Automatic generation method and device for pet allergen analysis reports

Abstract

This application relates to the field of data processing technology and discloses a method and apparatus for automatically generating pet allergen analysis reports. The method includes: constructing an allergen association knowledge graph storing homology relationships, cross-reaction relationships, and co-exposure relationships and corresponding weights; acquiring in vitro allergen detection data from pets, quantifying risk levels into risk scores, constructing association subgraphs based on association edges, and clustering to obtain one or more allergen clusters; for each allergen cluster, calculating the sum of risk scores within the cluster as a base risk value, calculating the sum of the products of the association weights between each allergen pair within the cluster and the smaller of the two risk scores as a synergistic contribution value, and using the product of the base risk value and the synergistic contribution value as a synergistic risk index to derive the synergistic exposure risk; and generating a comprehensive report. This invention, by constructing allergen clusters and quantifying synergistic exposure risks, solves the problems of isolated information and neglected superposition effects of weak positive signals in existing reports, thus improving clinical guidance.

Description

Technical Field

[0001] This application relates to the field of data processing technology, specifically to a method and apparatus for automatically generating pet allergen analysis reports. Background Technology

[0002] Pet allergies are becoming increasingly common in clinical practice. Accurate identification of allergens and the development of targeted intervention plans are crucial for improving treatment outcomes. Currently, pet allergen testing primarily utilizes immunological techniques such as chemiluminescence immunoassay, which can simultaneously detect dozens to hundreds of environmental and food allergens. After testing, the system automatically generates a report, listing each allergen in tabular form, along with its risk level and qualitative assessment (negative / weakly positive / positive / strongly positive). Figure 1 Example shown.

[0003] Existing reports typically only interpret and present each allergen independently, failing to consider the biological connections between multiple allergens. For example, house dust mites and flour mites belong to the same mite family; house dust mites and shrimp exhibit cross-reactivity; and indoor mold and dust mites are commonly found in the same exposure scenarios. When multiple related allergens show positive or weakly positive results simultaneously, especially when multiple weakly positive signals are superimposed, the cumulative risk of synergistic exposure is often significantly higher than the risk level indicated by a single indicator, sufficient to trigger clinical symptoms. However, existing reports lack a mechanism for aggregated analysis of related allergens, failing to reflect the risk of synergistic exposure. This leads to the neglect of the cumulative effect of weakly positive signals, resulting in an underestimation of the actual risk.

[0004] Therefore, there is an urgent need for a report generation method that can identify allergen associations, automatically mine allergen clusters, and generate comprehensive risk alerts to solve the technical problems of isolated report information, missed detection of weak signals, and insufficient clinical guidance in existing technologies. Summary of the Invention

[0005] To address the technical problems mentioned in the background section, the purpose of this application is to provide a method, apparatus, and computer device for automatically generating pet allergen analysis reports.

[0006] According to the first aspect of this application, a method for automatically generating a pet allergen analysis report is provided, comprising the following steps:

[0007] S1, construct an allergen association knowledge graph that stores the association relationships between different allergens and their corresponding association weights; wherein, the association relationships include homology relationships, cross-reaction relationships, and co-exposure relationships;

[0008] S2, acquire pet in vitro allergen detection data, including allergen names and their risk levels, quantify the risk levels into risk scores, treat all detected allergens as nodes, construct a correlation subgraph based on the correlation edges between nodes with a correlation weight greater than a threshold, and cluster the correlation subgraph to obtain one or more allergen clusters.

[0009] S3. For each allergen cluster, calculate the sum of the risk scores of each allergen within the cluster as the base risk value, calculate the sum of the products of the association weight between each pair of allergens within the cluster and the smaller of the two risk scores as the synergistic contribution value, and use the product of the base risk value and the synergistic contribution value as the synergistic risk index. Compare the synergistic risk index with the risk threshold to obtain the synergistic exposure risk.

[0010] S4. Generate and output a comprehensive report based on the co-exposure risk, including a list of allergen test results and cluster risk warning information for allergen clusters with co-exposure risk.

[0011] According to a second aspect of this application, an automatic generation device for pet allergen analysis reports is provided, the device comprising:

[0012] The knowledge graph construction module is used to construct an allergen association knowledge graph that stores the association relationships between different allergens and their corresponding association weights; wherein, the association relationships include homology relationships, cross-reaction relationships, and co-exposure relationships;

[0013] The data acquisition and clustering module is used to acquire pet in vitro allergen detection data, including allergen names and their risk levels, quantify the risk levels into risk scores, treat all detected allergens as nodes, construct a correlation subgraph based on the correlation edges between nodes with a correlation weight greater than a threshold, and cluster the correlation subgraph to obtain one or more allergen clusters.

[0014] The collaborative risk calculation module is used to calculate the sum of the risk scores of each allergen within each allergen cluster as the base risk value, calculate the sum of the products of the association weight between each pair of allergens within the cluster and the smaller of the two risk scores as the collaborative contribution value, and use the product of the base risk value and the collaborative contribution value as the collaborative risk index. The collaborative risk index is compared with the risk threshold to obtain the collaborative exposure risk.

[0015] The report generation module is used to generate and output a comprehensive report based on the co-exposure risks, including a list of allergen detection results and cluster risk warning information for allergen clusters with co-exposure risks.

[0016] According to a third aspect of this application, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program that, when executed by the processor, implements the method as described in any of the preceding claims.

[0017] Compared with the prior art, the present invention has the following beneficial technical effects:

[0018] (1) This invention utilizes a knowledge graph to associate previously isolated allergens into allergen clusters based on homology, cross-reactivity, and co-exposure relationships, enabling the superposition effect of multiple weak positive signals to be quantified from two dimensions: basic risk and synergistic contribution. This can solve the technical problem of underestimation of synergistic exposure risk due to independent interpretation in existing reports;

[0019] (2) By using differentiated pruning based on association type, unreasonable merging caused by excessively long indirect association chains can be prevented. Furthermore, by using cluster size correction coefficients, the synergistic risk indices between clusters of different sizes can be made comparable, thereby providing a more accurate and instructive comprehensive risk assessment for clinical use while preserving the original detection information. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0021] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0022] Figure 1 This is a schematic diagram of a routine pet allergen analysis report.

