Method for constructing a chain of transmission of an infectious disease based on coupling of mechanism processes

Abstract

This invention relates to the field of data processing technology and discloses a method for constructing infectious disease transmission chains based on mechanistic process coupling. The method includes: representing the process with case data; calculating the initial transmission probability between pairs of cases based on the degree of overlap of their spatiotemporal trajectories; constructing an infectious disease knowledge graph to represent the infectious disease mechanism, and fusing case data into a Bayesian network model to perform mechanism-guided probability correction on the initial transmission probability to obtain the posterior transmission probability between pairs of cases; constructing a weighted directed propagation graph based on the posterior transmission probability, and reconstructing the infectious disease transmission chain through graph structure optimization. This method introduces infectious disease mechanisms into the transmission probability calculation process to constrain the determination of transmission relationships between cases, avoiding false transmission edges generated solely based on spatiotemporal trajectory overlap, and improving the reliability and interpretability of transmission chain inference.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method for constructing infectious disease transmission chains based on mechanistic process coupling. Background Technology

[0002] Determining the transmission chain of infectious diseases is a crucial step in infectious disease prevention and control decision-making. Existing methods primarily rely on the overlap of spatiotemporal trajectories between patients to determine transmission relationships. However, these methods lack sufficient integration with the mechanisms of infectious diseases, and the inference process is susceptible to interference from accidental spatiotemporal overlaps, leading to false transmission chains. When epidemiological investigation information is incomplete or transmission links are missing, traditional data-driven methods struggle to effectively utilize prior knowledge of infectious diseases to infer and complete transmission relationships, resulting in insufficient reliability and scientific rigor in the constructed transmission chains. Fundamentally, existing methods often separate data analysis from the mechanisms of infectious diseases, failing to achieve transmission relationship inference under the constraints of these mechanisms, thus limiting the accuracy and scientific validity of transmission chain reconstruction. Summary of the Invention

[0003] The purpose of this invention is to solve the above problems and provide a method for constructing infectious disease transmission chains based on the coupling of mechanisms and processes. This method uses a probabilistic graphical model as the fusion carrier of mechanisms and processes and introduces an epidemiological knowledge graph as prior knowledge. This can reduce false transmission relationships caused by relying solely on the overlap of spatiotemporal trajectories and improve the reliability and interpretability of transmission chain inference.

[0004] The technical solution adopted by this invention to solve its technical problem is: a method for constructing an infectious disease transmission chain based on mechanism-process coupling, the method comprising the following steps:

[0005] S1 acquires case data;

[0006] S2 uses case data as information for observing the transmission process and calculates the initial transmission probability between each pair of cases based on the degree of overlap in their spatiotemporal trajectories.

[0007] S3 constructs an infectious disease knowledge graph to characterize the mechanisms of infectious diseases;

[0008] S4 integrates case data and epidemiological knowledge graphs in a Bayesian network model to perform mechanism-guided probability correction on the initial transmission probability, thereby obtaining the posterior transmission probability between each pair of cases.

[0009] S5 constructs a weighted directed propagation graph based on posterior propagation probability and reconstructs the infectious disease transmission chain through graph structure optimization.

[0010] The case data in step S1 includes individual travel information, inter-individual relationship data, comprehensive individual information, and nucleic acid test information. The individual travel information includes trip ID, trajectory date, trip start and end time, POIs and their latitude and longitude coordinates, main mode of transportation used, and trip duration. The inter-individual relationship data includes relationship tags between two patients, which include at least: same building, colleagues, cohabitation, chatting together, traveling together, and simultaneous location. The comprehensive individual information includes POIs and latitude and longitude coordinates of residence and workplace, as well as trajectory event descriptions containing activity types. The activity types include dining together, living in the same room, and simultaneous location. The test information includes the date of diagnosis.

[0011] Furthermore, step S2 specifically includes:

[0012] S21 extracts individual travel information for each case from the case data;

[0013] S22 for any two cases and traversing all its travel records ,in Case The itinerary collection Case A collection of itineraries;

[0014] S23 Determine the travel distance With itinerary Does spatiotemporal overlap exist?

[0015] S24 If a certain itinerary is... If the spatiotemporal overlap condition is met, it is included in the effective spatiotemporal overlap set, and the original propagation score is calculated.

[0016] S25 normalizes the original transmission scores of all case pairs to obtain the initial transmission probability.

[0017] The individual travel information in step S21 includes multiple trip records, each trip record including trip start time, trip end time, trip origin geographical coordinates and trip destination geographical coordinates;

[0018] In step S23, the travel is determined. With itinerary Does spatiotemporal overlap exist? Specifically, it includes:

[0019] Time overlap determination: If the start time of the later segment of the two trips is earlier than the end time of the earlier segment of the two trips, then the time overlap duration is calculated; the time overlap duration is the difference between the end time of the earlier segment of the two trips and the start time of the later segment of the two trips.

[0020] Among them, the itinerary Including cases The Start time of the journey and the case The End time of the trip, itinerary Including cases The Start time of the journey and the case The End time of the trip;

[0021] The later start time of the two trips mentioned above refers to the later start time of each person's trip; the earlier end time of the two trips mentioned above refers to the earlier end time of each person's trip.

