Method and device for simulating transmission of infectious disease by fusing multi-dimensional dynamic risks, and medium

Abstract

The application provides an infectious disease transmission simulation method and device fusing multi-dimensional dynamic risks, and a medium, relates to the field of infectious disease transmission simulation, fundamentally changes the modeling paradigm of simplifying the infectious risk into a fixed distance threshold in the traditional simulation by introducing an individual dynamic risk vector and a dynamic environment risk field driven by multi-source external data, proposes a dynamic personal risk threshold and a risk-driven sparse contact detection mechanism, which is not an exhaustive calculation of all spatially adjacent subject pairs, but intelligently prioritizes the subject pairs with high comprehensive exposure risk for fine judgment, significantly reducing the computational complexity; through the real-time access of multiple source data such as Internet of Things and city information model through the interface, the simulation environment and the real world risk are dynamically synchronized, having dynamic perception and adaptive ability, and through the dynamic adjustment of the personal risk threshold and the adaptive update of the individual dynamic risk vector, intelligent response to different prevention and control strategies can be realized.

Description

Technical Field

[0001] This invention relates to the field of infectious disease transmission simulation technology, and in particular to infectious disease transmission simulation methods, equipment, and media that integrate multi-dimensional dynamic risks. Background Technology

[0002] Infectious disease transmission simulation technology is a key tool in the field of public health for epidemic prediction, evaluation of prevention and control strategies, and optimization of resource allocation. Simulation methods based on multi-agent modeling (ABM) can meticulously depict the dynamic transmission process of diseases in the population by simulating a large number of agents with autonomous behavioral rules and their interactions, and have become the core technological paradigm in this field.

[0003] However, existing multi-entity infectious disease transmission simulation methods, especially their core contact detection mechanisms, still face the following technical challenges when dealing with the application requirements of large-scale, high-fidelity, and real-time decision-making:

[0004] Current technologies generally employ contact determination methods based on geometric proximity. This approach mechanically simplifies the complex and ever-changing risk of infection into a single, static spatial distance function, completely ignoring the fact that the core of infection risk lies in the exposure dose. This dose depends not only on distance but also on a variety of heterogeneous factors such as environmental ventilation, population density, individual protective equipment levels, exposure duration, and pathogen load. Furthermore, this method treats all individuals equally, failing to reflect the individualized risk characteristics arising from differences in behavioral patterns, health conditions, and social roles (such as healthcare workers and public transportation drivers and passengers). This results in severely insufficient fidelity in simulation results, making it difficult to accurately simulate super-spreading events caused by risk aggregation in reality.

[0005] Existing simulation systems typically rely on pre-defined static parameter libraries to operate. However, real-world infectious disease risk environments are highly dynamic and perceptible. For example, the density of people in public places fluctuates in real time, preventative disinfection operations in localized areas can periodically reduce the viral load in the environment, and meteorological conditions (such as temperature, humidity, and wind speed) directly affect the transmission efficiency of droplets and aerosols. Current technologies lack effective and standardized data fusion interfaces and driving mechanisms, failing to dynamically transform these multi-source, heterogeneous, and real-time external data streams from IoT sensors, urban information systems, and public transportation traffic data into changes in the risk field within the simulation environment. This results in simulation systems becoming static sandboxes detached from real-time reality, significantly diminishing their predictive and early warning value in the face of rapidly changing epidemic situations.

[0006] Current systems are mostly used as offline analysis tools: users set parameters, run simulations, and obtain results. To evaluate different prevention and control strategies, multiple scenarios must be manually designed, numerous simulations executed sequentially, and results compared – a cumbersome and slow process. The system itself lacks the ability to dynamically adjust strategy parameters based on intermediate results during simulation and to quickly converge to a better solution. This makes it difficult to embed simulation technology into the real-time emergency command and decision-making closed loop that requires rapid response, limiting its potential as a proactive decision support system.

[0007] In summary, existing technologies face systemic challenges in the field of multi-entity infectious disease transmission simulation, including low model fidelity, disconnect from dynamic reality, and insufficient system intelligence.

[0008] Therefore, it is necessary to provide a new approach to solve the aforementioned technical problems. Summary of the Invention

[0009] To achieve the above-mentioned objectives and other advantages of the present invention, the first objective of the present invention is to provide a method for simulating the spread of infectious diseases by integrating multi-dimensional dynamic risks, comprising the following steps:

[0010] Construct a virtual environment driven by a multi-dimensional risk field; wherein the virtual environment includes a set of multiple subjects, each subject having a health status, spatial location coordinates, individual movement parameters, and individual dynamic risk vector; and, several dynamic environmental risk fields are defined within the virtual environment, the intensity and distribution of which are driven and updated in real time or periodically by an external data source.

[0011] Within each simulation time step, the spatial coordinates of all subjects are updated according to the individual movement parameters, and the comprehensive exposure risk value borne by the subject in the current time step is calculated synchronously.

[0012] A personal risk threshold is dynamically generated for each subject. Contact detection is performed based on the latest spatial location coordinates of all subjects and their comprehensive exposure risk values ​​to screen out neighboring subjects whose comprehensive exposure risk values ​​exceed the risk threshold, forming a sparse set of potential contact pairs to be inspected.

[0013] For each pair of subjects in the potential contact pair set, the risk-weighted contact intensity is calculated based on the weighted result of their comprehensive exposure risk value and spatial distance, and based on the intensity and the preset infection probability function, it is determined whether a disease transmission event has occurred.

[0014] The health status of relevant subjects is updated based on the transmission determination results, and their individual dynamic risk vectors are adaptively adjusted according to the updated health status.

[0015] The process is repeated from the steps described above: updating the spatial coordinates of all subjects based on the individual movement parameters within each simulation time step, and simultaneously calculating the comprehensive exposure risk value borne by the subject at the current time step; updating the health status of relevant subjects based on the transmission determination results; and adaptively adjusting their individual dynamic risk vectors based on the updated health status until the simulation ends. Finally, the process outputs the epidemic transmission network and risk heat map based on the evolution of the comprehensive exposure risk value.

[0016] Furthermore, the comprehensive exposure risk value is configured as a function of the subject's own individual dynamic risk vector, the intensity of the dynamic environmental risk field at its location, and the subject's current health status.

[0017] Furthermore, the contact detection step based on the latest spatial location coordinates of all subjects and their comprehensive exposure risk value includes:

[0018] A two-layer spatial index structure is constructed for neighborhood search; the first layer is a spatial index based on spatial location, used to quickly locate nearby subjects; the second layer is a risk-sensitive index built based on comprehensive exposure risk value, used to quickly screen spatially nearby subjects at the risk level.

