A complex network-based epidemic prevention and control resource configuration optimization method

Abstract

This invention discloses a method for optimizing the allocation of epidemic prevention and control resources based on complex networks. The method includes: acquiring big data from online social platforms to measure participants' attitudes and opinions on epidemics and corresponding prevention and control measures; designing and executing specific decision-making experiments to measure participants' feedback to different types of incentives and information; analyzing experimental results and data to extract individual behavioral rules in different research scenarios and constructing a new generation of agent-based epidemic transmission models; conducting multi-layer community analysis based on deep learning to achieve accurate community discovery within a multi-layer framework; combining community discovery with impact maximization research to identify key user sets in the multi-layer system, allocate prevention and control resources, explore the evolutionary patterns of epidemics, and achieve optimized control of the epidemic transmission process under multi-dimensional uncertainties in parameters and models. This invention contributes to the rational allocation of epidemic prevention and control resources, providing effective protection for the lives and health of the people.

Description

Technical Field

[0001] This invention relates to the field of resource allocation technology, and in particular to an optimization method for resource allocation in epidemic prevention and control based on complex networks. Background Technology

[0002] Over the past few decades, the world has witnessed numerous large-scale epidemics, posing a serious threat to human life and health and creating enormous challenges to healthcare and social management. Effectively controlling the spread of epidemics has always been a major concern for societies and governments. Given the limited economic and social resources, and considering both epidemic-related variables and social variables influencing their spread, designing and implementing effective public policies and epidemic prevention and control measures is a major challenge for societies and governments, and a severe hurdle for policymakers and those developing related control measures.

[0003] On the other hand, the reasonable but simplistic assumptions of traditional epidemiological mathematical models make it difficult to describe the complex processes of transmission in detail. While modern epidemiological models have extended the limitations of traditional models by considering differences in individual behavior to describe contact patterns between individuals, and the advent of the big data era has provided powerful resources for understanding people's behavior, especially those aspects influencing disease transmission such as commuting and interaction, the interaction between changes in people's behavior and emergency decision-making in the special social context of an epidemic has not been fully considered. Furthermore, most infectious disease models fail to capture the adaptive changes in individual behavior during an outbreak. Current theoretical models lack experimental support for the behavioral interaction rules between individuals, and there is a lack of research combining theory and experiment to provide more realistic and practical models.

[0004] Furthermore, the diverse interaction patterns among individuals allow some network members to exhibit characteristics and connections distinct from other community members. Discovering communities within networks can assist in pandemic immunization, public opinion monitoring, and the prevention and intervention of sudden events. However, traditional community detection methods are limited by computational and storage space and cannot be extended to large networks or networks with high-dimensional features. For large, complex networks, deep learning-based community detection methods can utilize larger amounts of data with minimal preprocessing to uncover deeper network information and complex relationships. However, most current deep learning models are applied to community detection in single-layer networks, lacking application to multi-layer contact networks with complex data structures. Summary of the Invention

[0005] The purpose of this invention is to provide a method for optimizing the allocation of epidemic prevention and control resources based on complex networks, so as to solve the above-mentioned problems.

[0006] The present invention solves the technical problem by adopting the following technical solution:

[0007] An optimization method for epidemic prevention and control resource allocation based on complex networks includes the following steps:

[0008] Step 1: Obtain big data from online social platforms (such as Weibo) to measure participants' attitudes and opinions on the epidemic and corresponding prevention and control measures;

[0009] Step 2: Design and execute specific decision-making experiments to obtain real-world behavioral strategies and measure participants' responses to different types of incentives and information;

[0010] Step 3: Analyze the experimental results and data, extract the rules of individual behavior in different research scenarios, and incorporate the individual behavior decision-making sub-model into the epidemic transmission model based on data-driven approach to build a new generation of agent-based epidemic transmission model;

[0011] Step 4: For the newly constructed agent-based epidemic transmission model, perform multi-layer community analysis based on deep learning to achieve accurate community detection on a multi-layer framework;

[0012] Step 5: Combine community findings with impact maximization research to identify key user sets in multi-layered systems, allocate prevention and control resources, explore the evolutionary patterns of the epidemic, and achieve optimized control of the epidemic transmission process with multi-dimensional uncertainties in parameters and models.

