Disease transmission prediction method and system based on hypergraph sair propagation model

Abstract

The application discloses a disease transmission prediction method and system based on a supergraph SAIR transmission model, relates to the technical field of big data analysis, and models individuals as nodes, constructs an initial supergraph according to community division, adds nodes or generates superedges according to probability, forms a high-fidelity supergraph model with community structure, heavy-tailed distribution and non-self-loop topology, divides the nodes into four states of susceptible, asymptomatic, infected and recovered, deduces state evolution by using a synchronous updating mechanism, applies three targeted strategies of behavior intervention, high-order structure intervention and key node isolation, corrects transmission parameters and outputs a prediction data set. The system is composed of four modules of network construction, supergraph generation, state evolution and intervention prediction. The supergraph topology of the application is in line with a real social network, realizes a great improvement in the efficiency of targeted control and can support efficient prevention and control decisions.

Description

Technical Field

[0001] This invention relates to the field of big data analytics, specifically to a method and system for predicting disease transmission based on the Hypergraph SAIR propagation model. Background Technology

[0002] In complex systems such as public health communication, information dissemination, and social behavior transmission, the transmission process typically relies on network structures for description and analysis. Traditional research usually uses graph network models to describe these systems, where nodes represent individuals and edges represent the contact relationships between individuals, which can only characterize the second-order relationship structure of pairwise interactions.

[0003] However, existing hypergraph generation models cannot simultaneously take into account the three core features of real social networks: community structure, heavy-tailed distribution, and no self-loop topology. Some models have problems such as self-loops, unreasonable connections, and uncontrollable cross-community connections. Traditional models do not consider asymptomatic infected persons, which is inconsistent with the real mechanism. Existing control strategies are only designed for traditional graph networks and do not utilize high-order hypergraph structure information, resulting in poor targeting and low efficiency.

[0004] In summary, existing technologies cannot accurately depict the high-order interaction processes of groups, have insufficient prediction accuracy, and have limited effectiveness of control strategies. There is an urgent need for an integrated solution that can combine high-fidelity hypergraph modeling and targeted control. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for predicting disease transmission based on the hypergraph SAIR transmission model, so as to solve the problems raised in the prior art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a disease transmission prediction method based on the hypergraph SAIR transmission model, the disease transmission prediction method comprising the following steps: Each individual in the system is recorded as a node, forming a node set. A community network is formed according to preset configuration parameters, and an initial hypergraph network is constructed. New nodes are added to the initial hypergraph network with a preset probability. If no new nodes are added, the nodes are connected to generate hyperedges and form a hypergraph model. The states of nodes are categorized in the hypergraph model into susceptible individuals, asymptomatic individuals, infected individuals, and recovered individuals, forming a hypergraph SAIR propagation model, and the states of all nodes are updated through evolution. When the state of all nodes is evolved and updated, a preset control strategy is applied and the propagation result dataset is output.

[0007] Furthermore, the specific process of constructing the initial hypergraph network is as follows: Each individual in the system is defined as a node, forming a node set. Based on preset configuration parameters, different community networks are divided, and a preset number of nodes are allocated in each community network to construct an initial hypergraph network. The configuration parameters include node generation probability, community selection vector, hyperedge generation tensor, hyperedge size sequence, priority connection bias parameter, cross-community connection parameter, baseline transmission probability of asymptomatic individuals, baseline transmission probability of infected individuals, probability of asymptomatic individuals turning into infected individuals, probability of infected individuals recovering, and time step.

[0008] The division of the community network first extracts the community selection vector, and the same community selection vector is recorded as a community preference. The corresponding number of different community networks are divided according to the number of community preferences.