[0019] For each subject, all candidate subjects located in the subject's spatial neighborhood are searched using the first-level spatial index to form an initial selection set;

[0020] Using the second-level risk-sensitive index, candidate subjects whose comprehensive exposure risk value is within a preset correlation range with the subject are retrieved from the initial selection set, forming a fine screening set; wherein, the preset correlation range is dynamically determined according to the subject's personal risk threshold;

[0021] For each candidate subject in the fine screening set, its correlation degree with the subject is calculated, and only candidate subjects with a correlation degree exceeding a preset threshold are included in the contact pair set to be inspected with the subject, thereby realizing the sparse generation of the contact network.

[0022] Furthermore, the dynamic generation of individual risk thresholds for each subject is specifically set based on any one or a combination of the subject's current health status, the subject's historical exposure level, and the simulation global target;

[0023] Among them, when the threshold is set based on the subject's current health status, the threshold for susceptible subjects is lower than that for recovering subjects;

[0024] When the threshold is set based on the subject's historical exposure level, the higher the subject's recent exposure risk, the higher the threshold.

[0025] When setting global objectives based on simulation, the risk thresholds for all subjects are increased to simulate social distancing when simulating strict control strategies.

[0026] Furthermore, the step of updating the strength and distribution of the risk field in real time or periodically driven by an external data source includes:

[0027] Establish real-time data interfaces with external city information models or Internet of Things platforms;

[0028] The received real-time data is mapped to the environmental risk field intensity of the corresponding geographical area in the virtual environment;

[0029] A spatiotemporal interpolation algorithm is used to generate a smooth and continuous environmental risk field distribution map covering the entire simulation area based on discrete real-time data points.

[0030] Furthermore, the method is executed on a heterogeneous computing architecture, including:

[0031] The dynamic calculation of environmental risk field updates and the dynamic adjustment of individual risk thresholds are deployed to the central processing unit for execution, forming a control flow;

[0032] The main position update, the construction of the two-layer spatial index, and the calculation of the risk-weighted contact strength of sparse contact pairs are deployed on the graphics processor or tensor processing unit for parallel execution, forming a data stream;

[0033] By using shared memory or high-speed interconnects, control flow can be injected and updated with low latency to key parameters in the data flow.

[0034] Furthermore, the step of calculating the risk-weighted contact intensity based on the weighted result of its comprehensive exposure risk value and spatial distance, and determining whether a disease transmission event has occurred based on this intensity and a preset infection probability function, includes:

[0035] Based on the comprehensive exposure risk values ​​of the first and second subjects respectively, a risk coupling factor is calculated through a preset weighting function, and a spatial proximity factor is calculated through a distance decay function based on the Euclidean distance between the two. The risk coupling factor and the spatial proximity factor are fused together to calculate the risk-weighted contact intensity, which characterizes the potential danger of this contact.

[0036] The calculated risk-weighted contact intensity is used as input and substituted into a preset contextualized infection probability function to calculate the instantaneous probability of disease transmission caused by this contact.

[0037] Generate a random number uniformly distributed within the interval [0,1]. If the random number is not greater than the instantaneous probability, it is determined that the contact caused a disease transmission event, and the health status of the first subject is updated to infected; otherwise, it is determined that the contact did not cause transmission.

[0038] Furthermore, it also includes the following steps:

[0039] During the simulation, for each disease transmission event that is determined to occur, a transmission event tuple is recorded and associated. The tuple includes at least: transmission time step, source of infection identifier, infected subject identifier, precise location coordinates when the transmission occurs, and risk-weighted contact intensity on which the transmission is triggered and the contribution of key risk factors constituting that intensity.

[0040] After the simulation, a complete and weighted dynamic directed propagation graph is constructed based on all recorded propagation event tuples.

[0041] In response to a user query for a specific subject or a specific spatiotemporal region, based on the dynamic propagation directed graph, at least one of the following analyses is performed: forward tracing, backward tracking, and scene clustering; wherein...

[0042] The forward tracing is configured to start from the infected subject being queried, traverse backward along the directed edges, and identify all possible ingress and propagation chains and key infection nodes.

[0043] The backward tracing is configured to start from the queried source of infection and traverse forward along the directed edges to assess the scope and intensity of secondary transmission caused by it.

[0044] The scenario clustering is configured to cluster propagation events based on the location coordinates of the propagation event and a snapshot of the environmental risk field, thereby identifying high-risk physical scenarios where propagation occurs repeatedly.

[0045] A second objective of the present invention is to provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method.

[0046] A third object of the present invention is to provide a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described method.

[0047] Compared with the prior art, the beneficial effects of the present invention are:

[0048] This invention provides a method, equipment, and medium for simulating infectious disease transmission by integrating multi-dimensional dynamic risks. By introducing individual dynamic risk vectors and a dynamic environmental risk field driven by multi-source external data (such as environmental ventilation and population density), it fundamentally changes the traditional modeling paradigm that simplifies infectious risk to a fixed distance threshold. This invention upgrades each potential contact determination to a multi-factor coupled dynamic risk assessment event, fully considering key heterogeneous factors affecting transmission in reality, such as differences in individual protection, differences in location environment, and time decay effects. This enables the simulation system to more accurately simulate the conditions for super-spreading events, the identification of high-risk locations, and the real effects of different intervention measures (such as improving ventilation and restricting the flow of people in specific locations), thereby providing decision-makers with high-fidelity, realistic epidemic projection results.

[0049] This invention proposes a dynamic individual risk threshold and a risk-driven sparsity contact detection mechanism. Instead of exhaustively calculating all spatially adjacent subject pairs, this mechanism intelligently prioritizes and finely judges subject pairs with high overall exposure risk. This essentially introduces an early pruning strategy into the most time-consuming step of contact detection, significantly reducing computational complexity from the O(N²) level of traditional methods. Combined with collaborative querying using a two-layer spatial index, this method can achieve efficient, real-time or ultra-real-time simulation of cities with populations in the tens of millions while maintaining accuracy. It resolves the fundamental contradiction between scale, accuracy, and speed in existing technologies, making large-scale, high-resolution epidemic simulation possible.

[0050] This invention connects to multi-source data, such as the Internet of Things and urban information models, in real time through interfaces, enabling the simulation environment to dynamically synchronize with real-world risks. It possesses dynamic perception and adaptive capabilities, and through dynamic adjustment of individual risk thresholds and adaptive updates of individual dynamic risk vectors, it can intelligently respond to different prevention and control strategies. This transforms the simulation system from a passive simulator into an active decision support platform that can quickly evaluate the effects of various non-pharmaceutical intervention strategies and support online strategy optimization, greatly enhancing its practical value in emergency command and public health planning.

