Data and mechanism dual-driven early warning method for mosquito-borne infectious disease transmission
Patent Information
- Application Number
- CN202610945308.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-06-29
- Publication Date
- 2026-09-25
AI Technical Summary
然而，现有的传染病风险评估模型仍存在显著遗漏
本方法以 Ross-Macdonald 模型为基础，结合气象数据构建动态再生数，依托高斯混合模型测算城市防疫上限，利用隐马尔可夫模型修正病例数据偏差，再通过有序Logistic 回归完成校准，实现蚊媒传染病风险精准预警与干预阈值测算。
Smart Images
Figure CN122822399A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of infectious disease transmission early warning technology, specifically to a method for early warning of mosquito-borne infectious disease transmission driven by both data and mechanism. Background Technology
[0002] The increasing frequency and scale of outbreaks of diseases such as Kunkenya fever pose a serious threat to global public health security. Therefore, accurately and effectively assessing the outbreak risk of infectious diseases in target cities and calculating intervention needs is beneficial for governments and public health departments to allocate medical resources in a targeted manner and implement precise epidemic prevention and control. Infectious disease prediction models have always been an important tool for understanding and controlling the spread of epidemics. Since the introduction of classic calculus dynamics models and compartment models, mathematical models have become the core means of quantifying the laws of virus transmission; subsequently, researchers have continuously expanded upon them: for example, by introducing meteorological factors (temperature, precipitation) to quantify the objective physical limits of the natural environment on mosquito bites and reproduction; or by combining static social factors (such as land use, economic level, medical resources, etc.) to construct a comprehensive evaluation system with multiple indicators. However, existing infectious disease risk assessment models still have significant omissions. First, most existing studies isolate "deteriorating natural climate" and "weak defenses in single cities," failing to reveal the highly coupled nonlinear game relationship between the explosive force of natural meteorological ecology and the limits of human social intervention. Second, traditional dynamic models often lack rigid constraints on real-world social management capabilities and basic medical resources, making it impossible to accurately calculate the proportion of the population that needs to take proactive protective measures (such as using mosquito repellent, using mosquito nets, or getting vaccinated) to cut off the transmission chain. Finally, original infectious disease case sequences usually have significant lag and underreporting risks; directly substituting them into dynamic models can easily lead to distortion of the initial infection base, resulting in exponential shifts in prediction results. Based on these issues, to address the omissions and limitations of the former models, a comprehensive infectious disease risk assessment method based on multi-model coupling is proposed. This model innovatively decomposes the infectious disease risk of a target city into three core technical variables: environmentally driven transmission potential, social vulnerability constraint capacity, and real-time transmission status. This invention focuses on quantifying the dynamic baseline reproduction number (R0) using the classic Ross-MacDonald dynamic mechanism. Based on this, a Gaussian mixture model (GMM) is used to incorporate social vulnerability (SVI) to form rigid constraints, thereby determining the maximum limit of local proactive intervention. To better reflect the actual urban defense capabilities, a Hidden Markov Model (HMM) was used to decode the original case sequences, which have lag and underreporting risks, identifying the underlying evolutionary stages and transforming discrete observational data into a continuous transmission intensity index (TI), providing a more accurate initial infection base for the differential dynamics model. Finally, an ordered logistic regression model was used to perform nonlinear fitting and statistical calibration of the three indices, forming a semi-parametric risk model. When applied to complex urban public health management, the model achieves the fusion of hidden state decoding and dynamic calculation, and its risk level output and critical intervention threshold calculation are more closely aligned with actual decision-making needs.
[0003] To address this issue, we propose a data- and mechanism-driven early warning method for the spread of mosquito-borne infectious diseases. Summary of the Invention
[0004] To address the shortcomings of existing technologies, this invention provides a data- and mechanism-driven early warning method for the spread of mosquito-borne infectious diseases, thus solving the problems mentioned in the background section.
