Individual movement intervention method and device, storage medium and program product

Abstract

The invention discloses an individual movement intervention method and device, a storage medium and a program product, and relates to the technical field of reinforcement learning, and the method comprises the steps: obtaining epidemic disease related data in a plurality of space-time regions obtained by dividing a to-be-controlled city; the epidemic disease related data of the corresponding space-time areas are calculated through reinforcement learning algorithms corresponding to all the agents, target intervention measures corresponding to the space-time areas are obtained, each space-time area corresponds to one agent, all the agents can calculate the epidemic disease related data of all the space-time areas, and the target intervention measures corresponding to the space-time areas are obtained. The amount of data required to be calculated by each agent is reduced, so that the calculation burden is reduced; therefore, even under the condition of dynamic change of large-scale crowds, effective target intervention measures can be generated in real time, and movement intervention is carried out on the individuals in the corresponding space-time areas based on the target intervention measures, so that the individual movement intervention is carried out in time.

Description

Technical Field

[0001] The present application relates to the technical field of reinforcement learning, and in particular to an individual movement intervention method, device, storage medium and program product. Background Art

[0002] Currently, epidemic control strategies are usually customized for each individual, leveraging their specific behaviors and contact networks.

[0003] However, this approach usually faces a huge computational burden when dealing with a large number of individuals in a complex city, especially in the case of dynamic changes in large-scale populations. It is difficult to generate effective epidemic control strategies in real time and cannot conduct individual mobility interventions in a timely manner.

[0004] Therefore, there is a technical problem in the related art that the real-time performance of individual mobile intervention is poor. Summary of the invention

[0005] The main purpose of this application is to provide an individual mobile intervention method, device, storage medium and program product, aiming to solve the technical problem of poor real-time performance of individual mobile intervention.

[0006] To achieve the above objectives, the present application proposes an individual mobility intervention method, the method comprising:

[0007] Obtaining epidemic-related data in multiple spatiotemporal regions obtained by dividing the city to be controlled, each of the spatiotemporal regions corresponds to an intelligent agent;

[0008] By using the reinforcement learning algorithms corresponding to the intelligent agents, the epidemic-related data of the corresponding time and space regions are calculated to obtain target intervention measures for the corresponding time and space regions;

[0009] Based on the target intervention measures, mobile intervention is performed on individuals in the corresponding space-time area.

[0010] In one embodiment, the step of calculating the epidemic-related data of the corresponding spatiotemporal region by the reinforcement learning algorithm corresponding to each of the intelligent agents to obtain the target intervention measures of the corresponding spatiotemporal region includes:

[0011] Calculating global epidemic-related data based on the epidemic-related data in the multiple spatiotemporal regions;

[0012] Through the reinforcement learning algorithms corresponding to the intelligent agents, the epidemic-related data of the corresponding time and space regions and the global epidemic-related data are calculated to obtain the target intervention measures of the corresponding time and space regions.

[0013] In one embodiment, after the step of performing movement intervention on individuals in the corresponding spatiotemporal area based on the target intervention measure, the following steps are included:

[0014] Obtain the number of infected individuals in each spatiotemporal area after mobile intervention;

[0015] The reward value is calculated by a preset reward function based on the number of infected individuals corresponding to each of the spatiotemporal regions, the preset intervention cost, the preset balance weight between health and socio-economy, and the preset tendency weight, wherein the preset tendency weight is used to characterize the tendency to take intervention measures or the tendency to control costs;

[0016] Based on the reward value, the reinforcement learning algorithm is iteratively trained until the goal of minimizing the number of infected people and the intervention cost is reached to update the agent.

[0017] In one embodiment, the preset intervention cost is the sum of the costs of the target intervention measures taken in the corresponding time and space region; the preset balance weight is determined based on the bearing capacity of the economic system and the bearing capacity of the medical system in the corresponding time and space region.

[0018] In one embodiment, the step of performing movement intervention on individuals in the corresponding spatiotemporal area based on the target intervention measure includes:

[0019] Determining the proportion of individuals who can be intervened in each of the time-space regions based on the target intervention measures corresponding to each of the time-space regions and the preset intervention number limits of different intervention measures;

[0020] Based on the proportion of interveneable individuals in each of the space-time regions, target individuals in each of the space-time regions are determined, and mobile intervention is performed on the target individuals in the corresponding space-time regions.