[0051] The epidemic transmission network and risk heat map output by this invention, based on the evolution of comprehensive exposure risk values, provide insights far exceeding those of traditional infection number curves. It not only shows how many people are infected but also reveals how the risk evolves spatiotemporally and the key nodes and paths of the transmission chain. This output enables public health analysts to conduct retrospective root cause analysis, accurately identifying super-spreaders, high-risk locations, and vulnerable populations. This provides data support and visualization tools for implementing targeted and precise prevention and control measures, greatly enhancing the interpretability of simulation results and their action-oriented decision-making value.

[0052] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it according to the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Specific embodiments of the present invention are given in detail below with reference to the accompanying drawings. Attached Figure Description

[0053] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0054] Figure 1 Flowchart of a simulation method for infectious disease transmission that integrates multi-dimensional dynamic risks;

[0055] Figure 2 The flowchart for updating the strength and distribution of the risk field is driven in real time or periodically by external data sources;

[0056] Figure 3 A flowchart for contact detection based on the latest spatial coordinates of all subjects and their combined exposure risk values;

[0057] Figure 4 Flowchart for risk-weighted propagation determination;

[0058] Figure 5 Here is a flowchart of the method's execution on a heterogeneous computing architecture;

[0059] Figure 6 Flowchart for tracing the transmission chain and conducting backtracking analysis of high-risk scenarios;

[0060] Figure 7 This is a diagram of the epidemic transmission network based on the evolution of the comprehensive exposure risk value;

[0061] Figure 8 This represents the evolution sequence of the risk heatmap at three consecutive simulation time points;

[0062] Figure 9 This is a schematic diagram of a computer device.

[0063] Figure 10 This is a schematic diagram of a computer-readable storage medium. Detailed Implementation

[0064] The present invention will now be further described with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be noted that, without conflict, the various embodiments or technical features described below can be arbitrarily combined to form new embodiments.

[0065] Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this invention.

[0066] The drawing numbers in this application are only used to distinguish the steps in the scheme and are not used to limit the execution order of the steps. The specific execution order is as described in the specification.

[0067] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0068] This invention provides a method and system for simulating infectious disease transmission that integrates multi-source dynamic risk and proactive strategy optimization, addressing the problems of model distortion, low computational efficiency, disconnect from dynamic reality, and lack of intelligent decision support capabilities in existing technologies. The specific solution is as follows: Example 1

[0069] A method for simulating the spread of infectious diseases that integrates multi-dimensional dynamic risks, such as... Figure 1 As shown, it includes the following steps:

[0070] S100. Construct a virtual environment driven by a multi-dimensional risk field; wherein the virtual environment includes a set of multiple subjects, each subject having a health status, spatial location coordinates, individual movement parameters, and individual dynamic risk vector; and, several dynamic environmental risk fields are defined within the virtual environment, the intensity and distribution of which are driven and updated in real time or periodically by an external data source.

[0071] In this embodiment, a virtual city environment is constructed in a computer, and N virtual entities are generated. Each entity is assigned a health status (e.g., susceptibility), initial location, and movement parameters (e.g., commuting destination, movement speed model). Simultaneously, an individual dynamic risk vector is initialized for each entity. These factors represent risk adjustment coefficients derived from vaccination status, mask-wearing habits, and social activity levels. Furthermore, real-time pedestrian flow heat maps and ventilation data from key locations are obtained from urban data platforms and mapped onto a dynamic environmental risk field covering a virtual environment. ,in This indicates the horizontal position in the plane coordinate system of the simulation environment. This indicates the vertical position in the plane coordinate system of the simulation environment. This indicates the time step of the simulation.

[0072] To achieve dynamism and realism in the environmental risk field, the simulation system is deeply integrated with data streams from the external physical world. For example... Figure 2 As shown, the steps for updating the strength and distribution of the risk field, driven in real time or periodically by an external data source, include:

[0073] S110. Establish a real-time data interface with external city information models or Internet of Things platforms;

[0074] For example, a centralized data bus is first designed, and dedicated data adapters are developed for each type of external data source, such as sensor data using the MQTT protocol, urban pedestrian flow data from an HTTP RESTful API, and geographic information system layers in a database. Each adapter is responsible for protocol parsing, data cleaning and normalization, and anomaly handling. Protocol parsing refers to handling specific communication protocols, data cleaning and normalization refers to converting raw data into a unified internal data model, and anomaly handling refers to handling data packet loss and outliers, such as sensor fault values.

[0075] Then, a metadata database is established to record the spatiotemporal coverage, update frequency, data quality (i.e. uncertainty) indicators, and mapping relationship (i.e. affine transformation parameters) with the simulation environment coordinate system of each data source, so that the system can intelligently manage the life cycle of the data source.

[0076] S120. Map the received real-time data to the environmental risk field intensity of the corresponding geographical area in the virtual environment;

[0077] Specifically, mapping functions are defined for different risk types. For example, the basic mapping function corresponding to carbon dioxide concentration is: This maps common indoor carbon dioxide concentration ranges to risk levels [0,1]; where, The original carbon dioxide concentration value to be mapped. This is a clipping / clamping function.

[0078] The basic mapping function corresponding to personnel density (people / square meter) is: Where k is the saturation coefficient, reflecting the diminishing marginal effect of density on risk. Personnel density refers to the number of individuals per unit area.

[0079] Introducing a venue type correction factor For example, the risk multiplier for the same carbon dioxide concentration measured in a hospital fever clinic should be higher than that in an outdoor park. This coefficient is read from a GIS attribute table or a preset configuration.

[0080] Final environmental risk field basic strength The formula for calculating the location of data points is: ,in For the first The weights of these risk factors can be dynamically adjusted based on the different transmission routes of pathogens (droplets, aerosols). For risk factor indexing, For the first Mapping function of risk factors, For the first The raw data values ​​of the risk factors.

[0081] Each mapped risk intensity value is accompanied by a confidence level or variance, which is derived from the quality indicators of the data source and the uncertainty of the mapping function, providing a probabilistic framework for subsequent spatiotemporal interpolation and risk decision-making.

[0082] S130. Using a spatiotemporal interpolation algorithm, an environmental risk field distribution map that covers the entire simulation area and is smooth and continuous is generated based on discrete real-time data points.

[0083] This embodiment uses spatiotemporal kriging interpolation, which can not only handle spatial autocorrelation like traditional spatial interpolation (such as inverse distance weighted IDW), but also explicitly model the decay of data in the time dimension.

[0084] Optionally, construct a spatiotemporal mutation function. , where h is the spatial distance and Δt is the time difference. This function describes the spatiotemporal correlation structure of risk intensity.