[0009] The node generation probability is a probability parameter p, which is used to control the addition of new nodes. At each time step, a new node is added with probability parameter p. The community selection vector is M = (m1, m2, ..., m...). r ), where m r Let m represent the probability that a node belongs to community r, where m r >0, and m1+m2+...+m r =1; The hyperedge generating tensor P 2×2×...×2 Used to specify the probability of generating a hyperedge, where each element Pi1,…,i r When the community selects an indicator function of 1, it means that the hyperedge contains nodes from that community; The hyperedge size sequence X = (X0, X1, ..., X...) t The size and length distribution of the generated hyperedges are determined by the following: ... In this invention, the lengths of the hyperedges follow a Poisson distribution and are truncated within the range [2, 20]. The priority connection bias parameter γ is a smoothing parameter that determines the probability of generating a new hyperedge when adding a new node; the community node priority connection probability formula is that, during the hyperedge generation process, the probability of a node being selected can be calculated by normalizing the node's degree and the smoothing parameter: ; Where d t (u) represents the degree of node u at time t; d t (v) represents the degree of node v at time t; the priority connection bias parameter γ is a smoothing parameter that determines the probability of generating a new hyperedge when adding a new node, and the weighted degree can be denoted as d. t (u) + γ. Its function is to prevent d t When (u) = 0, the probability is 0, allowing new nodes to also have a chance to be selected. (Denominator) This represents the weighted sum of the degrees of all nodes in the candidate community. Its purpose is to normalize the probability sum to ensure that it equals 1.

[0010] Furthermore, the specific process of forming the hypergraph model is as follows: At each time step of the initial hypergraph network evolution, the current evolution event is determined based on the node generation probability. When the current event is determined to be node addition, the new node is assigned to the corresponding community through the community selection vector. When the current event is determined to be hyperedge generation, the set of communities crossed by the current hyperedge is generated through the hyperedge generation tensor and cross-community connection parameters. Based on the superedge size sequence, determine the total number of nodes for the current generated superedge, and decompose the total number of nodes for the superedge into the number of nodes selected for each community according to the number of communities in the selected community set; Based on the priority connection bias parameter, nodes are selected using the community node priority connection probability formula, and the selected nodes in each community are connected; based on the community size balance parameter, the distribution of the hyperedge generation probability in each community is adjusted to form a hypergraph model.

[0011] At each time step, a new node is introduced into the hypergraph with probability p. This node is assigned to a community based on the probability in the community selection vector. A tensor P is generated with hyperedges with probability 1-p. 2×2×...×2 We select N distinct communities. The total number of nodes in the hyperedge is determined by the hyperedge size sequence, which defines the size and length distribution of the newly generated hyperedge. We decompose the total number of nodes in the hyperedge into the number of nodes selected for each community, i.e., we decompose the random variable into N positive integers such that the sum of these N positive integers equals the random variable. From each selected community, we select a corresponding number of different nodes according to a degree-based priority connection rule. We then connect all the selected nodes to form a new hyperedge.

[0012] The further state division of the nodes within the hypergraph is as follows: The term "susceptible individuals within the supermap" refers to individuals who have not yet been infected with the pathogen and do not possess immunity; the term "asymptomatic infected individuals within the supermap" refers to individuals who have been infected with the pathogen but do not exhibit obvious clinical symptoms; the term "infected individuals within the supermap" refers to individuals who have been infected with the pathogen and exhibit obvious clinical symptoms; and the term "recovered individuals within the supermap" refers to individuals who have recovered from infection and acquired a certain level of immunity, and who no longer participate in the transmission process within a set timescale.

[0013] Furthermore, the specific process of the evolution and update of the hypergraph SAIR propagation model is as follows: If a node is in a susceptible state at time t, then at time t+1 it will have probability B i (t) transitions to an asymptomatic state with probability 1-B i (t) Maintain a susceptible state; If a node is in an asymptomatic state at time t, it will transition to an infected state with probability λ at time t+1 and remain in state A with probability 1-λ. If a node is in an infected state at time t, it will transition to a recovered state with probability μ at time t-1 and remain in an infected state with probability 1-μ. If a node is in a recovered state at time t, it will remain unchanged in subsequent time steps and will no longer participate in the transmission process.