[0022] Spatial nearest neighbor determination: Treat each segment of the journey as a line segment from the starting point to the ending point, and calculate the journey distance. and itinerary Shortest Euclidean distance ;like If they are spatially adjacent, then they are determined to be spatially adjacent. To set a spatial threshold, the value range is 50 meters to 500 meters;

[0023] Step S24 specifically includes treating each travel pair that simultaneously satisfies the conditions of temporal overlap and spatial proximity as a potential transmission contact event, and calculating its transmission contribution score based on the overlap duration and spatial distance: the longer the overlap time and the closer the distance, the higher the score. Finally, the scores of all valid travel pairs are summed to obtain the original transmission score of the case pair.

[0024] ,

[0025] in, For cases and cases The original propagation score, For the purpose of recording the itinerary, The set of all travel pairs that satisfy the spatiotemporal overlap condition. For the itinerary and itinerary Duration of time overlap It is the space decay factor. Itinerary and itinerary The shortest Euclidean distance, It is the spatial attenuation scale parameter, when When the spatial attenuation factor is 1, At that time, the spatial attenuation factor is 0.37. At that time, the spatial decay factor is 0.05;

[0026] Step S25 involves globally normalizing the original propagation score to obtain an initial propagation probability between 0 and 1:

[0027] ,

[0028] in, For cases and cases The original propagation score, For normalized cases and cases The initial propagation probability, It is the highest original transmission score among all case pairs. This is to prevent division by zero of extremely small positive numbers.

[0029] Compared to existing technologies, this invention does not rely solely on coarse-grained spatiotemporal co-occurrence judgments or simple intersection calculations based on case trajectories. Instead, it addresses the mechanisms of infectious diseases by introducing temporal overlap determination and spatial nearest neighbor constraints at the individual level. It treats each journey segment as a continuous line segment and calculates the shortest distance, achieving a more refined identification of potential transmission events that better reflects the characteristics of real-world contact-based infectious diseases. Furthermore, this invention overcomes the shortcomings of existing methods in reflecting the persistence of contact and spatial attenuation effects by constructing an effective temporal overlap set and accumulating transmission scores from multiple transmission events, and by employing an "overlap duration-spatial decay" function to characterize the changes in transmission risk over time and distance.

[0030] Furthermore, the infectious disease knowledge graph mentioned in step S3 specifically includes infectious disease epidemiology knowledge, which includes knowledge of the source of infection, transmission process, and susceptible population: the knowledge of the source of infection includes the infectious period, infectiousness time curve, and intergenerational time of infectious diseases; the knowledge of the transmission process includes the risk of place type and the risk of contact method; and the knowledge of the susceptible population includes basic susceptibility.

[0031] Furthermore, step S4 specifically includes:

[0032] S41 constructs infectious disease mechanism constraints based on an infectious disease knowledge graph and maps these infectious disease mechanism constraints to mechanism evidence variables in a probabilistic graphical model;

[0033] S42 targets any case Based on the aforementioned mechanism evidence variables, a mechanism propagation probability is constructed;

[0034] In the Bayesian network model, S43 uses the initial propagation probability obtained in step S2 as the observation likelihood and the mechanism propagation probability as the prior distribution for fusion and updating to obtain the posterior propagation probability between each pair of cases.

[0035] Step S41 specifically includes:

[0036] Subgraphs related to the mechanisms of infectious diseases are extracted from the knowledge graph of infectious diseases. Based on the semantic structure of entities and their relationships in the subgraphs, a probabilistic graphical model structure corresponding to the transmission process of infectious diseases is constructed. The subgraphs are mapped to random variables and their conditional dependencies in the probabilistic graphical model, thereby forming a model structure for characterizing the constraints of the mechanisms of infectious diseases.

[0037] The epidemiological mechanism constraints mentioned above include at least a structured representation of the characteristics of the source of infection, the characteristics of the transmission process, and the characteristics of population susceptibility; wherein the characteristics of the source of infection are jointly determined by the infectious period, the infectiousness time curve, and the intergenerational time distribution;

[0038] The characteristics of the transmission process are jointly determined by the risk of the type of place where the transmission occurs and the risk of the contact method between cases.

[0039] The susceptibility characteristics of the population are determined by baseline susceptibility;

[0040] The aforementioned infectious disease mechanism constraints are discretized into finite levels according to a preset mapping rule and mapped to mechanism evidence variables in a probabilistic graphical model. The value range of the mechanism evidence variables is [0,1] or a preset set of levels.

[0041] Step S42 specifically includes:

[0042] For any case Based on structured information related to the mechanisms of infectious diseases in the knowledge graph of infectious diseases, and constructing the mechanism transmission probability. The mechanism propagation probability satisfies:

[0043] ,

[0044] in, For cases Transmission to cases The propagation event variables; These are variables that provide evidence of the source of infection, determined by the infectious period, the infectious time curve, and the intergenerational time distribution. These are variables used as evidence of the transmission process, determined by the risk of location type and the risk of contact method; These are susceptibility evidence variables, determined by basic susceptibility; the conditional probabilities are predetermined by the mechanistic rules of the knowledge graph in the form of conditional probability tables or conditional probability distributions.

[0045] Step S43 specifically includes:

[0046] In the Bayesian network model, the initial propagation probability obtained in step S25 is used as the observation likelihood term, and the mechanism propagation probability obtained in step S42 is used as the prior distribution for fusion and updating.