[0085] For any scenario in the simulation environment where risk intensity needs to be estimated... The grid points, among which To find all observed data points in the spatiotemporal neighborhood of the specific time point corresponding to the nth time step during the actual execution of the simulation program, an optimal weight is assigned to each observation point by solving the Kriging equations. ,in The index is used to traverse and distinguish all known observations used in this interpolation, ensuring that the estimate is unbiased and has the minimum variance. Simultaneously, this method directly outputs the estimated variance, serving as a measure of uncertainty in the risk field.

[0086] When new data points arrive, local reinterpolation is performed only on grid points within a certain spatiotemporal radius centered on the new data point. For regions without new data, the risk field evolves smoothly according to a pre-defined time decay model, simulating the natural dissipation of risk.

[0087] Finally, an environmental risk intensity matrix and an optional uncertainty matrix covering the entire simulation area and stored in a grid format are obtained for subsequent calculation of the comprehensive exposure risk value.

[0088] S200. Within each simulation time step, update the spatial coordinates of all subjects according to the individual movement parameters, and simultaneously calculate the comprehensive exposure risk value borne by the subject in the current time step.

[0089] Preferably, the comprehensive exposure risk value is configured as a function of the subject's own individual dynamic risk vector, the intensity of the dynamic environmental risk field at its location, and the subject's current health status.

[0090] For example, at each simulation time step First, update the position based on the subject's movement model. Then, the overall exposure risk value for each subject j is calculated. The calculation formula is:

[0091]

[0092] in, This represents the aggregated scalar value of an individual's dynamic risk vector. The intensity of the environmental risk field at the current location of the main body. and These are the weighting coefficients. To be consistent with the subject's current health status Related functions include those that assume a risk to others if an infected individual does not pose a risk to themselves, but no longer calculate the exposure risk.

[0093] S300. Dynamically generate a personal risk threshold for each subject, and perform contact detection based on the latest spatial location coordinates of all subjects and their comprehensive exposure risk values ​​to screen out neighboring subjects whose comprehensive exposure risk values ​​exceed the risk threshold, forming a sparse set of potential contact pairs to be inspected.

[0094] Specifically, such as Figure 3 As shown, the contact detection step based on the latest spatial location coordinates of all subjects and their comprehensive exposure risk value includes:

[0095] S310. Construct a two-layer spatial index structure for neighborhood search; wherein, the first layer is a spatial index based on spatial location, used to quickly locate nearby subjects; the second layer is a risk-sensitive index constructed based on comprehensive exposure risk value, used to quickly screen nearby subjects at the risk level;

[0096] In this embodiment, a uniform grid spatial hash is first constructed based on the positions of all subjects as the first-level index. Simultaneously, based on the positions of all subjects... Values ​​can be used to build risk-sensitive indexes, such as balanced binary search trees (BSTs) or range indexes, to support fast risk value range queries.

[0097] S320. For each subject, use the first-level spatial index to find all candidate subjects located in the subject's spatial neighborhood to form a preliminary selection set.

[0098] S330. Using the second-level risk-sensitive index, candidate subjects whose comprehensive exposure risk value with the subject is within a preset correlation range are retrieved from the initial selection set to form a fine screening set; wherein, the preset correlation range is dynamically determined according to the subject's personal risk threshold.

[0099] For each subject j, calculate the individual risk threshold based on their health status and historical exposure. For example, susceptible subjects Lower, frequently exposed subjects It can be temporarily increased.

[0100] For subject A, spatial grid hashing is used to quickly find all candidate subjects in its own grid and adjacent grids, forming a set. Through risk-sensitive indexing, from Quickly find the V value falling within the range The main body inside forms a fine screening collection. , where K is a coefficient. This step directly filters out spatial neighbors with excessively large differences in risk values.

[0101] S340. For each candidate subject in the fine screening set, calculate its correlation degree with the subject, and only include candidate subjects whose correlation degree exceeds a preset threshold into the contact pair set to be inspected with the subject, thereby realizing the sparse generation of the contact network.

[0102] For example, for For each subject B in the dataset, calculate the precise degree of correlation, such as... and The weighted sum is the product of the distance decay function. Only if the correlation is greater than... Only then will Add to the final set of contact pairs to be checked. This set is much smaller than the full pair set of traditional methods.

[0103] Preferably, the step of dynamically generating a personal risk threshold for each subject is specifically set based on any one or a combination of the subject's current health status, the subject's historical exposure level, and the simulation global target;

[0104] Among them, when the threshold is set based on the subject's current health status, the threshold for susceptible subjects is lower than that for recovering subjects;

[0105] Optionally, a multiplier can be calculated or looked up based on the subject's current health status label. For example, when the health status factor equals 1.0, it represents the baseline state; when the health status factor equals 0.8, the threshold is slightly lowered to simulate the incubation period where the individual may already be infectious but not yet identified, requiring slightly more sensitive detection; when the health status factor equals 0.5, the threshold is significantly lowered. As an infected person is a source of risk, a lower personal risk threshold means the system will detect potential contacts around them with higher sensitivity. Even if others have low risk values, they are more likely to be included in the infected person's contact pair set, thus actively capturing their transmission chain and simulating the behavior of tracking close contacts of infected individuals in reality.

[0106] When the health status factor equals 2.0, the threshold is significantly increased. Recovered individuals possess immunity, resulting in a higher personal risk threshold. This means the system will ignore many low-risk exposures, drastically reducing exposure detection calculations and improving efficiency.

[0107] When the health status factor equals 10.0, the threshold is extremely high. Under physical isolation, almost no effective contact pairs will be generated, achieving zero waste of computing resources.

[0108] This mechanism enables computational resources to be tilted toward high-risk individuals, significantly reducing redundant computation for low-risk individuals while ensuring the accuracy of propagation chain capture.

[0109] When the threshold is set based on the subject's historical exposure level, the higher the subject's recent exposure risk, the higher the threshold.

[0110] Optionally, the time integral or moving average of the comprehensive exposure risk value of subject j within the most recent time window (e.g., the past 48 hours) is calculated as the recent exposure load.

[0111] Individuals who have recently been frequently exposed to high-risk environments have a historical exposure factor greater than 1.0, meaning their individual risk threshold has been adaptively increased. This simulates an individual's risk avoidance behavior or a public health system's focused surveillance, where the system deems the individual's recent risk saturated or warrants close monitoring, thus temporarily raising their exposure criteria and only recognizing exposure when the risk is even higher.

[0112] When setting global objectives based on simulation, the risk thresholds for all subjects are increased to simulate social distancing when simulating strict control strategies.

[0113] Optionally, the global strategy factor is used to simulate non-pharmacological interventions at the societal level on a macroscopic scale. Under normal conditions, the global strategy factor equals 1.0. When the simulation recommends maintaining social distancing, the global strategy factor smoothly rises to the target value (e.g., 1.5) after the strategy takes effect. This means that the individual risk threshold of all subjects is globally increased, resulting in more stringent sparsity screening and a sparser contact network, directly and efficiently simulating the effect of reducing contact across society.