[0013] Furthermore, the method for acquiring big data from social media platforms as described in step 1 includes: Since social media is a unique data collection tool, data collection is primarily based on the Weibo platform, with a focus on in-depth analysis of posts on social media related to the epidemic, prevention and control measures, and related opinions. The obtained data will be screened according to different criteria relevant to the research. Although the data is anonymized, it may contain geolocation information and can provide valuable information about information flow and its impact on individual and collective behavior.

[0014] Furthermore, the design and execution of specific decision-making experiments described in step 2 will involve designing an adapted public goods game to measure participants' prosocial behavioral decisions during an epidemic. The goal is to measure participants' responses to different types of incentives and information (e.g., subsidies, commuting, personal willingness, etc.). The experiments will be conducted through the open platform oTree (otree.org).

[0015] Furthermore, the methods for conducting experiments include: applying the method of separation of variables when designing experiments, considering the impact of individual variables on research results, developing and designing corresponding code and interfaces to ensure that participants understand the experiment, and selecting participants based on socio-demographic variables to minimize sample bias.

[0016] Furthermore, the method for constructing a new generation of agent-based epidemic transmission models described in step 3 includes: exploring compartmentalized models (such as SIS, SIR, SEIR, ...) capable of describing different types of epidemics based on complex network theory. In these models, interactions between agents are modeled through multi-layered networks, with each layer representing a mode of interaction (family, workplace, school, social and cultural venues, etc.). There will be feedback between the epidemiological and behavioral sub-models: the latter determines social activities, thus determining epidemic transmission; conversely, transmission will also influence social activities. Finally, the study investigates the ultimate impact of different measures on epidemics, exploring how to optimize resource allocation from aspects such as public awareness campaigns, subsidies, obligations, and sanctions.

[0017] Furthermore, the method for performing multi-layer community analysis based on deep learning described in step 4 includes: conducting a comparative study with traditional community detection methods, and performing community detection based on convolutional networks. The network structure of each layer in the multi-layer network is used as the data input of the deep learning algorithm to extend the existing algorithm, thereby realizing community detection of multi-layer contact networks.

[0018] Furthermore, step 5, which combines community discovery with impact maximization research, identifies key user sets in multi-layered systems, allocates prevention and control resources, explores the evolutionary patterns of epidemics, and achieves optimized control of the epidemic transmission process under multi-dimensional uncertainties such as parameters and models. The method includes: first, applying the impact maximization algorithm to each layer of the multi-layered framework to find the key node set; then, extending the algorithm to multi-layered networks to explore the set of nodes that maximize impact across the entire social network; based on this, allocating limited resources; exploring the transmission mechanism of epidemics under various conditions; and finally, combining the discovered patterns to achieve optimal resource allocation.

[0019] Beneficial effects:

[0020] This invention optimizes the allocation of resources for epidemic prevention and control by utilizing complex network theory and data mining techniques. Existing technologies typically simulate epidemic transmission by simply improving epidemic models and optimizing transmission parameters. In contrast, this invention combines theory and experiment, involving different scientific disciplines. It integrates agent-based modeling, behavioral experiments, complex network theory, deep learning, data analysis, and mathematical and physical research methods to explore network modeling and epidemic transmission mechanisms. This allows for the prediction of which control measures can effectively curb the spread of an epidemic in its early stages, enabling policymakers to assess the effectiveness of measures early and accurately invest limited resources in appropriate measures. This interdisciplinary research method and technology not only contribute to the rational allocation of epidemic prevention and control resources but also, from another perspective, provide effective protection for the lives and health of the people, and is of great significance to improving society's health protection capabilities. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the method flow described in this invention;

[0022] Figure 2 This is an example diagram illustrating the use of a Weibo data collection tool to acquire data in an embodiment of the present invention.