[0014] The probability calculation process, taking the transition from a susceptible state to an asymptomatic state as an example, is given by the following formula: ; Where E i Represents node v i The set of superedges to which it belongs; h is an element of the superedge set; X j (t) represents node v j Is Y a characteristic function of asymptomatic individuals at time t? j (t) represents node v j Is it an indicator function of symptomatic infected individuals at time t? j h represents node v j Whether a node is an indicator function within a hyperedge h is used to characterize the relationship between nodes and the community interaction structure, and is a core foundation for high-order network modeling; therefore, This represents the number of nodes in the hyperedge ℎ that are in an asymptomatic infected state. This represents the number of nodes in the dominant infected state within the hyperedge ℎ.

[0015] Furthermore, a mean-field approximation method can be used to construct a mean-field model. This mean-field approximation method is as follows: ; ; ; ; The probability can be further updated, where E(·) represents the mathematical expectation operator, used to describe the average behavior of various node states during random propagation. Represents node V i The rate of change of the expected value in each state with time t, where the numerator represents the change in state per unit time and the denominator represents the change in time, reflects the dynamic process of node state transition. P(S) i (t)=1) indicates that node v i The probability of being in a susceptible state at any given time; P(X i (t)=1) indicates that node v i The probability of being asymptomatic at any given time; P(Y) i (t)=1) indicates that node vi The probability of being in an symptomatic infection state at any given time.

[0016] To avoid the update order affecting the propagation process, this invention employs a synchronous update mechanism: at time t, the state transition probabilities of all nodes are calculated; based on these probabilities, the states of all nodes are simultaneously updated to time t+1. This mechanism ensures the stability and repeatability of the propagation process.

[0017] The propagation process terminates when the maximum time step T is reached or when there are no nodes in the network in state A or I.

[0018] At each time step, the proportions of susceptible nodes (S(t), asymptomatic nodes (A(t), infected nodes (I(t), and recovered nodes (R(t))) are statistically analyzed. The maximum infection scale (maxI(t), the peak infection time (t), and the final recovery rate (R(T)) are also recorded.

[0019] Furthermore, the specific process of forming the propagation result dataset is as follows: The preset control strategies include behavioral intervention strategies, higher-order structure intervention strategies, and key node isolation strategies. The baseline transmission probability of asymptomatic individuals and the baseline transmission probability of infected individuals are corrected based on behavioral intervention strategies. The existence state of hyperedges is adjusted based on higher-order structure intervention strategies. The transmission participation state of nodes is changed based on key node isolation strategies. This targeted application operation directly modifies the constraints of transmission evolution. Throughout the entire process of node state evolution and update after the application of the preset control strategy, the node state and propagation evolution sequence are continuously collected. The collected data is then processed according to the preset statistical dimensions to form a propagation result dataset.

[0020] The proposed behavioral intervention strategy reduces the probability of transmission by increasing the probability of individual self-protection, for example, by lowering the baseline probability of transmission, thereby reducing the probability of transmission occurring. First, the effect of applying self-protection only to symptomatic nodes is investigated, followed by the effect of applying dual protection to both asymptomatic and symptomatic nodes.

[0021] When only infected nodes take preventative measures, their transmission probability decreases from β2 to τβ2, where 0 < τ < 1. At any given time t, the probability that a susceptible node may transform into an asymptomatic node through hyperedge interaction is: ; The first item represents the risk of exposure to asymptomatic state nodes, and the second item represents the risk of exposure to infected state nodes with reduced infectivity.

[0022] When preventative measures are implemented on nodes in the asymptomatic and infected states, the basic transmission rate β1 of asymptomatic infected nodes decreases to τβ1, and the basic transmission rate β2 of symptomatic infected nodes decreases to τβ2. At time t, each susceptible node can become an asymptomatic carrier through contact with asymptomatic and infected nodes in the same hyperedge, with the probability being: .

[0023] The higher-order structure intervention strategy is to identify large-scale group interaction structures in hypergraph networks and restrict or remove hyperedges, such as randomly removing hyperedges between communities, thereby reducing group propagation channels.

[0024] The critical node isolation strategy identifies critical nodes by their degree values ​​in the hypergraph. When a node's degree value exceeds a set threshold, isolation measures are implemented to temporarily remove it from the propagation network, thereby disrupting the main propagation path and reducing the propagation scale.