[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for early warning of mosquito-borne infectious disease transmission driven by both data and mechanisms, comprising the following steps: The target city's environmental-driven dissemination potential, social vulnerability constraints, and real-time dissemination status are broken down into three core technological variables. Based on the classic Ross-Macdonald dynamic model, a biological response function driven by meteorological factors is introduced to construct a dynamic basic meteorological regeneration number. : A social vulnerability index is constructed by clustering and weighting static social indicators using a Gaussian mixture model. Based on this, the maximum limit of active intervention in the city is set. ; Hidden Markov Models are used to perform state decoding on noisy case sequences and output the propagation intensity index. , and serve as the initial input for the dynamic model; Using ordered Logistic Regression , , Nonlinear fitting is performed to output the comprehensive risk level and critical intervention threshold.
[0006] Preferably, the dynamic basic meteorological regeneration data The construction includes: Characterizing vector bite rate based on the Briere equation ; Media mortality rate based on quadratic polynomial ; Characterizing Media Abundance Based on Multivariable Coupled Equations Furthermore, the precipitation lag effect and the normal distribution characteristics of temperature are introduced; Substituting the above function into the basic regeneration number formula of the Ross-Macdonald model, we obtain the weather-driven dynamic regeneration number.
[0007] Preferably, the social vulnerability index The construction includes: Obtain static indicators for the target city, including land cover indicators and socio-economic indicators; The indicators are standardized to construct a static feature matrix; Identify potential urban vulnerability patterns using Gaussian mixture models; Calculate the inter-group variance of each indicator and normalize it to obtain the weights, then perform weighted aggregation to obtain the final weights. ; Through the negative mapping function To reach the maximum limit of proactive intervention in the city.
[0008] Preferably, the propagation intensity index The generation includes: Historical case sequences from multiple cities were used as global panel data to jointly train a hidden Markov model. Initialize the globally shared state transition matrix and observation distribution parameters; Perform the forward-backward algorithm on the target city to calculate the posterior probability of each hidden state; The median of observations under each state is extracted as the baseline intensity, and a continuous propagation intensity index is obtained by weighted synthesis.
[0009] Preferably, the dynamic model incorporates a human intervention rate. As the third core state variable, and to construct a dynamic control regeneration number:
[0010] in The effective intervention equivalent ratio is calculated after adjusting for protection efficiency, and it is assumed that the behavior is constant within a short meteorological window.
[0011] Preferably, the method for calculating the critical intervention threshold includes: Solve The lowest social intervention rate at that time ; like It is determined to be in a local self-healing state; like If the local area is blocked, the intervention target is output. like It was determined to be a cross-domain spillover crisis and marked as a high-risk radiation source.
[0012] Preferably, the comprehensive risk index is defined as follows: Based on the disaster risk paradigm "risk = hazard × vulnerability", and combined with the Cobb-Douglas function form, a multiplicative model is constructed:
[0013] Where α, β, and γ are elasticity coefficients estimated by ordered logistic regression, used to reflect the nonlinear contribution of each dimension to the overall risk.
[0014] Preferably, the method further includes a closed-loop feedback mechanism: The propagation intensity index obtained by decoding the Hidden Markov Model Mapped to the scale of asymptomatic infections, serving as the proportion of infection in the population in the Ross-Macdonald model for the next cycle. The initial value input enables dynamic correction of the initial values of the dynamic model.
[0015] Compared with existing technologies, this invention provides a method for early warning of mosquito-borne infectious disease transmission driven by both data and mechanisms, which has the following beneficial effects: This method is based on the Ross-Macdonald model, combines meteorological data to construct a dynamic reproduction number, relies on a Gaussian mixture model to calculate the upper limit of urban epidemic prevention, uses a hidden Markov model to correct the bias of case data, and then completes the calibration through ordered logistic regression, so as to realize accurate early warning and intervention threshold calculation of mosquito-borne infectious disease risk. Detailed Implementation
[0016] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. 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.