[0023] Figure 3 This is an example diagram of a public goods game decision-making experiment in an embodiment of the present invention;

[0024] Figure 4 This is a visualization of a multi-layered complex network constructed in an embodiment of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, 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.

[0026] Due to the multidimensional uncertainty of the spread of epidemics, optimizing the allocation of prevention and control resources under complex circumstances with limited economic and social resources poses a significant challenge to medical and health care and social management. Traditional methods face the problem of failing to comprehensively grasp the complexity of individual behavior and organically combine it with the dynamic transmission patterns of epidemics. Therefore, this invention overcomes the shortcomings of existing technologies and provides a method for optimizing the allocation of prevention and control resources based on complex network theory to solve the above problems. This invention combines the design of epidemic models with individual and social behaviors by utilizing complex network theory, game theory, and data mining techniques. It considers the heterogeneity of individual micro-attributes and behaviors during epidemics and improves agent-based transmission models to establish a complete and detailed description of the behavior and interactions of individuals in the population. In addition, it extends deep learning algorithms and influence maximization algorithms to achieve accurate community discovery on multi-layer contact networks and explore the key node set in the entire complex system in order to rationally allocate relevant resources. This method can help policymakers assess which epidemic prevention and control measures are more conducive to investing existing resources. By combining research methods and technologies from different fields, this approach not only helps to optimize the allocation of epidemic prevention and control resources and achieve efficient epidemic control, but also provides effective protection for the lives and health of the people, and is of great significance to improving the society's health protection capabilities.

[0027] This invention provides a method for optimizing the allocation of epidemic prevention and control resources based on complex network theory. It incorporates individual decisions and public policies into the mathematical model of epidemics, primarily based on data obtained from controlled experiments, combined with information obtained from social networks through data mining. This establishes a realistic agent-based epidemic transmission model, conducts multi-layered community discovery, identifies the set of key nodes with the greatest influence, and then studies the social impact of different prevention and control measures and their ultimate impact on the spread of the epidemic. It analyzes and determines more effective epidemic intervention and control measures in real society, and conducts in-depth research on the optimal incentives for successfully implementing these measures. This allows for the prediction of the effectiveness of high-cost prevention and control measures in the early stages of epidemic transmission, improving the efficiency of epidemic control and the rational allocation of prevention and control resources.

[0028] like Figure 1 As shown, this invention provides a method for optimizing the allocation of epidemic prevention and control resources based on complex network theory, comprising the following steps:

[0029] Step 1: Collect data based on social media platforms to pre-measure participants' attitudes and opinions on the epidemic and corresponding prevention and control measures.

[0030] Since information and cognition form the basis of individual decision-making, it is undeniable that online social networks have transformed how people communicate and interact, exerting a unique influence during large-scale or global events (such as pandemics). To make modeling more realistic, the uncertainty of people's behavioral responses in a pandemic environment needs to be considered. This behavior must be measured based on epidemiological circumstances. Since both incentive and coercive measures have associated social and economic costs and will affect the development of society and the pandemic, before designing decision-making experiments, it is advisable to first mine people's decision-making feedback on prevention and control measures through social platforms. For example, the choice of commuting mode can influence the spread of the disease to some extent. In-depth analysis can be conducted on posts about the epidemic, prevention and control measures, and related opinions from social platforms such as Weibo. The obtained data will be filtered according to different criteria relevant to the research. Data can be scraped using Python scripts or raw data can be obtained and analyzed using Weibo data collection tools, such as GooSeeker. Figure 2 As shown.

[0031] Step 2: Design a decision experiment of an adapted public goods game to measure participants’ responses to different incentives and information (such as subsidies, commuting, personal willingness, etc.).