[0114] When simulating the lockdown of a specific high-risk area, the global strategy factor for the entities within that area is set to a maximum value (e.g., 100) so that they have almost no external contact and no impact on entities outside the area.

[0115] The global strategy factor can be slightly reduced (e.g., to 0.9) to simulate the overall increase in the system's sensitivity to potential contact due to the widespread use of detection.

[0116] This mechanism provides an efficient and parameterized way to simulate macro-intervention policies without the need for complex modifications to the micro-behavioral rules of each subject, enabling one-click policy injection and effect evaluation.

[0117] S400. For each pair of subjects in the potential contact pair set, calculate the risk-weighted contact intensity based on the weighted result of their comprehensive exposure risk value and spatial distance, and determine whether a disease transmission event has occurred based on the intensity and a preset infection probability function.

[0118] Specifically, such as Figure 4 As shown, the steps of calculating the risk-weighted contact intensity based on the weighted result of its comprehensive exposure risk value and spatial distance, and determining whether a disease transmission event has occurred based on this intensity and a preset infection probability function, include:

[0119] S410. Based on the comprehensive exposure risk values ​​of the first subject and the second subject, a risk coupling factor is calculated using a preset weighting function. Based on the Euclidean distance between the two subjects, a spatial proximity factor is calculated using a distance decay function. The risk coupling factor and the spatial proximity factor are then fused to calculate the risk-weighted contact intensity, which characterizes the potential danger level of this contact.

[0120] For example, for each pair in the generated set of sparse contact pairs Calculate risk-weighted contact intensity ;

[0121] in, , These are the combined exposure risk values ​​for subjects A and B at the current time step. and This represents the weighting coefficient. For example, set it to... , indicating average risk; it can also be set to asymmetric weights, for example, when simulating one-way transmission (e.g., A is a definite source of infection), we can let Approaching 0 Close to 1. Let Euclidean distance be the distance between subjects A and B. λ is the distance attenuation coefficient, which controls the rate at which the contact strength decreases with increasing distance. The larger the value of λ, the more sensitive the effect of distance. The distance decay function quantizes spatial proximity as a multiplier between [0,1].

[0122] S420. Using the calculated risk-weighted contact intensity as input, substitute it into the preset contextualized infection probability function to calculate the instantaneous probability of disease transmission caused by this contact.

[0123] The calculated Input a preset infection probability function to calculate the instantaneous probability of disease transmission resulting from this contact. The fundamental probability function is:

[0124] in, The basic transmission rate scaling factor is used to abstract the intensity of contact. This maps to the actual biological probability of infection. This represents the maximum infection probability, such as setting it to 0.99.

[0125] This basic function can be extended to more complex contextualized infection probability functions. ,For example:

[0126]

[0127] in, The basic transmission rate of the disease, The environmental risk coefficient of the contact site can be obtained from the dynamic environmental risk field. The individual susceptibility multiplier for subject A can be derived from its individual dynamic risk vector. The individual infectivity multiplier for subject B (as the source of infection) is related to their health status and viral load model.

[0128] S430. Generate a random number uniformly distributed in the interval [0,1]. If the random number is not greater than the instantaneous probability, it is determined that this contact has caused a disease transmission event, and the health status of the first subject is updated to infected. S440. Otherwise, it is determined that this contact has not caused transmission.

[0129] Optionally, for the current contact pair The pseudo-random number generator is invoked to generate a random number R uniformly distributed within the interval [0,1]. If the random number is less than the instantaneous probability, it is determined that this contact led to a disease transmission event, and subject A (if a susceptible person) was infected by subject B; otherwise, it is determined that this contact did not lead to transmission.

[0130] The results of this determination, along with the timestamp, subject ID, location, calculated probability, and random number, can be recorded in the log, providing valuable data for subsequent model calibration, uncertainty analysis, and propagation chain tracing.

[0131] S500. Update the health status of relevant subjects based on the propagation determination results, and adaptively adjust their individual dynamic risk vectors according to the updated health status.

[0132] If transmission is determined to have occurred, the health status of the corresponding subject is updated. Simultaneously, based on the event outcome (e.g., infection) and simulation time, the individual's dynamic risk vector is adaptively adjusted, such as increased infectivity after infection and decreased infectivity after recovery.

[0133] S600. Repeat steps S200 to S500 until the simulation ends, and output the epidemic transmission network and risk heat map based on the evolution of the comprehensive exposure risk value.

[0134] In the active policy optimization mode, the simulation ends when the optimization algorithm converges or finds the optimal policy that satisfies all constraints.

[0135] Optionally, if it is not in strategy optimization mode, or if the main conditions are not triggered for a long time, the simulation will end when the simulation duration reaches the preset upper limit, or when the epidemic naturally disappears (no new infections for N consecutive steps).

[0136] Figure 7 This demonstrates a structured epidemic transmission network constructed by the simulation system from micro-level transmission events. This network is automatically constructed based on recorded transmission event tuples, where each node represents a participant in the transmission, each directed edge represents a successful infection event, and edge attributes fully record the transmission intensity and occurrence time step. The average comprehensive exposure risk value is also shown. The node entity during its infectious period The average risk-weighted contact intensity directly reflects the overall risk level of an individual. A risk-weighted contact intensity greater than 1.2 indicates high-intensity transmission; a risk-weighted contact intensity greater than or equal to 0.8 and less than or equal to 1.2 indicates medium-intensity transmission; and a risk-weighted contact intensity less than 0.8 indicates low-intensity transmission. Side labels indicate the specific risk-weighted contact intensity value and transmission time step. Figure 7The network clearly illustrates a typical transmission pattern starting from a single source of infection (A), spreading through two super-spreaders (B, F) to multiple generations and branches. Because this transmission network is not a simple connection graph but a weighted temporal risk network, it enables precise source tracing. It allows for reverse tracing of the entry path of any case along the edges and can automatically and quantitatively identify super-spreaders and high-risk individuals, based entirely on the internal risk calculations derived from this invention. , The effectiveness of different prevention and control strategies can be quantitatively assessed by calculating changes in network metrics before and after removing key nodes (such as isolating B or F).

[0137] In this embodiment, the spatial locations of all subjects and their combined exposure risk values ​​are obtained. The simulated geospatial space is divided into regular grids. For each grid, all subjects within it are counted, and the composition of these subjects is calculated. The average value is used to obtain the regional risk index of the grid at time t. Then, based on the regional risk index of each grid at time t, the data is obtained from a pre-defined continuous chromatogram (...). Figure 8 The heatmap maps colors from low-risk green to high-risk red, and finally renders and generates a complete heatmap frame.