[0025] Furthermore, the disease transmission prediction system includes a network construction module, a hypergraph generation module, a state evolution module, and an intervention prediction module; The network construction module is used to complete the initial construction of system nodes and community network; the hypergraph generation module is used to allocate nodes, generate hyperedges and complete the construction of hypergraph model; the state evolution module is used to realize the temporal update and deduction of node state in disease transmission; the intervention prediction module is used to apply control strategies and output disease transmission prediction results.

[0026] Furthermore, the network construction module includes a node modeling unit and a community networking unit; The node modeling unit is used to define system individuals as independent nodes; the community networking unit is used to allocate nodes to corresponding communities to form an initial network based on community parameters; Furthermore, the hypergraph generation module includes a node allocation unit and a hyperedge generation unit; The node allocation unit is used to allocate nodes to the corresponding community network; the hyperedge generation unit is used to select nodes according to the priority connection rule and complete the hyperedge connection.

[0027] Furthermore, the state evolution module includes a state transition unit and a timing deduction unit; The state transition unit is used to complete the transition of four types of node states according to probability; the time-series deduction unit is used to perform continuous evolution and update of node states according to time steps; Furthermore, the intervention prediction module includes a strategy control unit and a data output unit; The strategy control unit is used to execute three types of intervention strategies and modify the propagation evolution constraints; the data output unit is used to collect evolution data and form a propagation result dataset.

[0028] Compared with the prior art, the beneficial effects of the present invention are: 1. The hypergraph model constructed by this invention can simultaneously take into account the three core features of real social networks: community structure, heavy-tailed degree distribution, and no self-loop topology. It can accurately control cross-community connections and modularity, and solve problems such as no self-loops and unreasonable connections. The topology fitting degree is significantly better than the existing mainstream hypergraph models, providing a high-fidelity network foundation for propagation prediction.

[0029] 2. This invention incorporates asymptomatic data and real-world patterns, and employs a synchronous update mechanism to avoid interference from the order of status updates, ensuring a stable and reproducible process.

[0030] 3. This invention proposes three targeted strategies: behavioral intervention, higher-order structure intervention, and key node isolation. Relying on the design of higher-order structures in hypergraphs, it solves the problems of poor targeting and low efficiency of traditional strategies. Attached Figure Description

[0031] Figure 1 This is a flowchart illustrating the disease transmission prediction method based on the Supergraph SAIR transmission model of this invention. Detailed Implementation

[0032] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0033] Example 1: As Figure 1 As shown, this invention provides a technical solution: a disease transmission prediction method based on the Hypergraph SAIR transmission model. The disease transmission prediction method includes the following steps: Each individual in the system is recorded as a node, forming a node set. The node sets are then classified according to the number of communities to construct the initial hypergraph network. The specific process of constructing the initial hypergraph network is as follows: Each individual in the system is defined as a node, forming a node set. Based on preset configuration parameters, different community networks are divided, and a preset number of nodes are allocated in each community network to construct an initial hypergraph network. The configuration parameters include node generation probability, community selection vector, hyperedge generation tensor, hyperedge size sequence, priority connection bias parameter, cross-community connection parameter, baseline transmission probability of asymptomatic individuals, baseline transmission probability of infected individuals, probability of asymptomatic individuals turning into infected individuals, probability of infected individuals recovering, and time step.

[0034] New nodes are added to the initial hypergraph network with a preset probability. If no new nodes are added, the nodes are connected to generate hyperedges and form a hypergraph model. The specific process of forming the hypergraph model: At each time step of the initial hypergraph network evolution, the current evolution event is determined based on the node generation probability. When the current event is determined to be node addition, the new node is assigned to the corresponding community through the community selection vector. When the current event is determined to be hyperedge generation, the set of communities crossed by the current hyperedge is generated through the hyperedge generation tensor and cross-community connection parameters. Based on the superedge size sequence, determine the total number of nodes for the current generated superedge, and decompose the total number of nodes for the superedge into the number of nodes selected for each community according to the number of communities in the selected community set; Based on the priority connection bias parameter, nodes are selected using the community node priority connection probability formula, and the selected nodes in each community are connected; based on the community size balance parameter, the distribution of the hyperedge generation probability in each community is adjusted to form a hypergraph model.

