysis was ≥ 2.

5. The method for predicting the shelf life of fresh-cut jackfruit based on characteristic metabolites according to claim 1, wherein The screening step of the variable importance projection VIP value>1 includes: Based on ≥3 independent biological replicate samples, each containing more than 5 fresh-cut fruit buds, the coefficient of variation (CV) of the metabolite peak area within the group was calculated, and only metabolites with CV ≤ 15% were retained for OPLS-DA modeling; For volatile metabolites detected by GC-MS, retention time RT drift was calculated using quality control samples. For metabolites with retention time RT deviation ≤ 0.15 min and relative standard deviation (RSD) < 2%, the variable importance projection (VIP) value was multiplied by a stability factor of 1.

2. For metabolites with fold changes ≥2 or ≤0.5 in fresh-cut jackfruit at different freshness levels, the variable importance projection VIP value was superimposed with a difference enhancement factor of 0.3; The standard deviation (SD) of the variable importance projection VIP values ​​of three independent experiments was calculated. Only metabolites with a standard deviation (SD) < 0.1 were retained, and candidate markers with poor reproducibility were eliminated. For metabolites with variable importance projection VIP>1, the peak shape symmetry factor As is required to be between 0.9 and 1.1, and the signal-to-noise ratio S / N>30 to exclude instrument noise interference; 200 permutation tests were introduced to ensure that metabolites with variable importance projection VIP>1 had a high predictive ability Q in the permutation model. 2 The intercept is <0.05 to avoid false positives caused by overfitting.

6. The method for predicting the shelf life of fresh-cut jackfruit based on characteristic metabolites according to claim 1, wherein The process of grinding each sample with liquid nitrogen and placing it in a headspace bottle, and adding saturated NaCl solution and internal standard solution specifically includes: Fresh-cut fruit buds were ground with liquid nitrogen using a cryo-grinder at a frequency of 18-20 Hz. The single grinding time was controlled within 30-45 s. Liquid nitrogen was continuously replenished during the grinding process to maintain the temperature ≤ -196°C, ensuring a cell disruption rate of ≥95% while preventing the degradation of volatile metabolites caused by high temperatures. After grinding, the sample was immediately transferred to a 50 mL headspace vial precooled to -80 °C, with a transfer time of <15 s; The solution was added at a ratio of 1:4 (sample mass g: saturated NaCl solution volume mL). The salting-out effect was used and the ionic strength was ≥ 6 mol / L to promote the transfer of volatile metabolites from the aqueous phase to the headspace phase. Add 20 μL of a 10 μg / mL deuterated internal standard solution whose retention time differs from that of the target metabolite by ≥ 2 min; 0.1% v / v ascorbic acid was added to the internal standard solution.

7. The method for predicting the shelf life of fresh-cut jackfruit based on characteristic metabolites according to claim 1, wherein HS-SPME detection conditions include: The extraction was performed using an SPME Arrow extraction head. The extraction conditions were as follows: constant temperature oscillation at 60°C for 5 min, an oscillation rate of 200 rpm; the extraction head was kept 5 mm away from the sample liquid surface during headspace extraction for 15 min; and desorption was performed at 250°C for 5 min.

8. The method for predicting the shelf life of fresh-cut jackfruit based on characteristic metabolites according to claim 1, wherein Methods for dynamically adjusting shelf life prediction results based on real-time monitoring of storage temperature and humidity parameters include: A temperature and humidity sensor array is deployed in the storage environment with a temperature accuracy of ± 0.1°C and a humidity accuracy of ± 2% RH. Real-time data is collected every 5 minutes and synchronized to the data processing terminal via the LoRa wireless transmission module. A three-dimensional environmental parameter grid covering the storage space is constructed with a spatial resolution of ≤ 0.5 m. 3 ; Based on the Arrhenius equation, the degradation rate of characteristic metabolites at different temperatures was fitted. When the real-time temperature deviated from the preset storage temperature by ±2°C, the metabolite concentration attenuation coefficient in the BP neural network was automatically adjusted with an adjustment step size of ≤ 0.05 min. When humidity is >85% RH, a mold growth warning is triggered, and a shelf life reduction factor of 0.8 to 0.9 is superimposed on the model output layer. When humidity is <60% RH, a water loss rate compensation parameter of 10%-15% is added to correct the quality prediction deviation caused by water loss. A sliding window incremental learning algorithm was used. Every 24 hours, the 50 most recently collected sets of temperature and humidity data and the corresponding differential metabolite concentrations were input into the BP neural network. The weight matrix related to environmental parameters was locally updated to ensure the model's adaptability to seasonal environmental fluctuations. The prediction error dynamically converged to ≤ 5%.

Images