Method and system for diagnosing bacterial contamination of tremella aurantialba based on characteristic metabolite detection

Abstract

The invention belongs to the technical field of agricultural intelligent detection, and particularly relates to a tremella aurantialba bacterial contamination diagnosis method and system based on characteristic metabolite detection.The method comprises the steps that firstly, quantitative values of three characteristic metabolites including azelaic acid, choline glycerophosphate and L-arabitol are accurately obtained through optimized ternary solvent extraction and targeted mass spectrometry; and in combination with sample growth stage information, calculating a standardized diagnosis index relative to a stage health benchmark so as to eliminate growth difference interference. Secondly, performing two-stage intelligent analysis: performing abnormal preliminary screening based on a first-class support vector machine model, and generating first contamination assessment information containing a specific contamination type tendency through similarity matching between metabolic characteristics and a historical contamination clustering mode; meanwhile, by calculating the mahalanobis distance between the to-be-detected sample and the same batch of samples, second contamination evaluation information representing the metabolism consistency of the group is generated. And finally, fusing the two types of evaluation information, querying the multi-dimensional decision rule table, and outputting a hierarchical diagnosis result.

Description

Technical Field

[0001] This invention belongs to the field of agricultural intelligent detection technology, specifically relating to a diagnostic method and system for fungal infection of Auricularia auricula-judae based on the detection of characteristic metabolites. Background Technology

[0002] Currently, the detection of fungal infections in edible fungi such as *Tremella aurantialba* mainly relies on traditional sensory morphological observation, pathogen isolation and culture, microscopic examination, and molecular biology techniques such as PCR / qPCR. Sensory identification, which involves observing changes in the color and odor of the fruiting body, is intuitive but highly subjective. Isolation and culture methods require several days for plate culture and morphological identification of pathogens, offering high accuracy but low efficiency. Molecular detection techniques, based on specific primers to amplify pathogen DNA, are highly sensitive but only applicable to known pathogens and cannot reflect changes in the host's physiological state under fungal stress. In recent years, metabolomics techniques, especially untargeted metabolomics, have begun to be applied to research on biological stress responses. They can unbiasedly detect overall changes in small molecule metabolites in organisms, providing the possibility of discovering biomarkers under disease or stress conditions.

[0003] However, in the existing technology system, sensory recognition and culture methods have a severely lagging detection window, making it impossible to achieve early warning of bacterial contamination; molecular detection technology is limited by known pathogen primers and lacks the ability to identify mixed infections or unknown bacteria; and existing metabolomics applications are mostly limited to the laboratory research stage, and the massive and complex data generated by them require professionals to conduct multivariate statistical analyses (such as PCA and OPLS-DA) that take several hours to several days, making it difficult to transform them into standardized diagnostic solutions applicable to the production site. Summary of the Invention

[0004] The purpose of this invention is to provide a diagnostic method and system for fungal infection of Auricularia auricula based on the detection of characteristic metabolites, so as to solve the problem of how to establish an early detection technology for fungal infection of Auricularia auricula that can be directly applied to the production line, with standardized operation procedures and rapid interpretation, without the need for prior identification of specific pathogens.

[0005] The present invention achieves the above objectives through the following technical solutions: Firstly, the present invention proposes a diagnostic method for fungal infection of *Auricularia auricula-judae* based on the detection of characteristic metabolites, the method comprising: Mass spectrometry data of the *Auricularia auricula-judae* sample to be tested are obtained, and quantitative response values ​​of characteristic metabolites are determined based on the mass spectrometry data. The characteristic metabolites include azelaic acid, glycerophosphate choline, and L-arabinitol. Based on the quantitative response values ​​of the characteristic metabolites and the growth stage information of the test golden ear samples, a standardized diagnostic index relative to the health benchmark of the corresponding growth stage is determined. Anomaly detection and pattern matching analysis were performed on the standardized diagnostic index to obtain the first bacterial infection assessment information; Obtain the metabolic consistency information of the same batch of the cultivation unit to which the test golden ear sample belongs, and determine the second contamination assessment information based on the metabolic consistency information; Based on the first and second bacterial infection assessment information, the final diagnostic result of the bacterial infection status of *Auricularia auricula-judae* is determined.

[0006] Furthermore, determining the quantitative response value of the characteristic metabolite based on the mass spectrometry detection data includes: The mass spectrometry detection data were subjected to quality control verification; the mass spectrometry detection data was obtained from the test *Auricularia auricula-judae* sample by ultra-high performance liquid chromatography-tandem mass spectrometry. Based on the preset characteristic metabolite identification information, the signal intensity of the corresponding target characteristic peak is extracted from the calibrated mass spectrometry data; The signal intensity of the extracted target feature peak is corrected by using the signal intensity of the internal standard corresponding to each characteristic metabolite to obtain the quantitative response value of the target metabolite.

[0007] Furthermore, the extraction of signal intensity based on preset characteristic metabolite identification information includes: Based on the retention time window and precise mass-to-charge ratio information associated with each characteristic metabolite, the corresponding chromatographic peaks are located from the mass spectrometry chromatogram; Integrate the located chromatographic peaks and calculate their peak area or peak height; The peak area or peak height obtained by integration is used as the signal strength of the target characteristic peak.