[0035] The states of nodes are categorized in the hypergraph model into susceptible individuals, asymptomatic individuals, infected individuals, and recovered individuals, forming a hypergraph SAIR propagation model, and the states of all nodes are updated through evolution. The state of the node is divided into: The term "susceptible individuals" refers to individuals who have not yet been infected with the pathogen and do not possess immunity; the term "asymptomatic infected individuals" refers to individuals who have been infected with the pathogen but do not exhibit obvious clinical symptoms; the term "infected individuals" refers to individuals who have been infected with the pathogen and exhibit obvious clinical symptoms; and the term "recovered individuals" refers to individuals who have recovered from the infection and acquired a certain level of immunity, and who no longer participate in the transmission process within a set time scale. The specific process of the evolution and update of the hypergraph SAIR propagation model is as follows: If a node is in a susceptible state at time t, then at time t+1 it will have probability B i (t) transitions to an asymptomatic state with probability 1-B i(t) Remain in a susceptible state; if a node is in an asymptomatic state at time t, it will transition to an infected state at time t+1 with probability λ, and remain in state A with probability 1-λ; if a node is in an infected state at time t, it will transition to a recovered state at time t-1 with probability μ, and remain in an infected state with probability 1-μ; if a node is in a recovered state at time t, it will remain unchanged in subsequent time steps and will no longer participate in the transmission process. When the state of all nodes is evolved and updated, a preset control strategy is applied and the propagation result dataset is output; The specific process of forming the propagation result dataset: The preset control strategies include behavioral intervention strategies, higher-order structure intervention strategies, and key node isolation strategies. The baseline transmission probability of asymptomatic individuals and the baseline transmission probability of infected individuals are corrected based on behavioral intervention strategies. The existence state of hyperedges is adjusted based on higher-order structure intervention strategies. The transmission participation state of nodes is changed based on key node isolation strategies. This targeted application operation directly modifies the constraints of transmission evolution. Throughout the entire process of node state evolution and update after the application of the preset control strategy, the node state and propagation evolution sequence are continuously collected. The collected data is then processed according to the preset statistical dimensions to form a propagation result dataset.

[0036] For example, based on the defined hypergraph model, with an initial time t=0, network nodes are divided into r non-overlapping communities. At each time step, a new node is added with probability p, and its community affiliation is assigned using a community selection vector M. New hyperedges are generated with probability 1-p, and nodes are selected into hyperedges following a degree-based priority connection principle. By adjusting the cross-community connection parameter α, the modularity of the hypergraph is precisely controlled. Experiments have verified that, within the range of 2-20 communities and 1000-100000 nodes, the power-law goodness of fit of the degree distribution exceeds 0.95, the modularity adjustment error is less than 0.02, no self-loop structures are generated, and the fit of the topological features to real social networks is significantly improved compared to existing mainstream hypergraph models, providing a high-fidelity topological foundation for subsequent propagation modeling.

[0037] The specific parameters are described in the table below:

[0038] Based on the proposed hypergraph model, various hypergraphs can be generated to study their degree distributions. The probability of cross-community connections is fixed at α=0.2, the probability of node generation is set to p=0.5, and the number of hyperedges is defined as X. t ~Poisson (2≤|e|≤20), number of communities r=5. In SuperMap 10 3 The node and the hypergraph contain 10 4There are 10 nodes. The degree distribution and goodness-of-fit distribution of the hypergraph both follow a power-law distribution.