[0047] ,

[0048] in, It is the initial propagation probability obtained in step S25. It is the mechanism propagation probability obtained in step S42. To integrate the updated propagation probabilities,

[0049] ,

[0050] in, To integrate the updated propagation probabilities, It is the normalized posterior propagation probability. Cases from the same source Pointing to all potentially susceptible cases The summation of the fusion components, It is a very small positive number to avoid division by zero errors caused by a denominator of 0.

[0051] Compared to existing technologies, this invention does not solely rely on spatiotemporal co-occurrence or similarity indices of case trajectories to determine transmission relationships. Instead, it introduces an epidemiological knowledge graph to represent the constraints of epidemiological mechanisms, structurally mapping the characteristics of the source of infection, the characteristics of the transmission process, and the characteristics of susceptible populations into mechanistic evidence variables in a probabilistic graphical model, and constructing mechanistic transmission probabilities based on this. Within a Bayesian network framework, this invention achieves a fusion update mechanism of "spatiotemporal observation likelihood and mechanistic constraints," obtaining a normalized posterior transmission probability output. This enables the transmission results to possess both data-driven and mechanistic interpretability, overcoming the limitations of existing methods that only provide co-occurrence strength and black-box learning weights, lacking sufficient epidemiological mechanism constraints and probability consistency.

[0052] Furthermore, step S5 specifically includes:

[0053] S51 constructs a propagation probability matrix based on the pairwise posterior propagation probabilities between cases obtained in step S4;

[0054] S52 constructs a weighted directed propagation graph based on the propagation probability matrix, where cases are nodes and edges represent propagation relationships. The weight of each edge is the posterior propagation probability of the corresponding case pair.

[0055] S53 On the weighted directed propagation graph, solve for the propagation chain structure that satisfies the preset structural constraints. The structural constraints include at least: each node corresponds to at most one most likely source node; the propagation chain structure is an acyclic directed structure.

[0056] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the aforementioned method for constructing an infectious disease transmission chain based on mechanistic process coupling.

[0057] A computer-readable storage medium having computer instructions stored thereon, which, when executed by a processor, implement the aforementioned method for constructing an infectious disease transmission chain based on mechanistic process coupling.

[0058] Beneficial effects:

[0059] Compared with existing technologies, this invention introduces the mechanism constraint of infectious disease knowledge graph in the transmission probability inference stage, and integrates case data and infectious disease mechanisms within the framework of a probabilistic graphical model. This overcomes the problems of existing technologies that rely solely on spatiotemporal co-occurrence to determine transmission relationships, are prone to generating false transmission edges, and lack mechanistic explanations. It has the following beneficial effects:

[0060] (1) Overcome the problem of misjudgment of correlation caused by relying solely on spatiotemporal co-occurrence.

[0061] Existing technologies typically infer propagation edges based on the spatiotemporal overlap or co-occurrence intensity of trajectories. However, spatiotemporal co-occurrence only reflects correlation and does not necessarily indicate actual propagation. It is easily affected by factors such as accidental co-occurrence, positioning errors, or missing trajectories, resulting in a large number of false propagation edges. This invention corrects the initial propagation probability results by introducing epidemiological mechanisms of infectious diseases, effectively reducing misjudgments.

[0062] (2) To overcome the problem of insufficient utilization of information on the mechanisms of infectious diseases,

[0063] Existing methods generally lack a structured characterization of the source of infection, the risks of the transmission process, and differences in population susceptibility, resulting in inferences lacking mechanistic support. This invention extracts mechanistic subgraphs from an epidemiological knowledge graph and maps them to mechanistic evidence variables in a probabilistic graphical model, enabling transmission patterns to explicitly participate in the inference process.

[0064] (3) Case data and mechanisms are used in synergistic initiation to improve the reliability of inferences.

[0065] This invention uses the initial propagation probability obtained from trajectory data as the observation likelihood term and the mechanistic propagation probability as the prior distribution for fusion and updating. When case pairs simultaneously possess strong evidence of spatiotemporal contact and meet the conditions of an infectious disease mechanism, their propagation probability increases; conversely, when co-occurrence exists but does not meet the characteristics of the source of infection, the transmission process, or the susceptibility of the population, the propagation probability decreases, thereby achieving probability filtering of spurious associations.

[0066] (4) Ensure that the propagation chain results simultaneously satisfy data consistency and mechanism rationality.

[0067] Through the aforementioned fusion inference mechanism, this invention can theoretically ensure that the transmission chain inference results conform to both the observed facts of case trajectories and the epidemiological mechanisms of infectious diseases, avoiding errors in transmission direction or unreasonable transmission chain structures caused solely by co-occurrence intensity.

[0068] (5) Enhance the scientific rigor, interpretability, and application value of tracing the transmission chain.

[0069] This invention can effectively improve the reliability and stability of propagation chain reconstruction, making the inference results more scientific and interpretable. Attached Figure Description

[0070] Figure 1 This is a flowchart of a method for constructing infectious disease transmission chains based on mechanism-process coupling.

[0071] Figure 2 This is a schematic diagram illustrating the calculation of the initial propagation probability based on the overlap of spatiotemporal trajectories in this embodiment.

[0072] Figure 3 This is a schematic diagram of the structure of the epidemiological knowledge graph in this embodiment.

[0073] Figure 4 This is a schematic diagram of the propagation probability fusion update based on the Bayesian network model in this embodiment.

[0074] Figure 5 This is a flowchart illustrating the generation of the propagation chain structure based on the maximum spanning tree algorithm in this embodiment.

[0075] Figure 6 This is a schematic diagram illustrating the generation of the propagation chain structure based on the maximum spanning tree algorithm in this embodiment.