[0008] Furthermore, the determination of a standardized diagnostic index relative to a health benchmark for the corresponding growth stage, based on the growth stage information of the *Eurydon edulis* sample to be tested, includes: Based on the growth stage information, the average healthy azelaic acid value corresponding to that stage is obtained from a pre-constructed segmented healthy metabolite benchmark library. with standard deviation glycerophosphate choline healthy mean with standard deviation and the average health value of L-arabinitol with standard deviation ; Calculate the following standardized diagnostic indices: First Index , ; in, and Health samples at the corresponding stages The mean and standard deviation of the values; Second Index , , where is the quantitative response value of L-arabinol.

[0009] Furthermore, the step of performing anomaly detection and pattern matching analysis on the standardized diagnostic index to obtain first contamination assessment information includes: The first index Zr and the second index Zc are input into a pre-trained pattern recognition model for abnormal state discrimination. The pattern recognition model is a support vector machine model trained based on healthy golden ear sample data; If the pattern recognition model determines that the first index Zr and the second index Zc are within the range of healthy data distribution, then the first bacterial contamination assessment information is generated to characterize the normal metabolic state of the sample. If the pattern recognition model determines that the first index Zr and the second index Zc deviate from the distribution range of health data, then the vector [Zr, Zc] formed by the first index Zr and the second index Zc is matched with the similarity of multiple predefined typical bacterial infection pattern prototype vectors. The state label associated with the prototype vector of the most similar typical bacterial infection pattern is determined as the first bacterial infection assessment information.

[0010] Furthermore, the step of performing similarity matching between the vector [Zr, Zc] formed by the first index Zr and the second index Zc and multiple predefined typical bacterial infection pattern prototype vectors includes: The cosine similarity Sk between the vector [Zr, Zc] and the prototype vector Pk of the k-th typical bacterial infection pattern is calculated as follows: ; in, Represents the vector dot product. The Euclidean norm of a vector; The matching result is to obtain the prototype vector Pk corresponding to the maximum cosine similarity max(Sk) and its associated state label; The typical bacterial infection pattern prototype vector Pk is predefined through the following steps: Collect M historically confirmed bacterial infection samples, each sample i corresponding to a feature vector. ; For the M feature vectors { Perform cluster analysis based on density peaks; For the k-th cluster obtained from cluster analysis, its prototype vector Pk is calculated as follows: ; in, Let be the number of samples belonging to the k-th cluster. Let be the set of all samples in the k-th cluster; Assign a corresponding state label to the prototype vector Pk based on the main pathogen type or infection stage corresponding to the sample in the k-th cluster.

[0011] Furthermore, the status label includes at least one of the following: "defense response-dominated", "membrane damage-dominated", "osmotic stress response", or "mixed metabolic disorder".

[0012] Furthermore, the metabolic consistency information of the same batch population within the cultivation unit to which the tested *Auricularia auricula-judae* sample belongs is obtained, and the second contamination assessment information is determined based on the metabolic consistency information, including: Obtain mass spectrometry data from at least N other golden ear fungus samples from the same cultivation unit and the same harvesting cycle, where N≥3; Based on the mass spectrometry data of the other golden ear samples, their respective first index Zr and second index Zc are determined; Calculate the covariance matrix of the dataset consisting of the [Zr, Zc] vectors of the other golden ear samples, and calculate the Mahalanobis distance of the [Zr, Zc] vector of the golden ear sample to be tested relative to the dataset based on this covariance matrix. Based on the magnitude of the Mahalanobis distance, a second bacterial contamination assessment is generated, characterizing the degree of consistency between the metabolic patterns of the sample under test and the same batch of population.

[0013] Furthermore, determining the final diagnostic result of the fungal infection status of *Auricularia auricula-judae* based on the first and second fungal infection assessment information includes: The assessment status contained in the first contamination assessment information is mapped to a first confidence score, and the consistency level contained in the second contamination assessment information is mapped to a second confidence score. The first confidence score and the second confidence score are weighted and summed according to preset weighting coefficients to obtain a comprehensive decision index; Based on the status label type contained in the first contamination assessment information, the numerical range of the comprehensive decision index, and the consistency level of the second contamination assessment information, query the pre-constructed multidimensional decision rule table; Based on the query results of the multidimensional decision rule table, the corresponding final diagnostic results are output.

[0014] Secondly, the present invention proposes a diagnostic system for fungal infection of *Auricularia auricula-judae* based on the detection of characteristic metabolites, used to implement the steps of the rapid diagnostic method for fungal infection of *Auricularia auricula-judae* as described above, the system comprising: The mass spectrometry data acquisition unit is used to acquire mass spectrometry detection data of the *Eurydon edulis* sample to be tested. A data preprocessing unit is used to determine the quantitative response value of a characteristic metabolite based on the mass spectrometry detection data, wherein the characteristic metabolite includes azelaic acid, glycerophosphate choline, and L-arabinitol; The index calculation unit is used to determine a standardized diagnostic index relative to the health benchmark of the corresponding growth stage based on the quantitative response value of the characteristic metabolite and the growth stage information of the test golden ear sample. The first assessment unit is used to perform anomaly detection and pattern matching analysis on the standardized diagnostic index to obtain the first bacterial infection assessment information. The second evaluation unit is used to obtain the metabolic consistency information of the same batch of the cultivation unit to which the test golden ear sample belongs, and to determine the second contamination evaluation information based on the metabolic consistency information. The decision fusion unit is used to determine the final diagnostic result of the fungal infection status of *Auricularia auricula-judae* based on the first fungal infection assessment information and the second fungal infection assessment information.