[0039] To evaluate the robustness of the model results, a repeated experimental design was employed. For each parameter pair, 10 independent hypergraph samples were generated, and the modularity was calculated using the Leiden algorithm. The arithmetic mean was used as the Q-modality value for the corresponding parameter set. The results show that the model can precisely control the community characteristics of the hypergraph: as α decreases, the modularity Q-value increases monotonically from 0.27 at α=0.9 to 0.93 at α=0.01, confirming that the model can continuously adjust the community structure from weak to strong by adjusting α. Furthermore, at a fixed α, adjusting the preset number of communities can further optimize the modularity. This is reflected in the Q-value first increasing and then stabilizing with the increase of the number of communities, indicating that the model can simultaneously control community size and partitioning quality.

[0040] In the generated hypergraph network, a randomly selected number of nodes are assigned to the asymptomatic infected state. All other nodes in the network are assigned to the susceptible state. The baseline propagation probabilities of asymptomatic infected nodes and infected nodes to susceptible nodes are set as β1 and β2, respectively. The infection probability of a susceptible node is determined by the combined contribution of all asymptomatic infected nodes and infected nodes within its hyperedge, thus forming a nonlinear propagation mechanism based on the population structure. After initializing the node states in the hypergraph, the iterative process begins. First, the probability matrix of susceptible nodes transitioning to asymptomatic infected states is calculated, and higher-order state transitions are performed accordingly. Subsequently, asymptomatic infected nodes transition to infected states with probability λ, and infected nodes transition to recovered states with probability μ. Recovered nodes no longer participate in the propagation process. The iteration ends when the system reaches a preset termination condition. To improve the statistical stability of the results, the above process is repeatedly tested independently, and the results are averaged to obtain numerical simulation results based on the Monte Carlo method. Furthermore, without loss of generality, this invention constructs a corresponding mean-field model to analytically approximate the system evolution process.

[0041] The results show that the mean-field model maintains good consistency with the Monte Carlo simulation results in terms of overall trend and key state proportion evolution, and can provide a stable upper bound estimate of the realization process. This consistency indicates that the mean-field model can effectively characterize the macroscopic behavior in high-order interactive networks, while the Monte Carlo simulation verifies the rationality of the model from the perspective of microscopic stochastic evolution. The two complement each other at different scales, thereby enhancing the theoretical reliability and interpretability of the model. The behavioral intervention strategy introduces a protective awareness parameter η∈[0,1]. Simultaneous self-protection measures at both asymptomatic and infected states achieve a proportional reduction effect; in the calculation formula, β1 and β2 are replaced with (1-η)β1 and (1-η)β2. As η increases, the gap between the two strategies widens significantly. The dual-node synchronous protection strategy of this invention is not simply about expanding the protected population, but is based on the principle revealed by the SAIR model of this invention that "asymptomatic individuals are the core chain of continuous spread." When η=0.95, under the dual-state strategy, a reduction of 75.71% is achieved compared to the single-state method. This result indicates the existence of a key cognitive threshold; beyond this threshold, even a small increase in preventative awareness can bring disproportionate control benefits. For the behavioral awareness synergistic regulation strategy, strengthening preventative behaviors and symptom control in asymptomatic individuals is equally important for effectively alleviating the actual situation. This invention implements a high-order structure intervention strategy by simulating social distancing policies, randomly removing cross-community hyperedges or large clustered hyperedges with a length greater than or equal to a specific threshold (set to 5) to break the chain. The hyperedge-targeted removal strategy is designed entirely based on the community structure and hyperedge length characteristics of the hypergraph model, clearly prioritizing cross-community hyperedges and then large-scale hyperedges with a length ≥5. As the removal ratio increases, the diffusion radius gradually decreases. When the intervention intensity of randomly removing cross-community hyperedges is high, the control effect is better than other intervention policies; however, as the intervention intensity decreases, the control effect rapidly weakens, indicating that strict isolation measures are needed to achieve the desired effect in community isolation. Secondly, removing large-scale hyperedges with a length ≥5, at the same hyperedge removal ratio, improves the suppression effect by more than three times compared to the policy of indiscriminate hyperedge removal, while minimizing the impact on normal social activities, thus solving the core problems of the existing indiscriminate social distancing control policy, which is characterized by a one-size-fits-all approach and high costs. The critical node isolation strategy calculates the participation degree of each node in the hypergraph network and prioritizes the removal of a specified proportion of nodes with the highest degree, simulating fixed-point isolation. The node-targeted isolation strategy of this invention is designed based on the heavy-tailed degree distribution characteristics of the hypergraph model. It prioritizes the isolation of hypergraph height hub nodes supporting the core chain. Removing only 40% of the height nodes can reduce the final impact radius by 78.74%, achieving a control efficiency more than twice that of random isolation at the same proportion. This achieves the optimal control effect of "small-proportion removal, large-proportion suppression." Example 2: This invention provides a disease transmission prediction system based on the hypergraph SAIR propagation model. The disease transmission prediction system includes a network construction module, a hypergraph generation module, a state evolution module, and an intervention prediction module. The network construction module is used to complete the initial construction of system nodes and community network; the hypergraph generation module is used to allocate nodes, generate hyperedges and complete the construction of hypergraph model; the state evolution module is used to realize the temporal update and deduction of node state in disease transmission; the intervention prediction module is used to apply control strategies and output disease transmission prediction results.