[0076] Figure 7 This is a schematic diagram of the reconstructed propagation chain result in this embodiment.

[0077] Figure 8 This is a structural diagram of an electronic device in this embodiment. Detailed Implementation

[0078] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments, so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby making a clearer and more explicit definition of the scope of protection of the present invention.

[0079] Identifying the transmission chain of infectious diseases is crucial for understanding the scale and mechanism of transmission, and is a key foundation for infectious disease prevention and control. Existing methods primarily rely on the overlap of spatiotemporal trajectories between cases to determine transmission relationships. However, these methods lack sufficient integration with epidemiological mechanisms, and the inference process is susceptible to interference from accidental spatiotemporal overlaps, leading to false transmission chains. When epidemiological information is incomplete or transmission links are missing, traditional data-driven methods struggle to effectively utilize prior knowledge of epidemiology to infer and complete transmission relationships, resulting in insufficient reliability and scientific rigor in the constructed transmission chains. Fundamentally, existing methods often separate data analysis from epidemiological mechanisms, failing to achieve transmission relationship inference under the constraints of these mechanisms, thus limiting the accuracy and scientific validity of transmission chain reconstruction.

[0080] To address the aforementioned issues, this application proposes a method for constructing infectious disease transmission chains based on mechanism-process coupling. By introducing infectious disease mechanisms into the transmission probability calculation process, this method constrains the determination of transmission relationships between cases, avoiding false transmission edges generated solely based on overlapping spatiotemporal trajectories, and improving the reliability and interpretability of transmission chain inference. For more technical implementation details of this application's embodiments, please refer to the specific embodiments below.

[0081] Example: This invention provides a method for constructing infectious disease transmission chains based on mechanism-process coupling (see...). Figure 1 The method includes the following steps:

[0082] S1 acquires case data.

[0083] First, patient case data is collected through telephone surveys, big data positioning, serological and molecular epidemiology, and other methods. Since this data typically records patient activity trajectories in natural language, it is unstructured data and therefore requires structuring to transform it into computable data. The processing flow includes: extracting key information such as patient identifiers (IDs), activity time, and location using a natural language processing model; then automatically cleaning and formatting the data, converting time data into standardized timestamps, and using geocoding services to convert location data into latitude and longitude coordinates and standardized addresses; finally, all data is indexed by patient ID and organized into a standardized form, thus completing the transformation from unstructured epidemiological investigation text to structured data. In this embodiment, the acquired case data includes:

[0084] Individual travel information: This includes multiple trip records. Each trip record contains the trip start time, end time, origin and destination POIs and their latitude and longitude coordinates, location type, mode of transportation, etc.

[0085] Inter-individual relationship data: This includes relationship tags between cases, which at least include cohabitation, same building, same travel, same time and space, etc.

[0086] Individual comprehensive information: including the case's place of residence, place of work (POI), and description of activity type;

[0087] Testing information: including the time of diagnosis of the case.

[0088] S2 uses case data as information for observing the transmission process and calculates the initial transmission probability between pairs of cases based on the degree of overlap in their spatiotemporal trajectories.

[0089] See Figure 2 In this embodiment, based on the structured data obtained in step S1, the initial transmission probability between each pair of cases is calculated. The specific process is as follows:

[0090] S21 extracts individual travel information for each case from the case data;

[0091] Taking Case 33 and Case 1 as examples, their travel records were extracted from the individual travel information form;

[0092] Case 33's travel itinerary includes:

[0093] journey : 2022-08-28 21:10–22:20, POI=A venue (entertainment venue).

[0094] Case 1's travel itinerary includes:

[0095] journey : 2022-08-28 21:30–21:58, POI=A Venue (Entertainment Venue),

[0096] journey : 2022-08-29 09:10–09:40, POI=B Location (Workplace).

[0097] S22 for any two cases and traversing all its travel records ,in Case The itinerary collection Case A collection of itineraries;

[0098] In this embodiment, the itinerary pair set of Case 33 and Case 1 is as follows: Each trip pair is considered as a potential contact event candidate for subsequent spatiotemporal overlap determination.

[0099] S23 Determine the travel distance With itinerary Does spatiotemporal overlap exist?

[0100] In this embodiment, Time-based judgment was made, among which, the first day of case 33 The first segment of the journey started at 21:10 and ended at 22:20; Case 1's... The first trip starts at 21:30 and ends at 21:58. The second trip starts at 21:30, and the first trip ends at 21:58. Since the second trip starts at 21:30 earlier than the first trip ends at 21:58, the two trips are determined to have overlapping times. The duration of the overlap is calculated as: the first trip ending at 21:58 minus the second trip starting at 21:30, resulting in a difference of 28 minutes.

[0101] Similarly, for the itinerary When we perform a time-based judgment, since the time intervals do not overlap, we can directly exclude them.

[0102] Based on the POI information in the travel itinerary, the itinerary and The starting and ending points are both at location A, with a spatial distance of 0 meters, which is less than the preset spatial threshold. =200 meters, therefore it is judged to be spatially adjacent.

[0103] S24 If a certain itinerary is... If the spatiotemporal overlap condition is met, it is included in the effective spatiotemporal overlap set, and the original propagation score is calculated.