[0015] The beneficial effects of this invention are as follows: 1. This invention establishes a targeted detection system based on azelaic acid, glycerophosphate choline, and L-arabinitol, achieving precise capture of specific metabolic biomarkers for early stress response in *Auricularia auricula-judae*. A phased dynamic benchmark is proposed to effectively overcome the interference of metabolic background differences in *Auricularia auricula-judae* at different growth stages on interpretation. A two-stage analysis strategy is adopted: first, a support vector machine model is used to efficiently screen abnormal samples; then, metabolic pattern matching is used to achieve precise identification of the infestation type, providing instructive classification information while ensuring screening efficiency. 2. This invention introduces population metabolic consistency analysis for auxiliary verification, quantifying the relative position of individuals in a batch using Mahalanobis distance, improving the reliability of single-indicator judgment. Finally, a clear-cut diagnostic conclusion is output through multi-dimensional rule fusion, transforming complex laboratory analysis into a standardized on-site operation procedure. Attached Figure Description

[0016] Figure 1 This is a flowchart of a method for diagnosing fungal infection in *Auricularia auricula-judae* based on the detection of characteristic metabolites in this invention. Figure 2 This is a flowchart illustrating how the final diagnostic result is obtained based on the first and second bacterial infection assessment information in this invention. Figure 3 This is a flowchart of a two-stage analysis of standardized diagnostic indices in this invention. Detailed Implementation

[0017] The present application will now be described in further detail with reference to the accompanying drawings. It should be noted that the following specific embodiments are only used to further illustrate the present application and should not be construed as limiting the scope of protection of the present application. Those skilled in the art can make some non-essential improvements and adjustments to the present application based on the above application content.

[0018] This invention provides a method and system for diagnosing Auricularia auricula-judae infection based on the detection of characteristic metabolites. The method can be applied to portable or laboratory mass spectrometry analysis systems, as well as corresponding data analysis software and computer-readable storage media. This embodiment uses a rapid diagnostic method applied to a field rapid detection system integrating a mass spectrometer and an embedded analysis unit as an example.

[0019] Example 1 Combination Figures 1-3 The specific process will be described in detail below. The diagnostic method for fungal infection of *Auricularia auricula-judae* based on the detection of characteristic metabolites may include the following steps: S1. Obtain the mass spectrometry detection data of the *Auricularia auricula-judae* sample to be tested. Based on the mass spectrometry detection data, determine the quantitative response values ​​of the characteristic metabolites, which include azelaic acid, glycerophosphocholine, and L-arabitol.

[0020] Optionally, before performing step S1 to obtain mass spectrometry detection data, the method further includes standardized metabolite extraction of the *Auricularia auricula-judae* sample to be tested, in order to obtain a test solution that can be used for mass spectrometry injection, specifically including: Take an appropriate amount of freeze-dried and powdered *Auricularia auricula-judae* fruiting body tissue to be tested and place it in a pre-cooled centrifuge tube. Add pre-cooled metabolite extraction solution, which is a ternary mixed solvent prepared by water, acetonitrile, and isopropanol in a volume ratio of 1:1:1. This specific solvent system has been optimized to simultaneously and efficiently extract a wide range of endogenous metabolites from *Auricularia auricula-judae* tissue, including highly polar (such as sugars and organic acids), moderately polar, and weakly polar (such as lipids), making it suitable for *Auricularia auricula-judae* tissue rich in mucilage and polysaccharides.

[0021] In a preferred embodiment, determining the quantitative response value of a characteristic metabolite based on mass spectrometry detection data includes: The mass spectrometry detection data were quality controlled and verified to ensure the reliability and stability of the data. The mass spectrometry detection data were obtained by collecting the samples of *Auricularia auricula-judae* using an ultra-high performance liquid chromatography-tandem mass spectrometry (UHPLC-MS / MS) system equipped with an electrospray ionization source (ESI) in full scan / data-dependent secondary scan (Full MS / dd-MS2) mode.

[0022] Based on pre-defined characteristic metabolite identification information, the signal intensity of the corresponding target characteristic peak is extracted from the calibrated mass spectrometry data. For example, the pre-defined characteristic metabolite identification information includes, but is not limited to: azelaic acid identification parameters of retention time (RT) of approximately 4.20 minutes and mass-to-charge ratio (m / z) of 187.0975 [MH]⁻; glycerophosphocholine identification parameters of retention time of approximately 3.15 minutes and mass-to-charge ratio (m / z) of 258.1100 [M+H]⁺; and L-arabinitol identification parameters of retention time of approximately 2.80 minutes and mass-to-charge ratio (m / z) of 153.0759 [M+H]⁺. These parameters are determined by previous non-targeted metabolomics studies and standard validation, and can be fine-tuned according to specific instruments and analytical conditions.