[0032] Designing decision-making experiments on the otree platform, when a country adopts aggressive testing, contact tracing, and isolation policies, it is impossible to predict a priori whether isolated individuals would be willing to maintain isolation, without considering participants' employment, economic status, family circumstances, or other limitations. Since interventions based on individual prosocial responses can be implemented sustainably, the experiments aim to measure participants' prosocial behavioral decisions. For example, an experiment on commuting and personal habits: This involves incorporating information on commuting methods and their impact on epidemic transmission within a framework. Participants will choose commuting methods based on different costs (related to time and money). A simple experimental design would be as follows: Figure 3 As shown, the framework will set up different scenarios: with and without an epidemic, the latter involving feedback between decision-making and epidemic spread. Similarly, in each round of the experiment, participants' rewards are determined by their decisions and the evolution of the epidemic, exhibiting a certain functional relationship, which will in turn be influenced by the decisions of other participants. Similar settings will be used for other decision-making processes, such as remote work.

[0033] Step 3: Analyze the experimental results, incorporate the behavioral sub-model into the transmission model, and construct an agent-based epidemic transmission model.

[0034] First, we consider compartmentalized models of different types of epidemics (e.g., SIS, SIR, SEIR, ...). In these models, the interactions between agents are modeled through multi-layered networks. There is feedback between the epidemiological and behavioral sub-models: the latter determines social activities, thus determining the spread of the epidemic; conversely, the spread will also influence social activities. Through experiments and data mining, we study the ultimate impact of different measures on the epidemic, exploring how to optimize resource investment from aspects such as publicity campaigns, subsidies, obligations, and sanctions to make prevention and control measures more effective. This leads to the construction of a new generation of agent-based multi-layered frameworks, with the model structure of multi-layered complex networks as follows: Figure 4 As shown.

[0035] Step 4: Perform multi-level community discovery for the newly constructed multi-level model.

[0036] This study first explores the effectiveness of different community detection methods on constructed multi-layer contact networks, transitioning from traditional community detection methods to basic deep learning for community detection. During the gradual transition from single-layer to multi-layer networks, the following issues will be considered: the scalability of community detection algorithms in single-layer networks in terms of the number of layers, the differences between interaction types, and the varying sparsity across different layers. Although agents may belong to different communities at different layers, they can potentially play a crucial role in the overall complex system; therefore, the focus is on extending deep learning algorithms to achieve accurate multi-layer community detection.

[0037] Step 5: Combine community findings with impact maximization research to identify the key user set of the multi-layered system.

[0038] Under certain propagation patterns, identifying the key user subset that maximizes the spread of influence is crucial for the dissemination of epidemic prevention and control information and the efficient allocation of resources. Therefore, after achieving multi-layered community discovery, we will focus on expanding typical algorithms for maximizing influence (such as random heuristics and degree-centered heuristics) to explore the key user set of the entire system, allocate limited resources, investigate the patterns of epidemic propagation, and achieve optimal resource allocation.

[0039] Through the above steps, a method for optimizing the allocation of epidemic prevention and control resources based on complex network theory can be constructed. In the early stages of an epidemic, the interaction between individual decision-making feedback on different prevention and control measures and the spread of the epidemic can be pre-assessed. This allows for the exploration of areas to optimize resource investment, allocate resources to key user sets, achieve optimal resource allocation, and improve the effectiveness of prevention and control measures and the efficient allocation of resources. In this technique, we utilize complex network theory to optimize the allocation of limited prevention and control resources. This method can quantitatively assess the interaction between individual decision-making feedback on different prevention and control measures and the spread of the epidemic in the early stages of an outbreak. This allows for the exploration of areas to optimize resource investment, allocate resources to key user sets, achieve the effects of high-cost measures, and ultimately achieve optimal resource allocation, improving the effectiveness of prevention and control measures and the efficient allocation of resources. Furthermore, for large-scale epidemics with complex spatiotemporal structures, this method is more robust than traditional agent-based models.