[0042] The network construction module includes a node modeling unit and a community networking unit; The node modeling unit is used to define system individuals as independent nodes; the community networking unit is used to allocate nodes to corresponding communities to form an initial network based on community parameters; The hypergraph generation module includes a node allocation unit and a hyperedge generation unit; The node allocation unit is used to allocate nodes to the corresponding community network; the hyperedge generation unit is used to select nodes according to the priority connection rule and complete the hyperedge connection.

[0043] The state evolution module includes a state transition unit and a timing deduction unit; The state transition unit is used to complete the transition of four types of node states according to probability; the time-series deduction unit is used to perform continuous evolution and update of node states according to time steps; The intervention prediction module includes a strategy control unit and a data output unit; The strategy control unit is used to execute three types of intervention strategies and modify the propagation evolution constraints; the data output unit is used to collect evolution data and form a propagation result dataset.

[0044] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A disease transmission prediction method based on the hypergraph SAIR transmission model, characterized in that: The disease transmission prediction method includes the following steps: Each individual in the system is recorded as a node, forming a node set. A community network is formed according to preset configuration parameters, and an initial hypergraph network is constructed. New nodes are added to the initial hypergraph network with a preset probability. If no new nodes are added, random nodes are connected to generate hyperedges and form a hypergraph model. The states of nodes are categorized in the hypergraph model into susceptible individuals, asymptomatic individuals, infected individuals, and recovered individuals, forming a hypergraph SAIR propagation model, and the states of all nodes are updated through evolution. When the state of all nodes is evolved and updated, a preset control strategy is applied and the propagation result dataset is output.

2. The disease transmission prediction method based on the hypergraph SAIR transmission model according to claim 1, characterized in that: The specific process of constructing the initial hypergraph network is as follows: Each individual in the system is defined as a node, forming a node set. Based on preset configuration parameters, different community networks are divided, and a preset number of nodes are allocated in each community network to construct an initial hypergraph network. The configuration parameters include node generation probability, community selection vector, hyperedge generation tensor, hyperedge size sequence, priority connection bias parameter, cross-community connection parameter, baseline transmission probability of asymptomatic individuals, baseline transmission probability of infected individuals, probability of asymptomatic individuals turning into infected individuals, probability of infected individuals recovering, and time step.

3. The disease transmission prediction method based on the hypergraph SAIR transmission model according to claim 1, characterized in that: The specific process of forming the hypergraph model: At each time step of the initial hypergraph network evolution, the current evolution event is determined based on the node generation probability. When the current event is determined to be node addition, the new node is assigned to the corresponding community through the community selection vector. When the current event is determined to be hyperedge generation, the set of communities crossed by the current hyperedge is generated through the hyperedge generation tensor and cross-community connection parameters. Based on the superedge size sequence, determine the total number of nodes for the current generated superedge, and decompose the total number of nodes for the superedge into the number of nodes selected for each community according to the number of communities in the selected community set; Based on the priority connection bias parameter, nodes are selected using the community node priority connection probability formula, and the selected nodes in each community are connected; based on the community size balance parameter, the distribution of the hyperedge generation probability in each community is adjusted to form a hypergraph model.