[0104] In this embodiment, the original transmission scores for Case 33 and Case 1 are calculated:

[0105] because ,Pick =50 meters, then:

[0106] ,

[0107] Similarly, taking case 21 as another candidate source of infection, the effective travel history between case 21 and case 1 is as follows:

[0108] With an overlap time of 30 minutes and a spatial distance of 15 meters, the original propagation score is:

[0109] ,

[0110] S25 normalizes the original transmission scores of all case pairs to obtain the initial transmission probability;

[0111] In this embodiment, the maximum propagation score is:

[0112] ,

[0113] The initial transmission probability between case 33 and case 1 is:

[0114] ,

[0115] The initial transmission probability between Case 21 and Case 1 is:

[0116] ,

[0117] Based solely on the overlap of spatiotemporal trajectories to determine the transmission relationship, both Case 33 and Case 21 were potential sources of infection for Case 1, and their transmission probabilities were similar, indicating an uncertainty regarding their transmission relationship.

[0118] S3 constructs an infectious disease knowledge graph to characterize the mechanisms of infectious diseases.

[0119] In this embodiment, knowledge of infectious disease epidemiology is obtained from infectious disease prevention and control guidelines, epidemiological literature, professional books, etc., including at least the following three categories:

[0120] Knowledge of the source of infection: infectious period, infectious time curve, intergenerational transmission time.

[0121] Knowledge of each stage of the transmission process: risks associated with different types of venues and risks associated with different modes of contact.

[0122] Knowledge about susceptible populations: basic susceptibility.

[0123] The infectivity-time curve is used to describe the functional relationship between the infectivity of a case and time. In this embodiment, a normalized function is used to represent the infectivity intensity. The baseline susceptibility is estimated from a serological survey, and the expression is: Baseline susceptibility = 1 − seropositivity rate.

[0124] The aforementioned mechanistic knowledge is stored in the form of rules or parameters, for example: entertainment venues → high risk (0.8), cohabitation → high risk (0.9), sharing a room → medium risk (0.6). See the epidemiological knowledge graph for details. Figure 3 .

[0125] S4 integrates case data and an epidemiological knowledge graph in a Bayesian network model to perform mechanism-guided probability correction on the initial transmission probability, thereby obtaining the posterior transmission probability between each pair of cases.

[0126] See Figure 4 In this embodiment, after calculating the initial transmission probability in step S2, mechanistic knowledge from the epidemiological knowledge graph is introduced to correct the transmission probability by means of mechanistic constraints, so as to eliminate the ambiguity of transmission based solely on spatiotemporal trajectories and obtain the posterior transmission probability between each pair of cases. The specific process is as follows:

[0127] S41 Extracts mechanism subgraphs related to transmission judgment from the epidemiological knowledge graph in step S3. The subgraphs include at least: source of infection mechanism subgraph (infectious period, infectiousness time curve, intergenerational transmission time), transmission process mechanism subgraph (location type risk, contact mode risk), and susceptible population mechanism subgraph (basic susceptibility).

[0128] Based on the aforementioned mechanism subgraph, a model structure is constructed to characterize the constraints of the epidemiological mechanism of infectious diseases. The model structure includes at least a structured representation of the characteristics of the source of infection, the characteristics of the transmission process, and the characteristics of population susceptibility.

[0129] The epidemiological mechanism constraints are discretized or normalized according to a preset mapping rule and mapped to mechanistic evidence variables in a probabilistic graphical model. These mechanistic evidence variables characterize the prior evidence of the epidemiological mechanism in the probabilistic inference process, and their values ​​range from [0,1]. In this embodiment, the following mapping rule is used:

[0130] If the contact occurs at the peak of the infectious period, then F1 = 1.0;

[0131] If it occurs on the edge of the infectious period, then F1 = 0.5;

[0132] If it is not during the infectious period, then F1=0;

[0133] For high-risk locations and high-risk exposure methods, F2 = 0.8;

[0134] For medium-risk locations or general contact methods, F2 = 0.5;

[0135] The susceptibility value for the general population is F3=0.7.

[0136] S42 targets any case Based on the aforementioned mechanism evidence variables, a mechanism propagation probability is constructed;

[0137] In this embodiment, taking Case 1 as an example, based on the mechanistic rules of case data and knowledge graph, Table 1 is obtained:

[0138] Table 1 Examples of transmission probabilities by mechanism among different cases

[0139]

[0140] Constructing mechanism propagation probabilities based on mechanism variables:

[0141] In this embodiment, the conditional probability is pre-defined in the form of a conditional probability table (CPT) based on the mechanistic rules in the knowledge graph, and is implemented using a weighted calculation method. The weights are determined by epidemiological priors and can be adjusted according to the disease.

[0142] ,

[0143] Substitute into the calculation:

[0144] ,

[0145] This yields the probability of mechanistic transmission between any two cases.

[0146] In the Bayesian network model, S43 uses the initial propagation probability obtained in step S25 as the observation likelihood and the mechanism propagation probability as the prior distribution for fusion and updating.

[0147] In this embodiment, Case 1 is taken as an example:

[0148] ,

[0149] Normalization yields:

[0150] ,

[0151] ,

[0152] Thus, the posterior propagation probability after mechanistic constraint correction is obtained.

[0153] S5 constructs a weighted directed propagation graph based on the posterior propagation probability, and reconstructs the infectious disease transmission chain through graph structure optimization.

[0154] In this embodiment, a weighted directed propagation graph is constructed based on the posterior propagation probabilities between each pair of cases obtained in step S4, and the propagation chain structure that satisfies the structural constraints is solved by the maximum spanning tree algorithm.