[0023] Figure 2 This is a flowchart illustrating a method for automatically generating a pet allergen analysis report, as provided in an embodiment of this application.

[0024] Figure 3 This is a schematic diagram of an automatic pet allergen analysis report generation device provided in an embodiment of this application.

[0025] Figure 4 This is a schematic diagram of the structure of the knowledge graph construction module 10 provided in an embodiment of this application.

[0026] Figure 5 This is a schematic diagram of the structure of the data acquisition and clustering module 20 provided in an embodiment of this application.

[0027] Figure 6 This is a schematic diagram of the structure of the collaborative risk calculation module 30 provided in an embodiment of this application. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0029] like Figure 2 As shown in the figure, this application discloses an automatic generation method 100 for pet allergen analysis reports, including the following steps:

[0030] S1, construct an allergen association knowledge graph that stores the association relationships between different allergens and their corresponding association weights; wherein, the association relationships include homology relationships, cross-reaction relationships, and co-exposure relationships;

[0031] Pet allergies present with diverse clinical manifestations, affecting multiple systems including the skin, respiratory tract, and digestive tract. For example, allergic dermatitis is common in pets, with symptoms including itching, rashes, hair loss, and lichenification; respiratory allergies manifest as sneezing, conjunctivitis, and bronchitis; and digestive allergies present as vomiting and diarrhea. While these symptoms may be triggered by a single allergen, clinical practice has shown that they are more often the result of co-exposure to multiple allergens. For instance, when pets are simultaneously exposed to multiple mites such as house dust mites, flour mites, and tropical clawless mites, the sensitization effect is often significantly higher than that of a single mite. Furthermore, pets allergic to house dust mites may also exhibit cross-reactions to crustaceans such as shrimp and crab. Additionally, indoor mold and house dust mites are commonly found in the same humid environment, exhibiting co-exposure characteristics, and their simultaneous presence easily induces respiratory symptoms. Individual analysis of a single allergen cannot reflect these relationships; therefore, this embodiment explicitly stores these relationships by constructing a knowledge graph.

[0032] In some embodiments, an allergen association knowledge graph is constructed that stores the association relationships between different allergens and their corresponding association weights, including:

[0033] S11. Based on allergen taxonomic information, immunological cross-reactivity literature, and environmental co-exposure statistics, homology subsets, cross-reactivity subsets, and co-exposure subsets are constructed respectively.

[0034] The association weights in the homology subset are set according to the kinship distance of the classification unit to which the allergen belongs;

[0035] Taking dust mite allergens as an example, the testing items include house dust mites, flour mites, tropical clawless mites, and rough-legged flour mites. According to biological taxonomic information, house dust mites and flour mites belong to the same genus *Dictyophora* and are very closely related, so their association weight can be set to 0.9; while house dust mites and tropical clawless mites, although both belong to the mite family, belong to different families and are more distantly related, so their association weight can be set to 0.6; rough-legged flour mites belong to the family Flouridae and are even more distantly related to the family *Dictyophora*, so their association weight can be set to 0.5.

[0036] For example, for cereal allergens, the testing items include wheat, barley, oats, corn, and rice. Wheat, barley, and oats all belong to the Poaceae family, Pooideae subfamily, and are closely related, so a homology weight of 0.8 can be set; although corn is also Poaceae, it belongs to the Sorghum subfamily, and is relatively distantly related to wheat, so an association weight of 0.5 can be set; rice (paddy rice) and wheat both belong to Poaceae but belong to different subfamilies, so an association weight of 0.5 can be set.

[0037] For example, nut allergens can be tested for peanuts, walnuts, pecans, almonds, cashews, etc. Peanuts belong to the legume family, walnuts and pecans belong to the jujube family, almonds belong to the Rosaceae family, and cashews belong to the Anacardiaceae family. They are not closely related, so the weight of homology can be set at 0.3-0.4.

[0038] The association weights in the cross-reactivity subsets are set based on antigen epitope similarity or the cross-reactivity strength reported in the literature.

[0039] Taking the cross-reactivity between mites and crustaceans as an example, the mites tested include house dust mites and flour mites, while the crustaceans include shrimp and crab. Based on immunological literature reports, the house dust mite Der p1 antigen and shrimp tropomyosin have structural similarity and a high cross-reactivity. Therefore, the association weight between house dust mites and shrimp can be set at 0.7, and the association weight between house dust mites and crab can be set at 0.6.

[0040] For example, the cross-reactivity between pollen and fruit can be tested. Plant pollen samples include wormwood, amaranth, and Bermuda grass, while fruit samples include apples, apricots, mangoes, and pineapples. Immunological literature reports cross-reactivity between wormwood pollen and fruits such as apples and peaches. Therefore, the association weight between wormwood and apples can be set at 0.6, and the association weight between wormwood and apricots at 0.5.

[0041] For example, cross-reactivity between pet dander samples can be tested, including cat dander, dog dander, and cow dander. Since cats, dogs, and cows are all mammals, the main allergenic proteins in their dander share some homology. Therefore, the association weight between cat dander and dog dander can be set at 0.5, and the association weight between cat dander and cow dander can be set at 0.4.

[0042] The association weights in the co-exposure relationship subset are set based on the environmental co-occurrence frequency or the epidemiological co-positive probability;

[0043] Taking the co-exposure of house dust mites and mold as an example, the fungi tested included Candida albicans, Penicillium, Aspergillus fumigatus, and Alternaria alternifolia. According to indoor environmental survey data, Aspergillus fumigatus and house dust mites have a high co-occurrence frequency in the same humid household environment. Therefore, the association weight between Aspergillus fumigatus and house dust mites can be set at 0.5, and the association weight between Alternaria alternifolia and house dust mites can be set at 0.4.

[0044] Taking co-exposure to pollen allergens as an example, Artemisia and Amaranthus spinosa are common pollen in summer and autumn. Their flowering periods overlap in the same area, and the frequency of co-exposure is high. The association weight between Artemisia and Amaranthus spinosa can be set to 0.6.

[0045] Taking co-exposure to food allergens as an example, common meats such as chicken, beef, and pork often appear together in pet food, with a high probability of co-exposure. The association weight between chicken and beef can be set to 0.5, and the association weight between chicken and pork can be set to 0.4.