4. The disease transmission prediction method based on the hypergraph SAIR transmission model according to claim 1, characterized in that: The states of the nodes within the hypergraph are classified as follows: The term "susceptible individuals within the supermap" refers to individuals who have not yet been infected with the pathogen and do not possess immunity; the term "asymptomatic infected individuals within the supermap" refers to individuals who have been infected with the pathogen but do not exhibit obvious clinical symptoms; the term "infected individuals within the supermap" refers to individuals who have been infected with the pathogen and exhibit obvious clinical symptoms; and the term "recovered individuals within the supermap" refers to individuals who have recovered from infection and acquired a certain level of immunity, and who no longer participate in the transmission process within a set timescale.

5. The disease transmission prediction method based on the hypergraph SAIR transmission model according to claim 1, characterized in that: The specific process of the evolution and update of the hypergraph SAIR propagation model is as follows: If a node is in a susceptible state at time t, then at time t+1 it will have probability B i (t) transitions to an asymptomatic state with probability 1-B i (t) Remain in a susceptible state; if a node is in an asymptomatic state at time t, it will change to an infected state at time t+1 with probability λ, and remain in state A with probability 1-λ; if a node is in an infected state at time t, it will change to a recovered state at time t-1 with probability μ, and remain in an infected state with probability 1-μ; if a node is in a recovered state at time t, it will remain unchanged in subsequent time steps and will no longer participate in the transmission process.

6. The disease transmission prediction method based on the hypergraph SAIR transmission model according to claim 1, characterized in that: The specific process of forming the propagation result dataset: The preset control strategies include behavioral intervention strategies, higher-order structure intervention strategies, and key node isolation strategies. The baseline transmission probability of asymptomatic individuals and the baseline transmission probability of infected individuals are corrected based on behavioral intervention strategies. The existence state of hyperedges is adjusted based on higher-order structure intervention strategies. The transmission participation state of nodes is changed based on key node isolation strategies. This targeted application operation directly modifies the constraints of transmission evolution. Throughout the entire process of node state evolution and update after the application of the preset control strategy, the node state and propagation evolution sequence are continuously collected. The collected data is then processed according to the preset statistical dimensions to form a propagation result dataset.

7. A disease transmission prediction system based on the hypergraph SAIR transmission model, which is applied to the disease transmission prediction method based on the hypergraph SAIR transmission model as described in any one of claims 1-6, characterized in that: The disease transmission prediction system includes a network construction module, a hypergraph generation module, a state evolution module, and an intervention prediction module. The network construction module is used to complete the initial construction of system nodes and community network; the hypergraph generation module is used to allocate nodes, generate hyperedges and complete the construction of hypergraph model; the state evolution module is used to realize the temporal update and deduction of node state in disease transmission; the intervention prediction module is used to apply control strategies and output disease transmission prediction results.

8. The disease transmission prediction system based on the hypergraph SAIR transmission model according to claim 7, characterized in that: The network construction module includes a node modeling unit and a community networking unit; The node modeling unit is used to define system individuals as independent nodes; the community networking unit is used to allocate nodes to corresponding communities to form an initial network based on community parameters; The hypergraph generation module includes a node allocation unit and a hyperedge generation unit; The node allocation unit is used to allocate nodes to the corresponding community network; the hyperedge generation unit is used to select nodes according to the priority connection rule and complete the hyperedge connection.

9. The disease transmission prediction system based on the hypergraph SAIR transmission model according to claim 7, characterized in that: The state evolution module includes a state transition unit and a timing deduction unit; The state transition unit is used to complete the transition of four types of node states according to probability; the time-series deduction unit is used to perform continuous evolution and update of node states according to time steps; The intervention prediction module includes a strategy control unit and a data output unit; The strategy control unit is used to execute three types of intervention strategies and modify the propagation evolution constraints; the data output unit is used to collect evolution data and form a propagation result dataset.

Images