[0138] Figure 8 The risk heatmap was shown at three consecutive simulation time points. , , The evolutionary sequence of the epidemic directly reflects the dynamic changes in regional risks during the development of the epidemic. (Initially) The heat map showed scattered high-risk points (red areas), indicating that the outbreak originated in local high-risk locations; (Diffusion phase) High-risk points increase, expand, and begin to connect, forming preliminary contiguous risk areas, showing the spatial spread trend of the epidemic along the main movement path; (At its peak) the formation of large-scale, continuous high-risk areas revealed the geographical center and scope of the epidemic's spread.

[0139] Figure 8 The calculated individual-level comprehensive exposure risk value Through scientific grid-based statistical methods, the data is transformed into a macroscopic and visualized spatiotemporal risk field. This visualization not only intuitively displays the spatial distribution and evolution of the epidemic, but more importantly, it directly reveals how risks accumulate and spread spatially. This provides public health departments with a highly valuable spatiotemporal situational awareness tool for regional risk classification, precise resource allocation, and dynamic control decisions.

[0140] To achieve real-time simulation of tens of millions of entities, a CPU-GPU heterogeneous computing architecture is adopted, precisely offloading computing tasks to the most suitable processor based on their characteristics. In some embodiments, such as Figure 5 As shown, the method is executed on a heterogeneous computing architecture and includes:

[0141] S710: The dynamic environmental risk field update calculation and the dynamic adjustment of individual risk thresholds are deployed to the central processing unit for execution, forming a control flow;

[0142] Optionally, for task decomposition: an independent thread or process on the CPU runs a data adapter, consumes external data from the message queue, and performs parsing, cleaning, and normalization.

[0143] In the spatiotemporal Kriging interpolation algorithm, the part that solves the Kriging equations (constructing and solving the linear system) is sequential and requires complex matrix operations. This part is efficiently executed by the CPU's mathematical core library. The CPU is responsible for collecting observation data points in the spatiotemporal neighborhood of each grid point that needs interpolation; constructing the variogram matrix; and solving for the weight vector.

[0144] The CPU logic determines which grid regions need reinterpolation and schedules these local interpolation tasks. The CPU then outputs the latest environmental risk field matrix calculated by the CPU.

[0145] For dynamic adjustment of individual risk thresholds: the CPU parses macro-level instructions from the strategy optimization module or the user and converts them into global modulation factors.

[0146] For each subject, calculate its recent The historical average or integral. Although this calculation can be parallelized, it involves random access to each subject's time series data, which is inefficient on GPUs and more suitable for batch computation by CPUs.

[0147] Calculating the final threshold for each subject is simple but requires reading the subject's health status, making it suitable for CPU execution. The CPU generates an array containing the latest personal risk thresholds for all subjects.

[0148] S720 deploys the main position update, two-layer spatial index construction, and risk-weighted contact strength calculation of sparse contact pairs on a graphics processor or tensor processing unit for parallel execution, forming a data stream;

[0149] Optionally, the GPU acts as the simulation engine, responsible for synchronizing and homogenizing operations on a massive number of subjects. For subject position updates: a CUDA kernel is written, with each GPU thread handling one or a group of subjects.

[0150] The thread reads the individual movement parameters of the subject (such as target position and velocity) and calculates the new coordinates based on a simple motion model (such as random walk or path-based movement). The position updates of hundreds of thousands to millions of subjects are completed simultaneously during a single kernel startup.

[0151] The position coordinates (x, y) and movement parameters of all subjects are stored contiguously in the GPU global memory as a structure array, ensuring merged memory access and maximizing memory bandwidth utilization.

[0152] For the construction of the two-level spatial index: one GPU core calculates the column and row coordinates of the grid to which each subject belongs, uses the GPU's atomic addition operation to count the number of subjects within each grid, and calculates a prefix sum to determine the starting offset of the subject list for each grid. The other kernel scatters the subject IDs into the corresponding grid list space based on the offsets.

[0153] A sorting-based construction method is used to combine the subject's ID with its... As key-value pairs. Use the sort_by_key operation from the GPU high-performance library, sorted by... The values ​​are globally sorted across all entities. After sorting, the entity IDs are ranked according to their... The values ​​are ordered. This ordered array of IDs is itself a highly efficient range index. To query the value of V in... For the main body within the interval, a continuous range of IDs can be obtained by performing two binary search operations on the sorted array.

[0154] For risk-weighted contact strength calculation of sparse contact pairs, one GPU thread is allocated for each contact pair to be inspected. The input thread reads the contact pair. Threads pass and Efficiently collect from global memory , Wait for the data, then calculate the distance. and contact strength Each thread will calculate Write it back to the corresponding position in the global array.

[0155] The S730 enables low-latency injection and updating of key parameters in the data stream by the control flow through shared memory or high-speed interconnect.

[0156] Alternatively, the key to collaboration lies in minimizing data movement and achieving low-latency synchronization. On systems that support unified memory, the CPU and GPU can share the same physical memory, with CPU updates... After the array is created, the GPU kernel can directly access the latest data without explicit copying.

[0157] On systems that do not support unified memory, asynchronous copying is used. After the CPU computes the array, it copies it asynchronously to the GPU's video memory. This copying operation is submitted to a specific CUDA stream and executed concurrently with the compute kernels running on the GPU, hiding the transmission latency. The list of infection events (subject ID pairs) computed by the GPU is sent back to the CPU via asynchronous copying, triggering a state update.

[0158] The CPU is responsible for preparing and launching the GPU kernel. For example, the CPU dynamically calculates the mesh size and block size of the GPU kernel based on the size of the contact pair set, and passes the calculated texture memory or constant memory, array, etc. in step S710 as parameters or through a global memory pointer to the GPU kernel.

[0159] The simulation proceeds in a CPU-GPU collaborative loop. The typical sequence for each time step is as follows: the CPU updates the risk field and threshold, asynchronously injects the risk field and threshold, performs position updates, index construction, contact detection, intensity calculation, Monte Carlo determination in parallel, asynchronously returns new infection events for this step, the CPU receives infection events, updates the subject's health status, prepares for the next step, uses events to ensure the GPU task for this step is completed, and then proceeds to the next step.

[0160] To transform the simulation process into a structured, queryable propagation knowledge graph, thereby enhancing the interpretability and decision support value of the simulation, in some embodiments, such as Figure 6 As shown, it also includes the following steps:

[0161] S810. During the simulation, for each disease transmission event that is determined to occur, a transmission event tuple is recorded and associated. The tuple includes at least: transmission time step, source of infection identifier, infected subject identifier, precise location coordinates when the transmission occurs, and risk-weighted contact intensity on which the transmission is triggered and the contribution of key risk factors constituting the intensity.