[0046] Taking co-exposure to indoor allergens as an example, cat dander and dog dander often coexist in the same home environment, and the association weight can be set to 0.5.

[0047] S12, combine the above three subsets to form the allergen association knowledge graph, and store it as an undirected weighted graph structure. In the graph, nodes correspond to allergens, edges correspond to association relationships, and edge weights are the corresponding association weights.

[0048] During the merging process, if multiple types of associations exist between the same pair of allergens, the maximum value among the association weights corresponding to each association is taken as the final weight of that edge. For example, house dust mites and flour mites may have both a homology relationship (weight 0.9) and a co-exposure relationship (weight 0.5), so the final edge weight is the maximum value of 0.9. If only one type of association exists between the same pair of allergens, the weight corresponding to that association is directly used.

[0049] The resulting allergen association knowledge graph is stored in an undirected weighted graph structure. Each node in the graph uniquely corresponds to an allergen (such as "house dust mite", "shrimp", "wheat", etc.). Each edge connects two allergen nodes with an association relationship, and the weight value of the edge reflects the strength of the association relationship.

[0050] S2, acquire pet in vitro allergen detection data, including allergen names and their risk levels, quantify the risk levels into risk scores, treat all detected allergens as nodes, construct a correlation subgraph based on the correlation edges between nodes with a correlation weight greater than a threshold, and cluster the correlation subgraph to obtain one or more allergen clusters.

[0051] In this embodiment, the in vitro allergen detection data for pets can be obtained using chemiluminescence immunoassay. The test report lists the allergen name, test result, and allergy risk level in tabular form. To facilitate quantitative calculation, this embodiment maps the risk level to a risk score. For example, a negative result is quantified as 0 points, a weak positive result as 1 point, a positive result as 2 points, and a strong positive result as 3 points. It should be understood that the above quantification method is only an example, and the score mapping relationship can be adjusted according to clinical needs in actual applications.

[0052] In some embodiments, all detected allergens are treated as nodes, and a subgraph of association is constructed based on the association edges between nodes whose association weights are greater than a threshold. The subgraph of association is then clustered to obtain one or more allergen clusters, including:

[0053] S21, take all detected allergens as nodes, extract the association edge between any two nodes from the allergen association knowledge graph, and when the weight of the association edge is greater than or equal to the association weight threshold, include the association edge in the association edge set.

[0054] Specifically, all allergens appearing in the test report are used as nodes. For example, the allergens detected include house dust mites, flour mites, tropical clawless mites, rough-legged flour mites, wormwood, amaranth, chicken, beef, corn, wheat, rice, sesame, flaxseed, etc.

[0055] From the aforementioned allergen association knowledge graph, for each pair of allergens in the nodes, we search for the existence of an association edge and its corresponding association weight. Considering that edges with low association weights may correspond to weak associations, including all of them could lead to excessive merging of allergen clusters. Therefore, this embodiment sets an association weight threshold (e.g., 0.3). An association edge is only included in the association edge set if its weight is greater than or equal to this threshold. If the weight of an association edge is lower than the threshold, it is considered a clinically insignificant association and is not included.

[0056] S22, construct an associated subgraph based on the node and the associated edge set, and use a connected component algorithm to cluster the associated subgraph, grouping all allergens belonging to the same connected component into one allergen cluster, to obtain the one or more allergen clusters.

[0057] Specifically, based on a defined set of nodes and a set of associated edges, an undirected graph, namely the association subgraph, is constructed. In this subgraph, nodes represent detected allergens, and edges represent associations between two allergens that are greater than or equal to an association weight threshold.

[0058] The connected component algorithm is used to perform cluster analysis on the aforementioned associated subgraphs. The connected component algorithm can identify all maximal connected subgraphs in the graph, where nodes belonging to the same connected component exist at least once along a path formed by an associated edge, thus grouping allergens with direct or indirect relationships together. For example, if there is an associated edge between house dust mites and flour mites, and between flour mites and tropical clawless mites, then house dust mites, flour mites, and tropical clawless mites will belong to the same connected component and be classified into the same allergen cluster.

[0059] In some embodiments, a connected component algorithm is used to cluster the associated subgraph, grouping all allergens belonging to the same connected component into one allergen cluster, resulting in one or more allergen clusters, including:

[0060] S221, obtain the association type of each associated edge in the associated subgraph, and set the corresponding edge retention threshold according to the association type, wherein the edge retention threshold corresponding to the homo-origin relationship is the lowest, and the edge retention threshold corresponding to the cross-reaction relationship is the highest.

[0061] Specifically, each association edge in the association subgraph is traversed to obtain the association type of that edge. Association types include homology, cross-reaction, and co-exposure.

[0062] Since homology reflects biological taxonomic kinship, its clinical relevance is the most direct, and it should be retained even if the weight is low. Therefore, the edge retention threshold corresponding to homology is set to the lowest. For example, the edge retention threshold for homology is set to 0.2, meaning that homology edges with an association weight greater than or equal to 0.2 are retained, while those with a weight lower than 0.2 are removed.

[0063] Cross-reactivity relationships depend on specific antigenic epitope similarity and require high weights to ensure clinical significance. Therefore, the edge retention threshold for cross-reactivity relationships is set to be the highest. For example, setting the edge retention threshold for cross-reactivity relationships to 0.5 means that an edge is only retained if the association weight of the cross-reactivity relationship is greater than or equal to 0.5.

[0064] Co-exposure relationships reflect environmental statistical regularities or epidemiological co-positive probabilities. Their clinical relevance lies between homology and cross-reactivity relationships. Therefore, setting a retention threshold for edges corresponding to co-exposure relationships is the next best approach. For example, setting the edge retention threshold for co-exposure relationships to 0.3 means that edges of co-exposure relationships with an association weight greater than or equal to 0.3 are retained, while those with a weight lower than 0.3 are removed.

[0065] It should be understood that the above thresholds (homology 0.2, co-exposure 0.3, cross-reactivity 0.5) are merely illustrative settings and can be dynamically adjusted based on clinical validation results, population characteristics, or allergen categories in practical applications. For example, for certain specific allergen categories (such as food allergens), the retention threshold for co-exposure relationships can be appropriately increased to reduce false pairings; for allergen pairs of high clinical concern, the retention threshold can be appropriately decreased to improve sensitivity.