[0017] A data- and mechanism-driven early warning method for the spread of mosquito-borne infectious diseases includes the following steps: The target city's environmental-driven dissemination potential, social vulnerability constraints, and real-time dissemination status are broken down into three core technological variables. Based on the classic Ross-Macdonald dynamic model, a biological response function driven by meteorological factors is introduced to construct a dynamic basic meteorological regeneration number. ; Among them, dynamic basic meteorological regeneration data The construction includes: Characterizing vector bite rate based on the Briere equation ; Media mortality rate based on quadratic polynomial ; Characterizing Media Abundance Based on Multivariable Coupled Equations Furthermore, the precipitation lag effect and the normal distribution characteristics of temperature are introduced; Substituting the above function into the basic regeneration number formula of the Ross-Macdonald model, we obtain the dynamic regeneration number driven by weather. A social vulnerability index is constructed by clustering and weighting static social indicators using a Gaussian mixture model. Based on this, the maximum limit of active intervention in the city is set. ; Among them, the social vulnerability index The construction includes: Obtain static indicators for the target city, including land cover indicators and socio-economic indicators; The indicators are standardized to construct a static feature matrix; Identify potential urban vulnerability patterns using Gaussian mixture models; Calculate the inter-group variance of each indicator and normalize it to obtain the weights, then perform weighted aggregation to obtain the final weights. ; Through the negative mapping function To achieve the maximum limit of proactive intervention in the city; Hidden Markov Models are used to perform state decoding on noisy case sequences and output the propagation intensity index. , and serve as the initial input for the dynamic model; Using ordered Logistic Regression , , Nonlinear fitting is performed to output the comprehensive risk level and critical intervention threshold.
[0018] Among them, the transmission intensity index The generation includes: Historical case sequences from multiple cities were used as global panel data to jointly train a hidden Markov model. Initialize the globally shared state transition matrix and observation distribution parameters; Perform the forward-backward algorithm on the target city to calculate the posterior probability of each hidden state; The median of observations under each state is extracted as the baseline intensity, and a continuous propagation intensity index is obtained by weighted synthesis. Furthermore, the rate of human intervention is introduced into the dynamic model. As the third core state variable, and to construct a dynamic control regeneration number:
[0019] in The effective intervention equivalent ratio is calculated after adjusting for protection efficiency, and it is assumed that the behavior is constant within a short meteorological window.
[0020] Furthermore, the methods for calculating the critical intervention threshold include: Solve The lowest social intervention rate at that time ; like It is determined to be in a local self-healing state; like If the local area is blocked, the intervention target is output. like It was determined to be a cross-domain spillover crisis and marked as a high-risk radiation source.
[0021] Furthermore, the comprehensive risk index is defined as follows: Based on the disaster risk paradigm "risk = hazard × vulnerability", and combined with the Cobb-Douglas function form, a multiplicative model is constructed:
[0022] Where α, β, and γ are elasticity coefficients estimated by ordered logistic regression, used to reflect the nonlinear contribution of each dimension to the overall risk.
[0023] This invention also includes a closed-loop feedback mechanism: The propagation intensity index obtained by decoding the Hidden Markov Model Mapped to the scale of asymptomatic infections, serving as the proportion of infection in the population in the Ross-Macdonald model for the next cycle. The initial value input enables dynamic correction of the initial values of the dynamic model.
[0024] The beneficial effects of this invention are as follows: This method integrates mechanistic and data models. First, based on the classic Ross-Macdonald mosquito-borne infectious disease transmission model, it combines the Briere equation, quadratic polynomial, and multivariate coupling equation to characterize the nonlinear effects of temperature and precipitation on mosquito biting rate, mortality rate, and population abundance, constructing a dynamic meteorological reproduction number and introducing the active intervention rate of the population to obtain a controlled reproduction number considering the role of prevention and control. Then, it uses a Gaussian mixture model to calculate the urban social vulnerability index, delineate the upper limit of epidemic prevention intervention in various regions, derive the critical intervention threshold for cutting off the transmission chain, and classify different risk prevention and control states. At the same time, it uses a hidden Markov model to perform state decoding on case data with underreporting and lag problems to obtain a smooth transmission intensity index, correcting the initial infection base of the dynamic model to improve the inference accuracy. Finally, it relies on ordered logistic regression to perform nonlinear fitting calibration on the three dimensions of environmental transmission potential, social vulnerability, and transmission intensity to form a semi-parametric comprehensive risk model, which can accurately assess the risk of urban mosquito-borne infectious disease outbreaks and calculate intervention needs, providing scientific support for precise public health prevention and control and resource allocation.