[0023] The extraction of signal strength specifically includes: Based on the retention time window and precise mass-to-charge ratio information associated with each characteristic metabolite, the corresponding chromatographic peak is located from the mass spectrometry chromatogram; the located chromatographic peak is integrated using an appropriate integration algorithm to calculate its peak area or peak height; the peak area or peak height value obtained by integration is used as the original signal intensity of the target characteristic peak.

[0024] The original signal intensity of the extracted target characteristic peak is corrected using the signal intensity of the internal standard corresponding to each characteristic metabolite to obtain the quantitative response value of the characteristic metabolite. The internal standard is preferably a stable isotope-labeled analogue, such as azelaic acid-d4 or glycerophosphate-choline-d9, and is added at the initial stage of sample pretreatment to correct for matrix effects and signal fluctuations during pretreatment and instrument analysis. The correction method involves dividing the signal intensity of the target characteristic peak by the signal intensity of its corresponding internal standard characteristic peak to obtain the relative response value after internal standard correction, which is used as the final quantitative response value.

[0025] S2. Based on the quantitative response values ​​of characteristic metabolites and combined with the growth stage information of the golden ear samples to be tested, determine the standardized diagnostic index relative to the health benchmark of the corresponding growth stage.

[0026] In a preferred embodiment, step S2 includes: First, obtain or determine the growth stage information of the *Auricularia auricula-judae* sample to be tested. The growth stage can be divided according to the appearance of the fruiting body (such as the primordia stage, extension stage, and maturity stage) or the number of days of cultivation after fruiting.

[0027] Preferably, growth stage information can be automatically inferred from the response patterns of specific "stage indicator metabolites" (such as certain amino acids, sterols, or sugars) in mass spectrometry detection data through a pre-trained lightweight classification model.

[0028] Based on growth stage information, the corresponding health benchmark statistics are retrieved from a pre-constructed segmented health metabolite benchmark library. The benchmark library is constructed by collecting a large number of healthy golden ear fungus samples covering different production areas, batches, and growth stages, detecting their metabolites using the standard method described in step S1, and then grouping and statistically analyzing them.

[0029] Specifically, obtain the following for this stage: the mean μA and standard deviation σA for azelaic acid health, the mean μG and standard deviation σG for glycerophosphocholine health, and the mean μC and standard deviation σC for L-arabinitol health. Simultaneously, obtain the mean μlog(R) and standard deviation σlog(R) of the log10(A / G) ratio (i.e., the logarithm of the ratio of the quantitative response values ​​of azelaic acid to glycerophosphocholine) for the healthy samples in this stage.

[0030] Then, standardized diagnostic indices are calculated; the quantitative response values ​​of characteristic metabolites of the test sample (obtained via S1) are compared with the health benchmark for the corresponding stage to eliminate the influence of differences in metabolic background at different growth stages, and the following core standardized diagnostic indices are calculated: First Index This reflects the standardized deviation from the balance between defense and damage. ; Where A and G are the quantitative response values ​​of azelaic acid and glycerophosphate choline in the test sample, respectively. and Health samples at the corresponding stages The mean and standard deviation of the values.

[0031] Understandably, the first index By standardizing the log10(A / G) values ​​of the test samples with the distribution of the healthy population (Z-scoreization), the degree to which the samples deviated from the healthy norm in the balance between defense signals (azelic acid) and membrane damage markers (glycerophosphate choline) was quantitatively characterized. A positive Zr value indicates a relatively strong defense response, while a negative value indicates that membrane damage is relatively dominant.

[0032] Second Index This reflects the standardized deviation of the osmotic regulation response. Where C is the quantitative response value of L-arabinol in the sample to be tested, and is the quantitative response value of L-arabinol.

[0033] Understandably, the second index By normalizing the L-arabinol response value using the Z-score, the degree to which the sample deviated from the healthy norm in the osmotic regulatory metabolic pathway was quantitatively characterized. A significantly positive Zc value generally indicates that the cell is experiencing osmotic or water stress.

[0034] Through step S2 above, this invention transforms the original absolute quantitative response value into relative diagnostic indices (Zr and Zc) that have clear statistical significance and biological interpretation and are adapted to the growth stage. These two indices together constitute the core feature vector for assessing the metabolic status of *Auricularia auricula-judae*, effectively eliminating the interference of inherent metabolic differences at different growth stages of *Auricularia auricula-judae* on the diagnostic results.

[0035] S3. Perform anomaly detection and pattern matching analysis on the standardized diagnostic index to obtain the first bacterial infection assessment information.