[0066] S222, traverse each associated edge in the associated subgraph. When the associated weight of the associated edge is greater than or equal to the edge retention threshold corresponding to the associated type, retain the associated edge; otherwise, remove the associated edge from the associated subgraph to obtain the pruned subgraph.

[0067] Specifically, the association weight of each obtained association edge is compared with the edge retention threshold corresponding to that association type. If the association weight is greater than or equal to the edge retention threshold, the edge is retained; if the association weight is less than the edge retention threshold, the edge is removed from the association subgraph.

[0068] For example, the association weight between house dust mites and flour dust mites is 0.9, which is greater than the homology retention threshold of 0.2, so it is retained; the association weight between house dust mites and tropical clawless mites is 0.6, which is greater than 0.2, so it is retained; the association weight between flour dust mites and tropical clawless mites is 0.6, which is greater than 0.2, so it is retained; the association weight between house dust mites and shrimp is 0.4, which is less than the cross-reactivity retention threshold of 0.5, so it is removed; the association weight between wormwood and prickly amaranth is 0.6, which is greater than the co-exposure retention threshold of 0.3, so it is retained; the association weight between chicken and beef is 0.5, which is greater than 0.3, so it is retained; the association weight between corn and wheat is 0.5, which is greater than 0.2, so it is retained; the association weight between wheat and rice is 0.5, which is greater than 0.2, so it is retained; the association weight between sesame and flaxseed is 0.8, which is greater than 0.2, so it is retained.

[0069] After the above pruning process, the pruned sub-image is obtained.

[0070] Through the above pruning process, edges with association weights lower than the retention threshold for their corresponding types are removed, thus severing weak links in indirect association chains. For example, if there exists an indirect association chain "house dust mite-shrimp-crab", and the cross-reaction edge between house dust mite and shrimp is removed because its weight is lower than the threshold, then even if there is an indirect association between house dust mite and crab, they cannot form a connected path through shrimp, thereby avoiding unreasonable merging caused by excessively long indirect association chains.

[0071] S223, The pruned subgraph is clustered using the connected component algorithm, and all allergens belonging to the same connected component are grouped into one allergen cluster.

[0072] Specifically, the basic principle of the connected component algorithm is: starting from any unvisited node, visit all nodes connected by edges through depth-first traversal or breadth-first traversal. These nodes constitute a connected component; repeat the above process until all nodes have been visited. Each connected component corresponds to an allergen cluster.

[0073] The clustering process of the connected component algorithm is described in detail, for example:

[0074] Suppose that a pet allergen test result was positive or weakly positive for the following allergens: house dust mite, flour mite, tropical clawless mite, rough-legged flour mite, wormwood, prickly amaranth, chicken, beef, corn, wheat, rice, sesame, and flaxseed. After pruning, the remaining associated edges are shown in Table 1 below:

[0075] Table 1

[0076]

[0077] These nodes and edges form the pruned subgraph, which is then clustered using the connected component algorithm, as detailed below:

[0078] 1. Initialization: Mark all nodes as unvisited.

[0079] 2. Starting with the first node (house dust mites):

[0080] Access the house dust mite and add it to the current connected component.

[0081] Find its adjacent nodes by edges: dust mite, tropical clawless mite, coarse-legged dust mite.

[0082] Recursively visit house dust mites: if the adjacent nodes have house dust mites (visited) and tropical clawless mites (unvisited), add tropical clawless mites.

[0083] Visiting Tropical Clawless Mites: Adjacent nodes contain House Dust Mites and Flour Dust Mites (both have been visited).

[0084] Visiting the coarse-legged dust mite: The adjacent node has house dust mites (visited).

[0085] The current connected component contains {house dust mite, flour mite, tropical clawless mite, rough-legged flour mite}, denoted as cluster 1.

[0086] 3. Continue from the unvisited node (Artemisia):

[0087] Visit Artemisia and add a new connected component.

[0088] Adjacent node: Amaranth (unvisited), added.

[0089] Visiting Amaranth: The only adjacent node is Artemisia (already visited).

[0090] The current connected component contains {Artemisia argyi, Amaranth}, and is denoted as cluster 2.

[0091] 4. Starting from the never-visited node (chicken):

[0092] Access the chicken and add a new connected component.

[0093] Adjacent node: Beef (unvisited), add.

[0094] Visiting beef: The only adjacent node is chicken (already visited).

[0095] The current connected component contains {chicken, beef}, and is denoted as cluster 3.

[0096] 5. Starting from the unvisited node (corn):

[0097] Visit the corn node and add a new connected component.

[0098] Adjacent node: Wheat (unvisited), added.

[0099] Visit wheat: If the adjacent nodes are corn (visited) and rice (unvisited), add rice.

[0100] Visiting Rice: The only adjacent node is Wheat (already visited).

[0101] The current connected component contains {corn, wheat, rice}, and is denoted as cluster 4.

[0102] 6. Starting from the unvisited node (Sesame):

[0103] Visit Sesame and add a new connected component.

[0104] Adjacent node: Flaxseed (unvisited), added.

[0105] Visiting Flaxseed: Its only adjacent node is Sesame (already visited).

[0106] The current connected component contains {sesame seeds, flax seeds}, and is denoted as cluster 5.

[0107] All nodes have been visited, and five allergen clusters were finally obtained: mite allergen clusters (house dust mite, dry dust mite, tropical clawless mite, rough-legged white dust mite), pollen allergen clusters (wormwood, prickly amaranth), meat allergen clusters (chicken, beef), grain allergen clusters (corn, wheat, rice), and oilseed allergen clusters (sesame, flaxseed).

[0108] Through the clustering process of the connected component algorithm described above, this embodiment aggregates allergens with direct or indirect relationships in the pruned subgraph into independent allergen clusters. This clustering method ensures that there is at least one path consisting of a related edge between any two allergens within the same cluster, thus grouping biologically related allergens into the same cluster. At the same time, because the pruning process removes key edges with weights below the threshold, weak links in indirect relationship chains are cut off, effectively preventing unreasonable merging due to excessively long indirect chains, thereby ensuring that the division of allergen clusters is more in line with clinical practice.