[0155] S51 constructs a propagation probability matrix based on the pairwise posterior propagation probabilities between cases obtained in step S4, where the matrix elements represent cases. To the case The transmission probability, where the diagonal elements are set to 0, indicating that the case does not transmit to itself;

[0156] S52 constructs a weighted directed propagation graph based on the propagation probability matrix, where cases are nodes and edges represent propagation relationships. The weight of each edge is the posterior propagation probability of the corresponding case pair.

[0157] S53 On the weighted directed propagation graph, solve for the propagation chain structure that satisfies the preset structural constraints. The structural constraints include at least: each node corresponds to at most one most likely source node; the propagation chain structure is an acyclic directed structure.

[0158] In this embodiment, the maximum spanning tree algorithm is used to generate the propagation chain. The structure of the propagation chain must satisfy the following constraints: each case node corresponds to at most one most likely source of infection node; the propagation structure is an acyclic directed structure; and the propagation chain as a whole has the maximum total propagation probability.

[0159] The problem of solving the above propagation chain structure can be approximated by solving a maximum spanning tree structure in a directed weighted graph. First, for each non-root node in the graph, select the edge with the largest weight. Then, perform cycle detection to check whether the selected set of incoming edges forms a directed cycle. If no cycle is formed, the current edge set is the maximum spanning tree structure we are looking for. If one or more directed cycles exist, we enter the cycle contraction stage, shrinking each cycle into a "pseudo-node" and adjusting the weights of the external edges connected to the cycle. The specific operation of weight adjustment is as follows: the new weight of the edge pointing from the external node to the node inside the cycle is set to the original weight minus the weight of the largest incoming edge entering the node inside the cycle. Then, the above process is recursively applied to the new graph after contraction until the selected edge set has no cycles. Finally, during the backtracking process, each pseudo-node is gradually expanded, and the optimal edge connection in the original graph is restored based on the weight adjustment information, thereby reconstructing the propagation chain.

[0160] See Figure 5 and Figure 6 In this embodiment, a set of weighted directed edges and their weights are given: (33→1: 0.49), (33→2: 0.45), (33→3: 0.42), (33→4: 0.5), (1→2: 0.2), (4→3: 0.7), (21→1: 0.18), (47→1: 0.19). Taking node 33 as the root node, for all non-root nodes (1, 2, 3, 4) in the graph, the incoming edge with the largest current weight is selected: node 1 selects 33→1 (0.49), node 2 selects 1→2 (0.2), node 3 selects 4→3 (0.7), and node 4 selects 33→4 (0.5), resulting in the initial set of incoming edges {33→1, 1→2, 4→3, 33→4}. A directed cycle detection was performed on the set, identifying a directed cycle 33→4→3, which does not meet the acyclic requirement of a tree graph. Therefore, cycle processing was performed based on the cycle contraction mechanism: nodes within the cycle were merged into super nodes, the weights of incoming edges outside the cycle were recalculated, the super node was selected with the largest incoming edge, and finally the cycle structure was expanded to eliminate the cycle. After completing the cycle processing, an acyclic, maximum spanning tree graph starting from the root node 33 and covering all target nodes was obtained. The final propagation chain consists of edges 33→1 (0.49), 33→3 (0.42), 33→4 (0.5), and 1→2 (0.2). The incoming edges 21→1 (0.18) and 47→1 (0.19) corresponding to nodes 21 and 47 were not included in the final propagation chain because the target node 1 had already been assigned the higher-weighted incoming edge 33→1 (0.49).

[0161] In other implementations, minimum cut, maximum matching, or integer programming methods can also be used to solve the propagation chain structure. Figure 7 A schematic diagram illustrating the results of reconstructing the propagation chain.

[0162] In cases where some case information is missing or uncertain, this embodiment uses the mechanistic constraints provided by the epidemiological knowledge graph to complete and correct the transmission relationship, thereby ensuring the stability of the transmission chain construction process.

[0163] This embodiment provides a schematic diagram of an electronic device structure, for reference... Figure 8 The electronic device 8 in this embodiment includes a processor 80, a memory 81, and a computer program 82 stored in the memory and operable on the processor. The computer program is capable of implementing all the steps of the above-described method for constructing an infectious disease transmission chain based on mechanism-process coupling.

[0164] The processor 80 can be a central processing unit, other general-purpose processors, digital signal processors, application-specific integrated circuits (ASICs), off-the-shelf programmable gate arrays (OPGs), discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0165] The memory 81 can be an internal storage unit of an electronic device, such as a read-only memory (ROM), random access memory (RAM), static storage device, dynamic storage device, etc. The memory can be used to store the operating system and other applications, and can also temporarily store data that has been output or will be output.

[0166] The computer program 82 can implement the steps of the method for constructing an infectious disease transmission chain based on mechanism-process coupling, and the processor 80 executes the computer program 82.

[0167] In this embodiment, the computer-readable storage medium may include, but is not limited to: read-only memory (ROM), random access memory (RAM), disk storage medium, optical disk storage medium, flash memory, or other non-volatile storage medium capable of storing program instructions.