[0040] Against the backdrop of the increasing spatiotemporal complexity and multidimensional uncertainty of epidemic transmission, and given limited economic and social resources, existing agent-based modeling methods struggle to grasp the impact of the interaction between changes in individual behavior and emergency decision-making in the specific social context of epidemics on the transmission mechanisms and evolutionary patterns of epidemics. This invention, based on complex network theory and data mining techniques, combines theory and experiment on an interdisciplinary basis to understand the impact of changes in individual behavior in the specific environment of epidemics on their transmission patterns and mechanisms. Furthermore, it provides a new perspective and approach to explore the increasingly complex spatiotemporal transmission processes of epidemics, enabling early effectiveness assessments of control measures and the allocation of limited resources.

[0041] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing epidemic prevention and control resource allocation based on complex networks, characterized in that, The method comprises the following steps: Step 1: Obtain big data from online social platforms to measure the attitudes and opinions of participants towards the epidemic and corresponding prevention and control measures; Step 2: Design and implement specific decision experiments to obtain real people's behavior strategies and measure the feedback of participants to different types of incentives and information; Step 3: Analyze the experimental results and data, extract individual behavior rules in different research scenarios, and incorporate individual behavior decision sub-models into the epidemic transmission model based on data-driven, to build a new generation of Agent-based epidemic transmission model; Step 4: Based on the newly built Agent-based epidemic transmission model, conduct multi-layer community analysis based on deep learning, and realize accurate community discovery on a multi-layer framework; Step 5: Conduct influence maximization research combined with community discovery, find the key user set of the multi-layer system, configure prevention and control resources, explore the evolution law of the epidemic, and realize the optimization control of the epidemic transmission process under the multi-dimensional uncertainty of parameters and models; The method for designing and implementing specific decision experiments in step 2 comprises: the experiment is conducted through a modified public goods game to measure their prosocial behavior decisions under the epidemic situation; the goal is to measure the feedback of participants to different types of incentives and information; the experiment is conducted through the open platform oTree; For the experiment, the method comprises: when designing the experiment, the separate variable method is applied to consider the influence of a single variable on the research results, and the corresponding code and interface are developed and designed to ensure that the participants understand the experiment, and the participants are selected according to the social demographic variables to minimize sample bias; The method for building a new generation of Agent-based epidemic transmission model in step 3 comprises: exploring the compartment model for describing different types of epidemics based on complex network theory; in these models, the interaction between Agents is modeled through a multi-layer network, and each layer represents one way of interaction; there is feedback between the epidemiological and behavioral sub-models: the latter determines social activity and thus determines epidemic transmission; conversely, transmission will also affect social activity; finally, the final impact of different measures on the epidemic is studied to explore how to optimize resource allocation from the aspects of propaganda activities, subsidies, obligations, and sanctions.

2. The method of claim 1, wherein, The method for obtaining big data from online social platforms in step 1 comprises: realizing data collection based on the microblog platform, focusing on in-depth analysis of posts on social media related to the epidemic, prevention and control measures, and related opinions, and the obtained data will be filtered according to different criteria related to the research; valuable information about information flow and its impact on individual and collective behavior is obtained.

3. The method of claim 1, wherein, The method for multi-layer community analysis based on deep learning in step 4 comprises: community discovery based on convolutional networks, taking the network structure of each layer in the multi-layer network as the data input of the deep learning algorithm, extending the existing algorithm, and then realizing community discovery in the multi-layer contact network.

4. The method of claim 1, wherein, The influence maximization research combined with community discovery in step 5 finds a multi-layer system key user set, configures prevention and control resources, explores popular disease evolution rules, and realizes the optimal control of the popular disease propagation process with multi-dimensional uncertainty of parameters, models and the like, and the method comprises the following steps: applying the influence maximization algorithm to each layer network of the multi-layer framework to find a key node set; extending the algorithm to the multi-layer network to explore the node set of the influence maximization of the entire social network, allocating limited resources based on the node set, exploring the propagation mechanism of the popular disease under various conditions, and realizing the optimal allocation of resources in combination with the discovered rules.

Images