[0109] S3. For each allergen cluster, calculate the sum of the risk scores of each allergen within the cluster as the base risk value, calculate the sum of the products of the association weight between each pair of allergens within the cluster and the smaller of the two risk scores as the synergistic contribution value, and use the product of the base risk value and the synergistic contribution value as the synergistic risk index. Compare the synergistic risk index with the risk threshold to obtain the synergistic exposure risk.

[0110] In this embodiment, the risk score of a single allergen only reflects the independent sensitization level of that allergen. However, when multiple related allergens coexist, their synergistic exposure risk is often higher than the risk level indicated by a single indicator. For example, house dust mites are strongly positive (risk score 3), house dust mites are positive (risk score 2), tropical clawless mites are positive (risk score 2), and rough-legged white mites are positive (risk score 2). If judged independently, each allergen would be identified as positive or strongly positive. However, if multiple mite allergens appear simultaneously, their synergistic exposure may trigger more severe clinical symptoms. Similarly, corn is weakly positive (risk score 1), wheat is positive (risk score 2), and rice is positive (risk score 2). A single weak positive signal is easily overlooked, but since all three belong to the cereal family and have a common origin, their synergistic exposure risk may be sufficient to trigger clinical symptoms. Therefore, this embodiment designs the following synergistic risk index calculation method to quantify the synergistic exposure risk of related allergens, specifically:

[0111] For each allergen cluster obtained from clustering, let the cluster contain n allergens, and let the risk score of each allergen be denoted as . .

[0112] Calculate the base risk value Defined as the sum of the risk scores of each allergen within a cluster:

[0113]

[0114] Understandably, the baseline risk value reflects the cumulative effect of the independent sensitization levels of each allergen within a cluster.

[0115] Calculate the collaborative contribution value It is defined as the sum of the products of the association weights among all allergen pairs within a cluster and the smaller of the two risk scores:

[0116]

[0117] in, The association weights between allergen i and allergen j are obtained from the allergen association knowledge graph.

[0118] Based on the above calculation formula, the synergistic effect of two allergens is limited by the one with the lower risk (i.e., the weakest link effect), and the greater the association weight, the greater the synergistic contribution. By summing all allergen pairs within a cluster, the contribution of multiple associations within the cluster to the synergistic risk can be comprehensively reflected.

[0119] Next, calculate the collaborative risk index. Defined as the collaborative contribution value multiplied by the base risk value:

[0120]

[0121] Based on the above calculation formula, the higher the basic risk value, the more significant the amplification effect of synergistic contribution; the higher the synergistic contribution value, the stronger the cumulative effect of basic risk.

[0122] In some embodiments, the product of the base risk value and the collaborative contribution value is used as the collaborative risk index, including:

[0123] S31, obtain the number of allergens in the allergen cluster, and determine the cluster size correction coefficient based on the number of allergens and the corresponding negative correlation.

[0124] Specifically, cluster size correction coefficient It is negatively correlated with the number of allergens n, meaning that the more allergens in a cluster, the lower the marginal effect of individual synergistic contributions. Therefore, it is necessary to suppress the synergistic risk index to avoid the risk index being artificially high due to the excessive cluster size.

[0125] Preferably, this embodiment adopts As a cluster size correction coefficient, this function's form aligns with clinical experience: as the number of allergens within a cluster increases from 1 to 4, the correction coefficient decreases from 1 to 0.5, reflecting the law of diminishing marginal returns. In other embodiments, [the following can also be used] The coefficient values ​​can be determined by looking up a table, but this invention does not specifically limit this method.

[0126] S32, calculate the product of the basic risk value and the collaborative contribution value, and multiply the product by the cluster size correction coefficient to obtain the collaborative risk index.

[0127] Specifically, the final formula for calculating the collaborative risk index is as follows: .

[0128] Example 1: Calculation of Mite Allergen Clusters

[0129] The cluster of mite allergens obtained from the clustering includes four allergens: house dust mite, flour dust mite, tropical clawless mite, and rough-legged flour mite. The risk scores for each allergen are as follows: House dust mite (Strongly positive), dust mites (Positive), Tropical Clawless Mite (Positive), coarse-legged flour mite (Positive). The associations between allergen pairs within this cluster are all homologous, with association weights of 0.9 (house dust mite and house dust mite), 0.6 (house dust mite and tropical clawless mite), and 0.5 (house dust mite and rough-legged flour mite), etc. To simplify the calculation, this example uses an average weight of 0.8 for the example calculation.

[0130] Calculate the base risk value: .

[0131] Calculate the collaborative contribution value: The total contribution within the cluster is... There are 10 allergen pairs, and the synergistic contribution of each allergen pair is... Taking house dust mites and flour mites as examples: House dust mites and tropical clawless mites: House dust mites and rough-legged flour mites: Dust mites and tropical clawless mites: Dust mites and rough-legged flour mites: Tropical clawless mites and rough-legged flour mites: Summing yields .

[0132] Calculate the uncorrected synergistic risk index: .

[0133] Introducing a cluster size correction factor: number of allergens within the cluster , .

[0134] Final Collaborative Risk Index: .

[0135] Example 2: Calculation of cereal allergen clusters

[0136] The clustering of cereal allergens yielded three allergens: corn, wheat, and rice. The risk scores for each allergen are as follows: Corn (Weak positive), wheat (Positive), Rice (Positive). The association between allergen pairs within this cluster is homologous. The association weight between corn and wheat is 0.5, the association weight between wheat and rice is 0.5, and the association weight between corn and rice is 0.5 (because corn and rice are not closely related).

[0137] Calculate the base risk value: .

[0138] Calculate the collaborative contribution value: Corn-Wheat: Corn-Rice: Wheat-Rice: Summing yields .

[0139] Calculate the uncorrected synergistic risk index: .

[0140] Introducing a cluster size correction factor: number of allergens within the cluster , .

[0141] Final Collaborative Risk Index: .