[0036] Please combine Figure 2 and Figure 3 In a preferred embodiment, step S3 includes: (1) First stage: Anomaly screening based on one-class support vector machine (SVM). The first exponent Zr and the second exponent Zc are used as two-dimensional features and input into a pre-trained pattern recognition model for anomaly state discrimination. The pattern recognition model is preferably a one-class support vector machine (SVM) model trained on a large and comprehensive dataset of healthy golden ear samples (their Zr and Zc values). This model constructs a closed decision region by learning the distribution boundary of healthy data in the feature space. If the [Zr, Zc] coordinates of the sample to be tested fall within this decision region, the model determines that its metabolic characteristics are within the range of healthy data distribution and then generates the first infection assessment information: "normal metabolic state"; this step can efficiently filter out the vast majority of healthy samples and achieve rapid initial screening.

[0037] (2) Second stage: Fine discrimination of bacterial infection type based on pattern matching; if a support vector machine model determines that the [Zr, Zc] coordinate point of the sample to be tested falls outside the health decision area, that is, it is determined that it deviates from the distribution range of health data, then the fine discrimination process is started.

[0038] First, the two-dimensional feature vector [Zr, Zc], composed of the first exponent Zr and the second exponent Zc, is matched with several predefined prototype vectors (Pk) representing typical bacterial infection patterns. Cosine similarity is preferred for similarity matching, as it measures the directional proximity of two vectors and is insensitive to their absolute length, making it suitable for assessing the similarity of metabolic response patterns. Specifically, the cosine similarity Sk between vector [Zr, Zc] and the k-th prototype vector Pk is calculated using the following formula: ; in, Represents the vector dot product. The Euclidean norm of a vector; The matching result is to obtain the prototype vector Pk corresponding to the maximum cosine similarity max(Sk) and its associated state label; The prototype vector Pk of a typical infection pattern is predefined through the following steps: Collect M historically confirmed bacterial infection samples, each sample i corresponding to a feature vector. .

[0039] For M feature vectors { Clustering analysis based on density peaks is performed; this algorithm can discover naturally formed sample clusters with similar metabolic response patterns in the data, without needing to pre-specify the number of clusters, and can effectively identify different bacterial metabolic phenotypes.

[0040] For the k-th cluster obtained from cluster analysis, the feature vectors of all samples within it are... The arithmetic mean of the vectors is used as the prototype vector Pk of the cluster, and the calculation formula is: ; in, Let be the number of samples belonging to the k-th cluster. It is the set of all samples in the k-th cluster; the prototype vector Pk represents the average metabolic characteristics or central tendency of this type of bacterial infection pattern.

[0041] Based on the main pathogen type (such as Trichoderma, Penicillium, etc.) or infection stage (such as early stage, active stage) corresponding to the samples in the kth cluster, assign the corresponding state label to the prototype vector Pk.

[0042] In a preferred embodiment, the status label includes at least one of the following: "defense response-dominant type" (corresponding to a significant increase in azelaic acid), "membrane damage-dominant type" (corresponding to a significant increase in glycerophosphate choline and an abnormal ratio), "osmotic stress response type" (corresponding to a significant increase in L-arabinol), or "mixed metabolic disorder".

[0043] Finally, the state label associated with the best prototype vector obtained from similarity matching is determined as the first infection assessment information. For example, outputting a defense response-dominated type of suspected infection or a membrane damage-dominated type of abnormal infection.

[0044] S4. Obtain the metabolic consistency information of the same batch of the cultivation unit to which the golden ear fungus sample belongs, and determine the second contamination assessment information based on the metabolic consistency information.

[0045] In a preferred embodiment, step S4 includes: (1)Obtain background population data: To evaluate the relative status of a test sample in its production environment, its context information needs to be obtained. Specifically, obtain the mass spectrometry detection data of at least N (N≥3) other Tremella aurantialba samples from the same cultivation unit (such as the same mushroom house, the same batch of spawn bags) and the same harvest cycle. These samples serve as the background population and represent the typical metabolic state range of this batch under the same management conditions.

[0046] (2)Calculate population metabolic characteristics: According to the mass spectrometry detection data of other Tremella aurantialba samples (background population), determine their respective first index Zr and second index Zc according to the same standard process as in the aforementioned S1 to S2. Thus, obtain the characteristic vector set of the background population: .

[0047] (3)To quantify the deviation degree of the metabolic pattern of the test sample from its background population, the present invention uses the Mahalanobis Distance as a consistency metric. The calculation process is as follows: First, based on the characteristic vector set of the background population, calculate its covariance matrix, which describes the variation degree of the two variables Zr and Zc in this population and their correlation.

[0048] Then, calculate the Mahalanobis Distance (MD) of the characteristic vector [Zr, Zc] of the test sample relative to the distribution of this background population. Its calculation formula is: ; where μ is the mean vector of the characteristic vectors of the background population, and Σ⁻¹ is the inverse matrix of the covariance matrix of the background population.

[0049] (4)Generate the second contamination assessment information: Based on the calculated magnitude of the Mahalanobis Distance MD, generate the second contamination assessment information characterizing the consistency degree of the metabolic pattern between the test sample and the population of the same batch. Specifically, by comparing the MD value with a preset threshold, its consistency degree can be divided into discrete levels.