[0168] The above-mentioned storage media types are all known to those skilled in the art, and the present invention does not limit the specific form of the computer-readable storage media. In this embodiment, the computer program instructions stored on the computer-readable storage medium are loaded into the processor and configured to execute the infectious disease transmission chain construction process described in this invention. The specific implementation includes: the processor first calls the case data processing module in the storage medium to perform structured parsing of epidemiological investigation text or trajectory data, generating a standardized case dataset containing individual case travel information, contact relationship information, and detection time information; the processor executes the initial transmission probability calculation instruction, traversing the travel records of any pair of cases and calculating the transmission contribution score based on temporal overlap and spatial proximity constraints, thereby obtaining the initial transmission probability between each pair of cases; the processor executes the mechanism constraint fusion instruction, calling the knowledge of the source of infection, transmission process, and susceptibility in the infectious disease knowledge graph, mapping the mechanism to mechanism evidence variables in the probabilistic graphical model, and fusing and updating the initial transmission probability under the Bayesian network framework, outputting the posterior transmission probability; the processor executes the transmission chain structure optimization instruction, constructing a weighted directed transmission graph based on the posterior transmission probability, and solving for the optimal transmission chain structure satisfying acyclic constraints through the maximum spanning tree algorithm or other graph structure optimization methods, thereby reconstructing the transmission path.

[0169] The computer-readable storage medium described in this embodiment can support the programmatic implementation of the method of the present invention in electronic devices, so that the transmission chain inference process not only relies on case trajectory observation information, but also integrates the constraints of epidemiological mechanisms, thereby improving the scientificity, reliability and solvability of the transmission chain construction results.

[0170] The computer-readable storage medium described in this embodiment can support the programmatic implementation of the method of the present invention in electronic devices, so that the transmission chain inference process not only relies on case trajectory observation information, but also integrates the constraints of epidemiological mechanisms, thereby improving the scientificity, reliability and solvability of the transmission chain construction results.

[0171] The preferred embodiments of the present invention have been described in detail above, but the present invention is not limited to the embodiments described. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention, and these equivalent modifications or substitutions are all included within the scope defined by the claims of this application.

Claims

1. A method for constructing infectious disease transmission chains based on mechanistic process coupling, characterized in that, The method includes the following steps: S1 acquires case data; S2 uses case data as a process to calculate the initial transmission probability between pairs of cases based on the degree of overlap in their spatiotemporal trajectories. S3 constructs an infectious disease knowledge graph to characterize the mechanisms of infectious diseases; S4 integrates case data and epidemiological knowledge graphs in a Bayesian network model to perform mechanism-guided probability correction on the initial transmission probability, thereby obtaining the posterior transmission probability between each pair of cases. S5 constructs a weighted directed propagation graph based on posterior propagation probability and reconstructs the infectious disease transmission chain through graph structure optimization.

2. The method for constructing an infectious disease transmission chain based on mechanism-process coupling as described in claim 1, characterized in that, The case data in step S1 includes individual travel information, inter-individual relationship data, individual comprehensive information, and nucleic acid test information; Individual travel information includes trip ID, track date, trip start and end time, trip origin and destination points (POI) and their latitude and longitude coordinates, main mode of transportation used, and trip duration; Inter-individual relationship data includes relationship tags between two patients, which include at least: same building, colleagues, cohabiting, chatting together, traveling together, and simultaneous location; Individual comprehensive information includes the POI and latitude and longitude of the place of residence and workplace, as well as trajectory event descriptions including activity types; activity types include eating together, living together, and being in the same place at the same time; The testing information includes the date of diagnosis.

3. The method for constructing an infectious disease transmission chain based on mechanism-process coupling as described in claim 2, characterized in that, Step S2 specifically includes: S21 extracts individual travel information for each case from the case data; S22 for any two cases and traversing all its travel records ,in Case The itinerary collection Case A collection of itineraries; S23 Determine the travel distance With itinerary Does spatiotemporal overlap exist? S24 If a certain itinerary is... If the spatiotemporal overlap condition is met, it is included in the effective spatiotemporal overlap set, and the original propagation score is calculated. S25 normalizes the original transmission scores of all case pairs to obtain the initial transmission probability.

4. The method for constructing an infectious disease transmission chain based on mechanism-process coupling as described in claim 3, characterized in that, The individual travel information in step S21 includes multiple trip records, each of which includes the trip start time, trip end time, geographical coordinates of the trip start point, and geographical coordinates of the trip end point. In step S23, determine the travel distance. With itinerary Does spatiotemporal overlap exist? Specifically, it includes: Time overlap determination: If the start time of the later segment of the trip is earlier than the end time of the earlier segment, then the time overlap duration is calculated; this time overlap duration is the difference between the end time of the earlier segment and the start time of the later segment. ; Among them, the itinerary Including cases The Start time of the journey and the case The End time of the trip, itinerary Including cases The Start time of the journey and the case The End time of the trip; The later start time of the two trips mentioned above refers to the later start time of each person's trip; the earlier end time of the two trips mentioned above refers to the earlier end time of each person's trip. Spatial nearest neighbor determination: Treat each segment of the journey as a line segment from the starting point to the ending point, and calculate the journey distance. and itinerary Shortest Euclidean distance ;like If they are spatially adjacent, then they are determined to be spatially adjacent. To set a spatial threshold, the value range is 50 meters to 500 meters; Step S24 specifically includes treating each travel pair that simultaneously satisfies the conditions of temporal overlap and spatial proximity as a potential transmission contact event, and calculating its transmission contribution score based on the overlap duration and spatial distance: the longer the overlap time and the closer the distance, the higher the score. Finally, the scores of all valid travel pairs are summed to obtain the original transmission score of the case pair. The score calculation formula is as follows: in, For cases and cases The original propagation score, For the purpose of recording the itinerary, The set of all travel pairs that satisfy the spatiotemporal overlap condition. For the itinerary and itinerary Duration of time overlap It is the space decay factor. Itinerary and itinerary The shortest Euclidean distance, It is the spatial attenuation scale parameter, when When the spatial attenuation factor is 1, At that time, the spatial attenuation factor is 0.