[0142] After obtaining the synergistic risk index of each allergen cluster, it is compared with a preset risk threshold to determine the synergistic exposure risk. The risk threshold can be determined based on statistical analysis of large clinical sample data, for example, by performing back-substitution analysis on the test data of known clinically diagnosed allergy cases and selecting the index value corresponding to the balance point of sensitivity and specificity as the threshold. In this embodiment, the risk threshold can be set to 10. Then: the synergistic risk index of the mite allergen cluster is 32.4 > 10, indicating a synergistic exposure risk; the synergistic risk index of the cereal allergen cluster is 5.77 < 10, indicating no synergistic exposure risk.

[0143] S4. Generate and output a comprehensive report based on the co-exposure risk, including a list of allergen test results and cluster risk warning information for allergen clusters with co-exposure risk.

[0144] In this embodiment, traditional test reports only list the allergen name, test result, and risk level item by item. When multiple weak positive or positive signals are related, clinicians find it difficult to intuitively judge the risk of co-exposure from the table. This embodiment, while retaining the original report format, adds cluster risk warning information, making the risk of co-exposure explicit.

[0145] Generate a section listing the allergen test results item by item. This section should maintain a table format consistent with existing test reports (e.g., ...). Figure 1As shown, this section facilitates clinicians' review of individual allergen testing results. For example, it includes allergen classifications (such as mites, hair follicles, fungi, insects, plant pollen, grains, vegetables, fruits, nuts, dairy products, meat, and others), allergen names (such as house dust mites, flour mites, tropical clawless mites, and rough-legged flour mites), test results (negative, weakly positive, positive, and strongly positive), and allergy risk levels (presented graphically). The report generated in this embodiment retains all of the above information, ensuring compatibility with existing test report formats and enabling clinicians to quickly locate the test results for individual allergens.

[0146] For allergen clusters with potential synergistic exposure risks, additional cluster risk alerts are generated. Based on the comparison results from step S3, alerts are only generated for allergen clusters with a synergistic risk index greater than or equal to the risk threshold; allergen clusters with a synergistic risk index below the threshold are not alerted in the report to avoid information redundancy. The rules for generating cluster risk alerts are as follows:

[0147] Cluster risk warning information should include the following elements:

[0148] Allergen cluster names: named according to the category of allergens within the cluster, such as "dust mite allergen cluster", "cereal allergen cluster", "pollen allergen cluster", etc.

[0149] Cluster Allergen List: Lists all allergens included in the cluster and their test results, facilitating verification by clinicians;

[0150] Collaborative Risk Index Value: Presented in numerical form after calculation For clinical reference;

[0151] Risk level assessment: Based on the ratio of the collaborative risk index to the risk threshold, risks are categorized into "low risk," "medium risk," and "high risk." For example, a risk level is set... Low risk Medium risk High risk;

[0152] Clinical management recommendations: Based on the type and risk level of the allergen cluster, targeted clinical recommendations are provided. The recommendations can be generated based on a pre-defined rule base.

[0153] Taking the mite allergen cluster calculated in step S3 as an example, this cluster includes house dust mite (strongly positive), dry dust mite (positive), tropical clawless mite (positive), and rough-legged white dust mite (positive), with a synergistic risk index of 32.4 and a risk threshold set at 10. This cluster is classified as high-risk. The generated cluster risk warning information for this cluster can be described as follows:

[0154] [Cluster Risk Warning] Mite Allergen Clusters

[0155] The test revealed multiple positive results for the following mite allergens: house dust mite (strongly positive), dry dust mite (positive), tropical clawless mite (positive), and rough-legged flour mite (positive). The co-risk index of this allergen cluster was 32.4 (risk threshold 10), indicating a high-risk level.

[0156] Clinical recommendations: Co-sensitization to multiple mite allergens indicates that the pet is highly sensitive to mites. Co-exposure may trigger or aggravate symptoms of allergic skin diseases (itching, rash, hair loss), allergic rhinitis, bronchitis, etc. The following measures are recommended: (1) Strengthen indoor environmental control, use a mite remover to clean beds, sofas and carpets regularly, and keep indoor humidity below 50%; (2) Replace with mite-proof bedding and avoid using upholstered furniture; (3) Wash pet bedding regularly, using high temperature (above 60℃) washing; (4) If symptoms persist, specific immunotherapy (desensitization therapy) may be considered.

[0157] If multiple weak positive signals overlap in an allergen cluster, such as weak positive results for chicken, beef, and corn, a single weak positive signal may be easily overlooked. However, the three may be related (e.g., chicken and beef are co-exposed to other meats; corn, while not directly related to chicken or beef, is connected through other pathways if they form the same cluster). Assume the cluster's co-risk index is calculated to be 10.4 according to step S3, and the risk threshold is set to 10. This cluster falls under the medium-risk category. The generated cluster risk warning information for this cluster can be described as follows:

[0158] [Cluster Risk Warning] Food Allergen Cluster (Meat and Grains)

[0159] The following food allergens were found to be weakly positive or positive in multiple tests: chicken (weakly positive), beef (weakly positive), and corn (weakly positive). Although the sensitization level of each individual allergen is not high, the synergistic risk index of this allergen cluster is 10.4 (risk threshold 10), which is at a medium risk level. Synergistic exposure may trigger clinical symptoms.

[0160] Clinical recommendations: It is recommended to check whether the pet's daily diet contains the above-mentioned ingredients. Try switching to a single protein source (such as rabbit meat, venison, or other unconventional proteins) or a hypoallergenic prescription diet for 2-4 weeks to conduct a dietary elimination trial and observe whether the skin itching and gastrointestinal symptoms (vomiting, diarrhea) improve. If the symptoms are significantly relieved, the co-exposure to the above-mentioned food allergens can be confirmed as the pathogenic factor.

[0161] The comprehensive report can be generated as a structured text output, or as a visual report in PDF or HTML format, with no specific limitation. In the visual report, cluster risk warnings can be placed after the table of individual test results or presented as a separate section, using prominent colors (e.g., red for high risk, orange for medium risk) to draw the attention of clinicians. For allergen clusters without co-exposure risk, no additional warnings are generated in the report to maintain its simplicity.