[0162] In step S400, when the determination result indicates that propagation has occurred, a structured data record, called a propagation event tuple, is immediately created in memory. In the GPU-executed kernel, when a thread determines that the random number is less than the instantaneous probability, in addition to updating the marked state, the thread also writes the aforementioned fields to a global propagation event list. This list is appended to in GPU memory using a dynamic parallel container. After each time step, the newly added event list is transferred to the CPU's main memory or a database for persistent storage via asynchronous copying. The source of infection identifier and the infected entity identifier are directly associated with the entity set; the precise location coordinates at the time of propagation are associated with the environmental risk field; and the location type ID is associated with geographic information system attributes.

[0163] S820. After the simulation is completed, a complete and weighted dynamic propagation directed graph is constructed based on all recorded propagation event tuples.

[0164] After all simulations are completed, the system constructs a global dynamic directed propagation graph using all propagation event tuples. Specifically, all entities that participated in at least one propagation event in the simulation (whether as a source or target of infection) constitute a node set. Each propagation event tuple corresponds to a directed edge, and the edge weight can be the risk-weighted contact strength corresponding to that edge or the normalized infection probability. Each edge has an attribute, namely the time step of the propagation.

[0165] All propagation event tuples are read from persistent storage. A high-performance graph computation library is used to input these tuples as an edge list, and a directed graph data structure with the aforementioned properties is constructed in parallel. For extremely large graphs, sharded storage can be employed. A hash index is created for nodes (body IDs), and a range index is created for edges based on timestamps to support efficient subsequent queries.

[0166] S830. In response to a user's query for a specific subject or a specific spatiotemporal region, based on the dynamic propagation directed graph, perform at least one of the following analyses: forward tracing, backward tracking, and scene clustering; wherein...

[0167] The forward tracing is configured to start from the infected subject being queried, traverse backward along the directed edges, and identify all possible ingress and propagation chains and key infection nodes.

[0168] The backward tracing is configured to start from the queried source of infection and traverse forward along the directed edges to assess the scope and intensity of secondary transmission caused by it.

[0169] The scenario clustering is configured to cluster propagation events based on the location coordinates of the propagation event and a snapshot of the environmental risk field, thereby identifying high-risk physical scenarios where propagation occurs repeatedly.

[0170] This embodiment provides a query interface that supports interactive, in-depth data mining. This analysis module can be deployed on a CPU and utilizes the constructed graph data structure for computation. For complex queries, parallel algorithms from the graph computation library can be invoked.

[0171] When performing forward tracing, the user enters the ID of the infected subject. Starting from that node, the system performs a reverse breadth-first search or a reverse random walk on the propagation graph, visualizing all possible transmission paths that could have entered the case in a tree or path format. It calculates the reverse betweenness centrality or in-degree weight of the nodes in the graph and identifies the most frequently occurring upstream sources of infection in the entry paths. These may be undetected super-spreaders.

[0172] When performing backward tracing, the user enters the ID of a known source of infection. Starting from that node, the system performs a forward breadth-first search or influence propagation model simulation, lists all subjects directly or indirectly infected by the source of infection, estimates the kernel density of the location of the propagation event on the map, generates a heat map of the geographical spread impact caused by the source of infection, and counts the number of people directly and indirectly infected.

[0173] When performing scenario clustering, users request analysis of high-risk propagation scenarios. For each propagation event tuple, its scenario feature vector is extracted, including spatial coordinates, the intensity of the environmental risk field at the time of occurrence, the location type, and the event risk intensity. A parallel clustering algorithm is used to cluster the feature vectors of all propagation events. The central features of each cluster are analyzed. Each identified cluster is descriptively defined, sorted according to the number of events within the cluster, average propagation intensity, etc., and the locations of events in different clusters are highlighted on a map. Example 2

[0174] A computer device 900, such as Figure 9 As shown, the system includes a memory 910, a processor 920, and a computer program 930 stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of a method for simulating the spread of infectious diseases by incorporating multi-dimensional dynamic risks. For a detailed description of the method, please refer to the corresponding description in the above method embodiments; it will not be repeated here. Example 3

[0175] A computer-readable storage medium, such as Figure 10 As shown, a computer program is stored thereon. When executed by a processor, the computer program implements the steps of a method for simulating the spread of infectious diseases by incorporating multi-dimensional dynamic risks. For a detailed description of the method, please refer to the corresponding description in the above method embodiments, which will not be repeated here.

[0176] The number of devices and processing scale described herein are for the purpose of simplifying the description of the invention. Applications, modifications, and variations of the invention will be readily apparent to those skilled in the art.

[0177] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and illustrations shown and described herein.

[0178] The apparatus, computer device, and non-volatile computer storage medium and method provided in the embodiments of this specification are corresponding. Therefore, the apparatus, computer device, and non-volatile computer storage medium also have similar beneficial technical effects as the corresponding method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the corresponding apparatus, computer device, and non-volatile computer storage medium will not be repeated here.

[0179] Those skilled in the art will also know that, besides implementing the controller in the form of purely computer-readable program code, the same functions can be achieved by logically programming the method steps, making the controller take the form of logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers (PLCs), and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the devices included within it for implementing various functions can also be considered structures within that hardware component. Alternatively, the devices for implementing various functions can be considered as both software units implementing the method and structures within a hardware component.

[0180] The systems, apparatuses, or units described in the above embodiments can be implemented by computer chips or physical entities, or by products with certain functions. For ease of description, the above apparatuses are described separately as various units based on their functions. Of course, when implementing one or more embodiments of this specification, the functions of each unit can be implemented in one or more software and / or hardware.

[0181] Those skilled in the art will understand that the embodiments of this specification can be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the embodiments of this specification can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0182] This specification is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this specification. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0183] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0184] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0185] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0186] This specification may be described in the general context of computer-executable instructions, such as program units, that are executed by a computer. Generally, program units include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This specification may also be practiced in distributed computing environments, where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program units may reside in local and remote computer storage media, including storage devices.

[0187] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0188] The above description is merely an embodiment of this specification and is not intended to limit the scope of one or more embodiments of this specification. Various modifications and variations can be made to one or more embodiments of this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of one or more embodiments of this specification.