[0050] For example, in a preferred embodiment: if MD ≤ 2.0, it is determined that the individual is highly consistent with the population metabolic pattern; if 2.0 < MD ≤ 4.0, it is determined that the individual slightly deviates from the population metabolic pattern; if MD > 4.0, it is determined that the individual significantly deviates from the population metabolic pattern. These thresholds can be set and optimized based on the distribution of the Mahalanobis Distance of the background population or actual empirical data.

[0051] S5. Determine the final diagnosis result of the contamination status of Tremella aurantialba according to the first contamination assessment information and the second contamination assessment information.

[0052] In a preferred embodiment, step S5 includes: The assessment status included in the first infection assessment information is mapped to a first confidence score (Score_A). For example, if the assessment status is "normal metabolic status", it can be mapped to a low score (e.g., 0.1); if it is "defense response-dominated suspicious", it can be mapped to a medium score (e.g., 0.5); if it is "membrane damage-dominated abnormal", it can be mapped to a high score (e.g., 0.8). The mapping rules can be set based on the correlation between different statuses and the final infection confirmation result in historical data.

[0053] The consistency level contained in the second infection assessment information is mapped to a second confidence score (Score_B). For example, "high consistency" is mapped to a low score (e.g., 0.1), "slight deviation" is mapped to a medium score (e.g., 0.4), and "significant deviation" is mapped to a high score (e.g., 0.9).

[0054] The first confidence score and the second confidence score are weighted and summed according to preset weighting coefficients to obtain the Composite Decision Index (CDI); for example, The weighting coefficients w1 and w2 can be optimized and determined through analysis of historical validation data.

[0055] The final diagnosis does not rely solely on a single comprehensive decision index value. Instead, it combines information from three dimensions—the specific status label type contained in the first infection assessment information (e.g., "defense response-dominated"), the numerical range of the calculated comprehensive decision index CDI (e.g., CDI < 0.3, 0.3 ≤ CDI < 0.6, CDI ≥ 0.6), and the consistency level of the second infection assessment information (e.g., "significant deviation")—as input keys to query a pre-constructed multidimensional decision rule table.

[0056] The multidimensional decision rule table is summarized and derived by analyzing historical cases (including information from the above three dimensions and the final results confirmed by culture or pathology). For example, an entry in the rule table can be defined as: IF (status label is "membrane damage dominant" AND CDI ≥ 0.7 AND consistency level is "significant deviation") THEN (final diagnosis = "highly suspected bacterial infection, immediate treatment recommended").

[0057] Based on the query matching results of the multidimensional decision rule table, output the corresponding and clear final diagnosis results and suggestions.

[0058] The language used to describe the final diagnosis should be clear and actionable, and may include different levels such as: "Confirmed healthy", "Metabolic status stable, routine management recommended", "Metabolic outlier, laboratory retest recommended", "Possible risk of infection, enhanced monitoring and retesting in the short term recommended", "Highly suspected infection, immediate isolation and prevention and control measures recommended".

[0059] According to the above embodiments, the method proposed in this invention first obtains precise quantitative values ​​of three characteristic metabolites—azelaic acid, glycerophosphate choline, and L-arabinitol—through optimized solvent extraction and targeted mass spectrometry analysis. Then, combining sample growth stage information, a standardized diagnostic index relative to the stage's health baseline is calculated, effectively eliminating the interference of growth differences on diagnosis. The index is processed through a two-stage analysis: first, an anomaly screening is performed using a support vector machine model trained on healthy data; then, for abnormal samples, similarity matching is performed between metabolic characteristics and typical patterns obtained from clustering historical contamination data to generate first contamination assessment information containing the tendency of specific contamination types. Simultaneously, population metabolic consistency analysis is introduced, and second contamination assessment information characterizing the degree of deviation in metabolic patterns is generated by calculating the Mahalanobis distance between the test sample and samples from the same batch. Finally, the two types of assessment information are fused, and a pre-constructed multidimensional decision rule table is queried to output a diagnostic result with clear guidance.

[0060] Example 2 Based on the same inventive concept, another preferred embodiment of the present invention proposes a diagnostic system for *Auricularia auricula-judae* infection based on the detection of characteristic metabolites, used to implement the steps of the rapid diagnostic method for *Auricularia auricula-judae* infection as described in Example 1. The system includes: The mass spectrometry data acquisition unit is used to acquire mass spectrometry detection data of the *Auricularia auricula-judae* sample to be tested; specifically, it includes ultra-high performance liquid chromatography-tandem mass spectrometry (UHPLC-MS / MS) hardware equipment and is equipped with control software to execute the rapid targeted detection method (such as specific gradient, ion pair monitoring mode) described in step S1 of Example 1 above.

[0061] The data preprocessing unit is used to determine the quantitative response values ​​of characteristic metabolites based on mass spectrometry detection data. The characteristic metabolites include azelaic acid, glycerophosphate choline, and L-arabinitol. This unit has a built-in quality control verification program, an automatic characteristic peak extraction and integration algorithm, and stores the identification information of characteristic metabolites (such as retention time and mass-to-charge ratio) and the corresponding internal standard information. It can perform internal standard correction and output the corrected quantitative response values.