[0162] Reference Figure 3 As shown in the illustration, this application also provides an automatic generation device 200 for pet allergen analysis reports, the device comprising:

[0163] The knowledge graph construction module 10 is used to construct an allergen association knowledge graph that stores the association relationships between different allergens and their corresponding association weights; wherein, the association relationships include homology relationships, cross-reaction relationships and co-exposure relationships;

[0164] The data acquisition and clustering module 20 is used to acquire pet in vitro allergen detection data, including allergen names and their risk levels, quantify the risk levels into risk scores, treat all detected allergens as nodes, construct a correlation subgraph based on the correlation edges between nodes with a correlation weight greater than a threshold, and cluster the correlation subgraph to obtain one or more allergen clusters.

[0165] The collaborative risk calculation module 30 is used to calculate the sum of the risk scores of each allergen in each allergen cluster as the basic risk value, calculate the sum of the products of the association weight between each pair of allergens in the cluster and the smaller of the two risk scores as the collaborative contribution value, and use the product of the basic risk value and the collaborative contribution value as the collaborative risk index. The collaborative risk index is compared with the risk threshold to obtain the collaborative exposure risk.

[0166] The report generation module 40 is used to generate and output a comprehensive report based on the co-exposure risk, including a list of allergen detection results and cluster risk warning information generated for allergen clusters with co-exposure risk.

[0167] In some embodiments, refer to Figure 4 As shown, the knowledge graph construction module 10 includes:

[0168] Subset construction unit 11 is used to construct homology subsets, cross-reactivity subsets, and co-exposure subsets based on allergen taxonomy information, immunological cross-reactivity literature, and environmental co-exposure statistics.

[0169] The association weights in the homology subset are set based on the phylogenetic distance of the taxonomic units to which the allergens belong; the association weights in the cross-reactivity subset are set based on antigen epitope similarity or cross-reactivity intensity reported in the literature; and the association weights in the co-exposure subset are set based on environmental co-occurrence frequency or epidemiological co-positive probability.

[0170] The graph merging unit 12 is used to merge the above three subsets to form the allergen association knowledge graph and store it as an undirected weighted graph structure. In the graph, nodes correspond to allergens, edges correspond to association relationships, and edge weights are the corresponding association weights.

[0171] In some embodiments, refer to Figure 5 As shown, the data acquisition and clustering module 20 includes:

[0172] The associated edge extraction unit 21 is used to extract the associated edge between any two nodes from the allergen association knowledge graph, taking all detected allergens as nodes. When the weight of the associated edge is greater than or equal to the associated weight threshold, the associated edge is included in the associated edge set.

[0173] Clustering unit 22 is used to construct an associated subgraph based on the nodes and the set of associated edges, and to cluster the associated subgraph using a connected component algorithm, so that all allergens belonging to the same connected component are grouped into one allergen cluster, thus obtaining one or more allergen clusters.

[0174] In some embodiments, continue to refer to Figure 5 As shown, the clustering unit 22 includes:

[0175] The threshold setting subunit 221 is used to obtain the association type of each associated edge in the associated subgraph and set the corresponding edge retention threshold according to the association type. The edge retention threshold corresponding to the homogeneous relationship is the lowest, and the edge retention threshold corresponding to the cross-reaction relationship is the highest.

[0176] The pruning subunit 222 is used to traverse each associated edge in the associated subgraph. When the associated weight of the associated edge is greater than or equal to the edge retention threshold corresponding to the associated type, the associated edge is retained; otherwise, the associated edge is removed from the associated subgraph to obtain the pruned subgraph.

[0177] The connected component clustering subunit 223 is used to cluster the pruned subgraph using the connected component algorithm, grouping all allergens belonging to the same connected component into one allergen cluster.

[0178] In some embodiments, refer to Figure 6 As shown, the collaborative risk calculation module 30 includes:

[0179] The size correction unit 31 is used to obtain the number of allergens in the allergen cluster and determine the cluster size correction coefficient based on the number of allergens and the corresponding negative correlation.

[0180] The index calculation unit 32 is used to calculate the product of the basic risk value and the collaborative contribution value, and multiply the product by the cluster size correction coefficient to obtain the collaborative risk index.

[0181] This application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the computer program, when executed by the processor, implements the method as described in any of the preceding claims.

[0182] The computer device can be an electronic device with data processing capabilities, such as a desktop computer, server, laptop computer, tablet computer, embedded device, or dedicated testing equipment. The computer device includes at least one processor, memory, communication interface, and bus system. The processor, memory, and communication interface are interconnected via the bus system to complete communication between them. The processor executes the computer program stored in the memory to implement the steps described in the above method embodiments.

[0183] The memory may include high-speed random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device, flash memory, or other solid-state storage device. The memory is used to store an allergen association knowledge graph, allergen detection data, a risk score mapping table, association weight thresholds, risk thresholds, and temporary data required by the processor to execute programs. In some embodiments, the memory also stores a preset cluster risk warning information template library, used to generate corresponding clinical recommendation text based on the type and synergistic risk level of the allergen cluster.

[0184] The processor may be a central processing unit (CPU), a graphics processing unit (GPU), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices.

[0185] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for automatically generating a pet allergen analysis report, characterized in that, Includes the following steps: S1, construct an allergen association knowledge graph that stores the association relationships between different allergens and their corresponding association weights; wherein, the association relationships include homology relationships, cross-reaction relationships, and co-exposure relationships; S2, acquire pet in vitro allergen detection data, including allergen names and their risk levels, quantify the risk levels into risk scores, treat all detected allergens as nodes, construct a correlation subgraph based on the correlation edges between nodes with a correlation weight greater than a threshold, and cluster the correlation subgraph to obtain one or more allergen clusters. S3. For each allergen cluster, calculate the sum of the risk scores of each allergen within the cluster as the base risk value, calculate the sum of the products of the association weight between each pair of allergens within the cluster and the smaller of the two risk scores as the synergistic contribution value, and use the product of the base risk value and the synergistic contribution value as the synergistic risk index. Compare the synergistic risk index with the risk threshold to obtain the synergistic exposure risk. S4. Generate and output a comprehensive report based on the co-exposure risk, including a list of allergen test results and cluster risk warning information for allergen clusters with co-exposure risk.