Claims

1. A method for simulating the spread of infectious diseases by integrating multi-dimensional dynamic risks, characterized in that, Includes the following steps: Construct a virtual environment driven by a multi-dimensional risk field; wherein the virtual environment includes a set of multiple subjects, each subject having a health status, spatial location coordinates, individual movement parameters, and individual dynamic risk vector; and, several dynamic environmental risk fields are defined within the virtual environment, the intensity and distribution of which are driven and updated in real time or periodically by an external data source. Within each simulation time step, the spatial coordinates of all subjects are updated based on the individual movement parameters, and the comprehensive exposure risk value borne by the subject at the current time step is calculated synchronously. The comprehensive exposure risk value is configured as a function of the subject's own individual dynamic risk vector, the intensity of the dynamic environmental risk field at its location, and the subject's current health status. Specifically, the comprehensive exposure risk value is calculated using the following formula: in, The aggregated scalar value of the individual dynamic risk vector of subject i. The dynamic environmental risk field intensity of subject i at its current location, where α and β are weighting coefficients. To match the current health status of subject i Related modulation functions; A personal risk threshold is dynamically generated for each subject. Contact detection is performed based on the latest spatial location coordinates of all subjects and their comprehensive exposure risk values ​​to screen out neighboring subjects whose comprehensive exposure risk values ​​exceed the risk threshold, forming a sparse set of potential contact pairs to be inspected. Specifically, the dynamic generation of individual risk thresholds for each subject is set based on any one or a combination of the subject's current health status, the subject's historical exposure level, and the simulation global target. When set based on the subject's current health status, the threshold for susceptible subjects is lower than the threshold for recovered subjects; When the threshold is set based on the subject's historical exposure level, the higher the subject's recent exposure risk, the higher the threshold. When setting global goals based on simulation, the risk thresholds for all subjects are increased to simulate social distancing when simulating strict control strategies; The contact detection step, based on the latest spatial coordinates of all subjects and their comprehensive exposure risk value, includes: A two-layer spatial index structure is constructed for neighborhood search; the first layer is a spatial index based on spatial location, used to quickly locate nearby subjects; the second layer is a risk-sensitive index built based on comprehensive exposure risk value, used to quickly screen spatially nearby subjects at the risk level. For each subject, all candidate subjects located in the subject's spatial neighborhood are searched using the first-level spatial index to form an initial selection set; Using a second-level risk-sensitive index, candidate subjects whose comprehensive exposure risk value is within a preset correlation range with the subject are retrieved from the initial selection set, forming a fine-screen set; wherein, the preset correlation range is dynamically determined based on the subject's individual risk threshold; For each candidate subject in the fine screening set, its correlation degree with the subject is calculated, and only candidate subjects whose correlation degree exceeds a preset threshold are included in the contact pair set to be inspected with the subject, thereby realizing the sparse generation of the contact network. For each pair of subjects in the potential contact pair set, a risk-weighted contact intensity is calculated based on the weighted result of their comprehensive exposure risk value and spatial distance. Based on this intensity and a preset infection probability function, it is determined whether a disease transmission event has occurred. Specifically, this includes: Based on the comprehensive exposure risk values ​​of the first and second subjects, a risk coupling factor is calculated using a preset weighting function. Then, based on the Euclidean distance between them, a spatial proximity factor is calculated using a distance decay function. The risk coupling factor and the spatial proximity factor are fused to calculate the risk-weighted contact intensity, which characterizes the potential danger of this contact. Specifically, the risk-weighted contact intensity is calculated using the following formula: in, , These are the combined exposure risk values ​​for subjects A and B at the current time step. and These are the weighting coefficients. Let λ be the Euclidean distance between subjects A and B, and λ be the distance attenuation coefficient. Using the calculated risk-weighted contact intensity as input, and substituting it into a preset contextualized infection probability function, the instantaneous probability of disease transmission resulting from this contact is calculated; contextualized infection probability function. ,in, The basic transmission rate of the disease, The environmental risk factor of the location where the contact occurred. The individual susceptibility multiplier for subject A. The individual infectivity multiplier for subject B, This represents the upper limit of the maximum probability of infection. Generate a random number uniformly distributed within the interval [0,1]. If the random number is not greater than the instantaneous probability, it is determined that this contact has caused a disease transmission event, and the health status of the first subject is updated to infected; otherwise, it is determined that this contact has not caused transmission. The health status of relevant subjects is updated based on the transmission determination results, and their individual dynamic risk vectors are adaptively adjusted according to the updated health status. The process is repeated from the steps described above: updating the spatial coordinates of all subjects based on the individual movement parameters within each simulation time step, and simultaneously calculating the comprehensive exposure risk value borne by the subject at the current time step; updating the health status of relevant subjects based on the transmission determination results; and adaptively adjusting their individual dynamic risk vectors based on the updated health status until the simulation ends. Finally, the process outputs the epidemic transmission network and risk heat map based on the evolution of the comprehensive exposure risk value.

2. The infectious disease transmission simulation method integrating multi-dimensional dynamic risk as described in claim 1, characterized in that, The intensity and distribution of the risk field are updated in real time or periodically driven by an external data source. The update steps include: Establish real-time data interfaces with external city information models or Internet of Things platforms; The received real-time data is mapped to the environmental risk field intensity of the corresponding geographical area in the virtual environment; A spatiotemporal interpolation algorithm is used to generate a smooth and continuous environmental risk field distribution map covering the entire simulation area based on discrete real-time data points.

3. The infectious disease transmission simulation method integrating multi-dimensional dynamic risk as described in claim 2, characterized in that, The method is executed on a heterogeneous computing architecture and includes: The dynamic calculation of environmental risk field updates and the dynamic adjustment of individual risk thresholds are deployed to the central processing unit for execution, forming a control flow; The main position update, the construction of the two-layer spatial index, and the calculation of the risk-weighted contact strength of sparse contact pairs are deployed on the graphics processor or tensor processing unit for parallel execution, forming a data stream; By using shared memory or high-speed interconnects, control flow can be injected and updated with low latency to key parameters in the data flow.

4. A method for simulating the spread of infectious diseases by integrating multi-dimensional dynamic risks as described in any one of claims 1 to 3, characterized in that, It also includes the following steps: During the simulation, for each disease transmission event that is determined to occur, a transmission event tuple is recorded and associated. The tuple includes at least: transmission time step, source of infection identifier, infected subject identifier, precise location coordinates when the transmission occurs, and risk-weighted contact intensity on which the transmission is triggered and the contribution of key risk factors constituting that intensity. After the simulation, a complete and weighted dynamic directed propagation graph is constructed based on all recorded propagation event tuples. In response to a user's query for a given subject or spatiotemporal region, based on the dynamic propagation directed graph, at least one of the following analyses is performed: forward tracing, backward tracing, and scene clustering; wherein, The forward tracing is configured to start from the infected subject being queried, traverse backward along the directed edges, and identify its transmission chain and key infection nodes. The backward tracing is configured to start from the queried source of infection and traverse forward along the directed edges to assess the scope and intensity of secondary transmission caused by it. The scenario clustering is configured to cluster propagation events based on the location coordinates of the propagation event and a snapshot of the environmental risk field, thereby identifying high-risk physical scenarios where propagation occurs repeatedly.

5. A computer 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 computer program, it implements the steps of the method as described in any one of claims 1 to 4.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 4.

Images