[0062] The index calculation unit is used to determine a standardized diagnostic index relative to the corresponding growth stage health benchmark based on the quantitative response value of the characteristic metabolites and the growth stage information of the test golden ear sample. The unit is connected to or has the staged health metabolite benchmark library built in, and can receive growth stage information manually input or automatically inferred from the data preprocessing results, and calculate the Zr and Zc indices according to the formula in step S2 of Example 1.

[0063] The first evaluation unit is used to perform anomaly detection and pattern matching analysis on the standardized diagnostic index to obtain the first bacterial infection evaluation information. This unit stores a trained one-class support vector machine (SVM) model and a typical bacterial infection pattern prototype vector library, and can perform anomaly screening and pattern matching to output "normal metabolic state" or a specific bacterial infection type label.

[0064] The second evaluation unit is used to obtain the metabolic consistency information of the same batch of the cultivation unit to which the golden ear fungus sample belongs, and to determine the second contamination evaluation information based on the metabolic consistency information. This unit can access the detection data of other samples in the same batch, or call historical data of the same batch from the local database, calculate the Mahalanobis distance, and output the consistency level according to the preset threshold.

[0065] The decision fusion unit is used to determine the final diagnosis result of the fungal infection status of *Auricularia auricula-judae* based on the first and second fungal infection assessment information. This unit has built-in confidence mapping rules, weight coefficients and the multidimensional decision rule table, and can automatically perform quantitative fusion and rule query to generate and output a report containing the final diagnosis conclusion and recommendations.

[0066] It should be noted that each module in the above rapid diagnostic system corresponds to steps S1 to S5 in implementing the above rapid diagnostic system. The instances and application scenarios implemented by multiple modules and their corresponding steps are the same, but are not limited to the content disclosed in Embodiment 1 above.

[0067] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0068] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit it; although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of this application.

Claims

1. A diagnostic method for fungal infection of *Auricularia auricula-judae* based on the detection of characteristic metabolites, characterized in that, The method includes: Mass spectrometry data of the *Auricularia auricula-judae* sample to be tested are obtained, and quantitative response values ​​of characteristic metabolites are determined based on the mass spectrometry data. The characteristic metabolites include azelaic acid, glycerophosphate choline, and L-arabinitol. Based on the quantitative response values ​​of the characteristic metabolites and the growth stage information of the test golden ear samples, a standardized diagnostic index relative to the health benchmark of the corresponding growth stage is determined. Anomaly detection and pattern matching analysis were performed on the standardized diagnostic index to obtain the first bacterial infection assessment information; Obtain the metabolic consistency information of the same batch of the cultivation unit to which the test golden ear sample belongs, and determine the second contamination assessment information based on the metabolic consistency information; Based on the first and second bacterial infection assessment information, the final diagnostic result of the bacterial infection status of *Auricularia auricula-judae* is determined.

2. The method for diagnosing fungal infection in *Auricularia auricula-judae* based on the detection of characteristic metabolites according to claim 1, characterized in that, The step of determining the quantitative response value of the characteristic metabolite based on the mass spectrometry detection data includes: The mass spectrometry detection data were subjected to quality control verification; the mass spectrometry detection data was obtained from the test *Auricularia auricula-judae* sample by ultra-high performance liquid chromatography-tandem mass spectrometry. Based on the preset characteristic metabolite identification information, the signal intensity of the corresponding target characteristic peak is extracted from the calibrated mass spectrometry data; The signal intensity of the extracted target feature peak is corrected by using the signal intensity of the internal standard corresponding to each characteristic metabolite to obtain the quantitative response value of the target metabolite.

3. The method for diagnosing fungal infection in *Auricularia auricula-judae* based on the detection of characteristic metabolites according to claim 2, characterized in that, The extraction of signal intensity based on preset characteristic metabolite identification information includes: Based on the retention time window and precise mass-to-charge ratio information associated with each characteristic metabolite, the corresponding chromatographic peaks are located from the mass spectrometry chromatogram; Integrate the located chromatographic peaks and calculate their peak area or peak height; The peak area or peak height obtained by integration is used as the signal strength of the target characteristic peak.

4. The method for diagnosing fungal infection in *Auricularia auricula-judae* based on the detection of characteristic metabolites according to claim 1, characterized in that, The process of determining standardized diagnostic indices relative to the corresponding growth stage health benchmark by combining the growth stage information of the tested golden ear samples includes: Based on the growth stage information, the average healthy azelaic acid value corresponding to that stage is obtained from a pre-constructed segmented healthy metabolite benchmark library. with standard deviation glycerophosphate choline healthy mean with standard deviation and the average health value of L-arabinitol with standard deviation ; Calculate the following standardized diagnostic indices: First Index , ; in, and Health samples at the corresponding stages The mean and standard deviation of the values; Second Index , , where is the quantitative response value of L-arabinol.