[0025] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.
Claims
1. A method for early warning of mosquito-borne infectious disease transmission driven by both data and mechanisms, characterized in that, Includes the following steps: The target city's environmental-driven dissemination potential, social vulnerability constraints, and real-time dissemination status are broken down into three core technological variables. Based on the classic Ross-Macdonald dynamic model, a biological response function driven by meteorological factors is introduced to construct a dynamic basic meteorological regeneration number. ; A social vulnerability index is constructed by clustering and weighting static social indicators using a Gaussian mixture model. Based on this, the maximum limit of active intervention in the city is set. ; Hidden Markov Models are used to perform state decoding on noisy case sequences and output the propagation intensity index. , and serve as the initial input for the dynamic model; Using ordered Logistic Regression Nonlinear fitting is performed to output the comprehensive risk level and critical intervention threshold.
2. The method for early warning of mosquito-borne infectious disease transmission driven by both data and mechanism as described in claim 1, characterized in that, The dynamic basic meteorological regeneration number The construction includes: Characterizing vector bite rate based on the Briere equation ; Media mortality rate based on quadratic polynomial ; Characterizing Media Abundance Based on Multivariable Coupled Equations Furthermore, the precipitation lag effect and the normal distribution characteristics of temperature are introduced; Substituting the above function into the basic regeneration number formula of the Ross-Macdonald model, we obtain the weather-driven dynamic regeneration number.
3. The method for early warning of mosquito-borne infectious disease transmission driven by both data and mechanism as described in claim 1, characterized in that, The social vulnerability index The construction includes: Obtain static indicators for the target city, including land cover indicators and socio-economic indicators; The indicators are standardized to construct a static feature matrix; Identify potential urban vulnerability patterns using Gaussian mixture models; Calculate the inter-group variance of each indicator and normalize it to obtain the weights, then perform weighted aggregation to obtain the final weights. ; Through the negative mapping function To reach the maximum limit of proactive intervention in the city.
4. The method for early warning of mosquito-borne infectious disease transmission driven by both data and mechanism as described in claim 1, characterized in that, The propagation intensity index The generation includes: Historical case sequences from multiple cities were used as global panel data to jointly train a hidden Markov model. Initialize the globally shared state transition matrix and observation distribution parameters; Perform the forward-backward algorithm on the target city to calculate the posterior probability of each hidden state; The median of observations under each state is extracted as the baseline intensity, and a continuous propagation intensity index is obtained by weighted synthesis.
5. The method for early warning of mosquito-borne infectious disease transmission driven by both data and mechanism as described in claim 1, characterized in that, The dynamic model incorporates a human intervention rate. As the third core state variable, and to construct a dynamic control regeneration number: in The effective intervention equivalent ratio is calculated after adjusting for protection efficiency, and it is assumed that the behavior is constant within a short meteorological window.
6. The method for early warning of mosquito-borne infectious disease transmission driven by both data and mechanism as described in claim 1, characterized in that, The method for calculating the critical intervention threshold includes: Solve The lowest social intervention rate at that time ; like It is determined to be in a local self-healing state; like If the local area is blocked, the intervention target is output. like It was determined to be a cross-domain spillover crisis and marked as a high-risk radiation source.
7. The method for early warning of mosquito-borne infectious disease transmission driven by both data and mechanism as described in claim 1, characterized in that, The definition of the comprehensive risk index is: Based on the disaster risk paradigm "Risk = Hazard × Vulnerability", and combined with the Cobb-Douglas function form, a multiplicative model is constructed: Where α, β, and γ are elasticity coefficients estimated by ordered logistic regression, used to reflect the nonlinear contribution of each dimension to the overall risk.
8. The method for early warning of mosquito-borne infectious disease transmission driven by both data and mechanism as described in claim 1, characterized in that, The method also includes a closed-loop feedback mechanism: The propagation intensity index obtained by decoding the Hidden Markov Model Mapped to the scale of asymptomatic infections, serving as the proportion of infection in the population in the Ross-Macdonald model for the next cycle. The initial value input enables dynamic correction of the initial values of the dynamic model.