37. At that time, the spatial decay factor is 0.05; Step S25 involves globally normalizing the original propagation score to obtain an initial propagation probability between 0 and 1: ， in, For cases and cases The original propagation score, For normalized cases and cases The initial propagation probability, It is the highest original transmission score among all case pairs. This is to prevent division by zero of extremely small positive numbers.

5. The method for constructing an infectious disease transmission chain based on mechanism-process coupling as described in claim 4, characterized in that, The infectious disease knowledge graph in step S3 specifically includes knowledge of infectious disease epidemiology, which includes knowledge of the source of infection, transmission process, and susceptible populations; knowledge of the source of infection includes the infectious period, infectiousness time curve, and intergenerational time of infectious diseases. Knowledge of the transmission process includes risks associated with different types of venues and methods of contact; knowledge of susceptible populations includes basic susceptibility.

6. The method for constructing an infectious disease transmission chain based on mechanism-process coupling as described in claim 5, characterized in that, The specific steps of step S4 include: S41 constructing infectious disease mechanism constraints based on the infectious disease knowledge graph, and mapping the infectious disease mechanism constraints to mechanism evidence variables in the probabilistic graphical model; S42 targets any case Constructing mechanism propagation probabilities based on mechanism evidence variables; In the Bayesian network model, S43 uses the initial propagation probability obtained in step S2 as the observation likelihood and the mechanism propagation probability as the prior distribution for fusion and updating to obtain the posterior propagation probability between each pair of cases.

7. The method for constructing an infectious disease transmission chain based on mechanism-process coupling as described in claim 6, characterized in that, Step S41 specifically includes: Subgraphs related to the mechanisms of infectious diseases are extracted from the knowledge graph of infectious diseases. Based on the semantic structure of entities and their relationships in the subgraphs, a probabilistic graphical model structure corresponding to the transmission process of infectious diseases is constructed. The subgraphs are mapped to random variables and their conditional dependencies in the probabilistic graphical model, thereby forming a model structure for characterizing the constraints of the mechanisms of infectious diseases. The constraints of epidemiological mechanisms of infectious diseases include at least a structured representation of the characteristics of the source of infection, the characteristics of the transmission process, and the characteristics of population susceptibility; among which, the characteristics of the source of infection are jointly determined by the infectious period, the infectiousness time curve, and the intergenerational time distribution; The characteristics of the transmission process are jointly determined by the risk of the type of place where the transmission occurs and the risk of the mode of contact between cases; Population susceptibility characteristics are determined by baseline susceptibility; The above-mentioned infectious disease mechanism constraints are discretized into finite levels according to a preset mapping rule and mapped to mechanism evidence variables in a probabilistic graphical model. The value range of the mechanism evidence variables is [0,1] or a preset set of levels. Step S42 specifically includes: For any case Based on structured information related to the mechanisms of infectious diseases in the knowledge graph of infectious diseases, and constructing the mechanism transmission probability. The mechanism propagation probability satisfies in, For cases Transmission to cases The propagation event variables; These are variables that provide evidence of the source of infection, determined by the infectious period, the infectious time curve, and the intergenerational time distribution. These are the evidence variables for the transmission process, determined by the risk of location type and the risk of contact method; These are susceptibility evidence variables, determined by basic susceptibility; the conditional probabilities are predetermined by the mechanistic rules of the knowledge graph in the form of conditional probability tables or conditional probability distributions. Step S43 specifically includes: In the Bayesian network model, the initial propagation probability obtained in step S25 is used as the observation likelihood term, and the mechanism propagation probability obtained in step S42 is used as the prior distribution for fusion and updating. ， in, It is the initial propagation probability obtained in step S2. It is the mechanism propagation probability obtained in step S42. The updated propagation probability is used for fusion. ， in, To integrate the updated propagation probabilities, It is the normalized posterior propagation probability. Cases from the same source Pointing to all possible cases of infection The summation of the fusion components, It is a very small positive number to avoid division by zero errors caused by a denominator of 0.

8. The method for constructing an infectious disease transmission chain based on mechanism-process coupling as described in claim 7, characterized in that, Step S5 specifically includes: S51 constructs a propagation probability matrix based on the pairwise posterior propagation probabilities between cases obtained in step S4; S52 constructs a weighted directed propagation graph based on the propagation probability matrix, where cases are nodes, edges represent propagation relationships, and the weight of an edge is the posterior propagation probability of the corresponding case pair. S53 solves for the propagation chain structure that satisfies the preset structural constraints on the weighted directed propagation graph. The structural constraints include at least the following: each node corresponds to at most one most likely source node; the propagation chain structure is an acyclic directed structure.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the infectious disease transmission chain construction method based on mechanism-process coupling as described in any one of claims 1 to 8.

10. A computer-readable storage medium storing computer instructions thereon, characterized in that, When the computer instructions are executed by the processor, they implement the method for constructing an infectious disease transmission chain based on mechanistic process coupling as described in any one of claims 1 to 8.

Images