5. The method for diagnosing fungal infection in *Auricularia auricula-judae* based on the detection of characteristic metabolites according to claim 4, characterized in that, The step of performing anomaly detection and pattern matching analysis on the standardized diagnostic index to obtain the first bacterial infection assessment information includes: The first index Zr and the second index Zc are input into a pre-trained pattern recognition model for abnormal state discrimination. The pattern recognition model is a support vector machine model trained based on healthy golden ear sample data; If the pattern recognition model determines that the first index Zr and the second index Zc are within the range of healthy data distribution, then the first bacterial contamination assessment information is generated to characterize the normal metabolic state of the sample. If the pattern recognition model determines that the first index Zr and the second index Zc deviate from the distribution range of health data, then the vector [Zr, Zc] formed by the first index Zr and the second index Zc is matched with the similarity of multiple predefined typical bacterial infection pattern prototype vectors. The state label associated with the prototype vector of the most similar typical bacterial infection pattern is determined as the first bacterial infection assessment information.

6. The method for diagnosing fungal infection in *Auricularia auricula-judae* based on the detection of characteristic metabolites according to claim 5, characterized in that, The step of matching the vector [Zr, Zc] formed by the first index Zr and the second index Zc with the similarity vectors of multiple predefined typical bacterial infection patterns includes: The cosine similarity Sk between the vector [Zr, Zc] and the prototype vector Pk of the k-th typical bacterial infection pattern is calculated as follows: ； in, Represents the vector dot product. The Euclidean norm of a vector; The matching result is to obtain the prototype vector Pk corresponding to the maximum cosine similarity max(Sk) and its associated state label; The typical bacterial infection pattern prototype vector Pk is predefined through the following steps: Collect M historically confirmed bacterial infection samples, each sample i corresponding to a feature vector. ; For the M feature vectors { Perform cluster analysis based on density peaks; For the k-th cluster obtained from cluster analysis, its prototype vector Pk is calculated as follows: ； in, Let be the number of samples belonging to the k-th cluster. Let be the set of all samples in the k-th cluster; Assign a corresponding state label to the prototype vector Pk based on the main pathogen type or infection stage corresponding to the sample in the k-th cluster.

7. The method for diagnosing fungal infection in *Auricularia auricula-judae* based on the detection of characteristic metabolites according to claim 6, characterized in that, The status label includes at least one of the following: "defense response-dominated", "membrane damage-dominated", "osmotic stress response", or "mixed metabolic disorder".

8. The method for diagnosing fungal infection in *Auricularia auricula-judae* based on the detection of characteristic metabolites according to claim 6, characterized in that, Obtain the metabolic consistency information of the same batch population within the cultivation unit to which the *Auricularia auricula-judae* sample belongs, and determine the second contamination assessment information based on the metabolic consistency information, including: Obtain mass spectrometry data from at least N other golden ear fungus samples from the same cultivation unit and the same harvesting cycle, where N≥3; Based on the mass spectrometry data of the other golden ear samples, their respective first index Zr and second index Zc are determined; Calculate the covariance matrix of the dataset consisting of the [Zr, Zc] vectors of the other golden ear samples, and calculate the Mahalanobis distance of the [Zr, Zc] vector of the golden ear sample to be tested relative to the dataset based on this covariance matrix. Based on the magnitude of the Mahalanobis distance, a second bacterial contamination assessment is generated, characterizing the degree of consistency between the metabolic patterns of the sample under test and the same batch of population.

9. The method for diagnosing fungal infection in *Auricularia auricula-judae* based on the detection of characteristic metabolites according to claim 8, characterized in that, The final diagnostic result for determining the fungal infection status of *Auricularia auricula-judae* based on the first and second fungal infection assessment information includes: The assessment status contained in the first contamination assessment information is mapped to a first confidence score, and the consistency level contained in the second contamination assessment information is mapped to a second confidence score. The first confidence score and the second confidence score are weighted and summed according to preset weighting coefficients to obtain a comprehensive decision index; Based on the status label type contained in the first contamination assessment information, the numerical range of the comprehensive decision index, and the consistency level of the second contamination assessment information, query the pre-constructed multidimensional decision rule table; Based on the query results of the multidimensional decision rule table, the corresponding final diagnostic results are output.

10. A diagnostic system for fungal infection of *Auricularia auricula-judae* based on the detection of characteristic metabolites, used to implement the steps of the rapid diagnostic method for fungal infection of *Auricularia auricula-judae* as described in any one of claims 1-9, characterized in that, The system includes: The mass spectrometry data acquisition unit is used to acquire mass spectrometry detection data of the *Eurydon edulis* sample to be tested. A data preprocessing unit is used to determine the quantitative response value of a characteristic metabolite based on the mass spectrometry detection data, wherein the characteristic metabolite includes azelaic acid, glycerophosphate choline, and L-arabinitol; The index calculation unit is used to determine a standardized diagnostic index relative to the health benchmark of the corresponding growth stage based on the quantitative response value of the characteristic metabolite and the growth stage information of the test golden ear sample. The first assessment unit is used to perform anomaly detection and pattern matching analysis on the standardized diagnostic index to obtain the first bacterial infection assessment information. The second evaluation unit is used to obtain the metabolic consistency information of the same batch of the cultivation unit to which the test golden ear sample belongs, and to determine the second contamination evaluation information based on the metabolic consistency information. The decision fusion unit is used to determine the final diagnostic result of the fungal infection status of *Auricularia auricula-judae* based on the first fungal infection assessment information and the second fungal infection assessment information.

Images