2. The method for automatically generating a pet allergen analysis report according to claim 1, characterized in that, Construct an allergen association knowledge graph that stores the association relationships and corresponding association weights between different allergens, including: S11. Based on allergen taxonomic information, immunological cross-reactivity literature, and environmental co-exposure statistics, homology subsets, cross-reactivity subsets, and co-exposure subsets are constructed respectively. The association weights in the homology subset are set based on the phylogenetic distance of the taxonomic units to which the allergens belong; the association weights in the cross-reactivity subset are set based on antigen epitope similarity or cross-reactivity intensity reported in the literature; and the association weights in the co-exposure subset are set based on environmental co-occurrence frequency or epidemiological co-positive probability. S12, combine the above three subsets to form the allergen association knowledge graph, and store it as an undirected weighted graph structure. In the graph, nodes correspond to allergens, edges correspond to association relationships, and edge weights are the corresponding association weights.

3. The method for automatically generating a pet allergen analysis report according to claim 1, characterized in that, All detected allergens are treated as nodes. A subgraph of association is constructed based on the association edges between nodes whose association weights are greater than a threshold. The subgraph of association is then clustered to obtain one or more allergen clusters, including: S21, take all detected allergens as nodes, extract the association edge between any two nodes from the allergen association knowledge graph, and when the weight of the association edge is greater than or equal to the association weight threshold, include the association edge in the association edge set. S22, construct an associated subgraph based on the node and the associated edge set, and use a connected component algorithm to cluster the associated subgraph, grouping all allergens belonging to the same connected component into one allergen cluster, to obtain the one or more allergen clusters.

4. The method for automatically generating a pet allergen analysis report according to claim 3, characterized in that, The connected component algorithm is used to cluster the associated subgraph, grouping all allergens belonging to the same connected component into one allergen cluster, resulting in one or more allergen clusters, including: S221, obtain the association type of each associated edge in the associated subgraph, and set the corresponding edge retention threshold according to the association type, wherein the edge retention threshold corresponding to the homo-origin relationship is the lowest, and the edge retention threshold corresponding to the cross-reaction relationship is the highest. S222, traverse each associated edge in the associated subgraph. When the associated weight of the associated edge is greater than or equal to the edge retention threshold corresponding to the associated type, retain the associated edge; otherwise, remove the associated edge from the associated subgraph to obtain the pruned subgraph. S223, The pruned subgraph is clustered using the connected component algorithm, and all allergens belonging to the same connected component are grouped into one allergen cluster.

5. The method for automatically generating a pet allergen analysis report according to claim 4, characterized in that, The product of the basic risk value and the collaborative contribution value is used as the collaborative risk index, which includes: S31, obtain the number of allergens in the allergen cluster, and determine the cluster size correction coefficient based on the number of allergens and the corresponding negative correlation. S32, calculate the product of the basic risk value and the collaborative contribution value, and multiply the product by the cluster size correction coefficient to obtain the collaborative risk index.

6. An automatic generation device for pet allergen analysis reports, characterized in that, The device includes: The knowledge graph construction module is used to construct an allergen association knowledge graph that stores the association relationships between different allergens and their corresponding association weights; wherein, the association relationships include homology relationships, cross-reaction relationships, and co-exposure relationships; The data acquisition and clustering module is used to acquire pet in vitro allergen detection data, including allergen names and their risk levels, quantify the risk levels into risk scores, treat all detected allergens as nodes, construct a correlation subgraph based on the correlation edges between nodes with a correlation weight greater than a threshold, and cluster the correlation subgraph to obtain one or more allergen clusters. The collaborative risk calculation module is used to calculate the sum of the risk scores of each allergen within each allergen cluster as the base risk value, calculate the sum of the products of the association weight between each pair of allergens within the cluster and the smaller of the two risk scores as the collaborative contribution value, and use the product of the base risk value and the collaborative contribution value as the collaborative risk index. The collaborative risk index is compared with the risk threshold to obtain the collaborative exposure risk. The report generation module is used to generate and output a comprehensive report based on the co-exposure risks, including a list of allergen detection results and cluster risk warning information for allergen clusters with co-exposure risks.

7. The automatic generation device for pet allergen analysis reports according to claim 6, characterized in that, The knowledge graph construction module includes: The subset construction unit is used to construct homology subsets, cross-reactivity subsets, and co-exposure subsets based on allergen taxonomic information, immunological cross-reactivity literature, and environmental co-exposure statistics. The association weights in the homology subset are set based on the phylogenetic distance of the taxonomic units to which the allergens belong; the association weights in the cross-reactivity subset are set based on antigen epitope similarity or cross-reactivity intensity reported in the literature; and the association weights in the co-exposure subset are set based on environmental co-occurrence frequency or epidemiological co-positive probability. The graph merging unit is used to merge the above three subsets to form the allergen association knowledge graph and store it as an undirected weighted graph structure. In the graph, nodes correspond to allergens, edges correspond to association relationships, and edge weights are the corresponding association weights.

8. The automatic generation device for pet allergen analysis reports according to claim 6, characterized in that, The data acquisition and clustering module includes: The associated edge extraction unit is used to extract the associated edge between any two nodes from the allergen association knowledge graph, taking all detected allergens as nodes. When the weight of the associated edge is greater than or equal to the associated weight threshold, the associated edge is included in the associated edge set. The clustering unit is used to construct an associated subgraph based on the nodes and the set of associated edges, and to cluster the associated subgraph using a connected component algorithm, so that all allergens belonging to the same connected component are grouped into one allergen cluster, thus obtaining one or more allergen clusters.

9. The automatic generation device for pet allergen analysis reports according to claim 8, characterized in that, The clustering unit includes: The threshold setting subunit is used to obtain the association type of each associated edge in the associated subgraph and set the corresponding edge retention threshold according to the association type. The edge retention threshold corresponding to the homology relationship is the lowest, and the edge retention threshold corresponding to the cross-reaction relationship is the highest. The pruning sub-unit is used to traverse each associated edge in the associated subgraph. When the associated weight of the associated edge is greater than or equal to the edge retention threshold corresponding to the associated type, the associated edge is retained; otherwise, the associated edge is removed from the associated subgraph to obtain the pruned subgraph. The connected component clustering subunit is used to cluster the pruned subgraph using the connected component algorithm, grouping all allergens belonging to the same connected component into one allergen cluster.

10. A computer device, characterized in that, The method includes a memory and a processor, wherein the memory stores a computer program, and the computer program, when executed by the processor, implements the method as described in any one of claims 1-5.

Images