[0021] In one embodiment, the step of determining the target individuals in each of the spatiotemporal regions based on the proportion of interveneable individuals in each of the spatiotemporal regions comprises:

[0022] Determine high-risk places within the first time period based on places visited by confirmed cases within the first time period in the epidemic-related data of each of the spatiotemporal regions;

[0023] determining, based on the epidemic-related data, an infection risk of a first individual who visited the high-risk location within a second time period;

[0024] Determining a social contact risk of the first individual based on the social attribute of the first individual;

[0025] Based on the infection risk, social contact risk and the proportion of intervened individuals, a target individual is determined from the first individual corresponding to each of the spatiotemporal areas.

[0026] In one embodiment, the step of determining the target individual from the first individual corresponding to each of the spatiotemporal regions based on the infection risk, the social contact risk, and the proportion of interveneable individuals includes:

[0027] The first individuals are sorted from high to low according to the infection risk, and when there are individuals with zero infection risk and / or individuals with the same infection risk among the first individuals, the individuals with zero infection risk and / or individuals with the same infection risk are sorted from high to low according to the social contact risk to obtain a sorted second individual, wherein the higher the ranking, the higher the intervention priority;

[0028] Based on the proportion of interveneable individuals and the intervention priority, a target individual is determined from the second individuals.

[0029] In addition, to achieve the above-mentioned objectives, the present application also proposes an individual mobile intervention device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the individual mobile intervention method as described above.

[0030] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the individual mobile intervention method described above are implemented.

[0031] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the steps of the individual movement intervention method described above are implemented.

[0032] One or more technical solutions proposed in this application have at least the following technical effects:

[0033] The present application obtains epidemic-related data in multiple space-time regions obtained by dividing the city to be controlled; through the reinforcement learning algorithms corresponding to each of the intelligent agents, the epidemic-related data in the corresponding space-time region are calculated to obtain target intervention measures for the corresponding space-time region. Since each of the space-time regions corresponds to an intelligent agent, each intelligent agent can calculate the epidemic-related data in each space-time region respectively, and the amount of data required to be calculated by each intelligent agent is reduced, thereby reducing the computational burden; therefore, even in the case of dynamic changes in a large-scale population, effective target intervention measures can be generated in real time, and based on the target intervention measures, mobile intervention is carried out on individuals in the corresponding space-time region, thereby realizing timely individual mobile intervention. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0035] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0036] Figure 1 A flowchart of the first embodiment of the individual mobility intervention method of the present application is provided;

[0037] Figure 2 A schematic diagram of a first scenario provided for the first embodiment of the individual mobility intervention method of the present application;

[0038] Figure 3 A schematic diagram of a second scenario provided in Example 1 of the individual mobility intervention method of the present application;

[0039] Figure 4 A flow chart of the second embodiment of the individual movement intervention method of the present application;

[0040] Figure 5 A schematic diagram of a third scenario provided for the second embodiment of the individual movement intervention method of the present application;

[0041] Figure 6 A schematic diagram of a fourth scenario provided in Example 2 of the individual movement intervention method of the present application;

[0042] Figure 7 This is a schematic diagram of the module structure of the individual mobile intervention device according to an embodiment of the present application;

[0043] Figure 8 Schematic diagram of the device structure of the hardware operating environment involved in the individual mobile intervention method in the embodiment of the present application.

[0044] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0045] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0046] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0047] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device capable of realizing the above functions, an individual mobile intervention device, etc. The following takes an individual mobile intervention device as an example to illustrate this embodiment and the following embodiments.

[0048] Based on this, the present application embodiment provides an individual movement intervention method, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the individual movement intervention method of the present application.

[0049] In this embodiment, the individual movement intervention method includes steps S10 to S30:

[0050] Step S10, obtaining epidemic-related data in multiple space-time regions obtained by dividing the city to be controlled, each of the space-time regions corresponds to an intelligent agent;

[0051] In this embodiment, the individual mobile intervention device may be subordinate to the individual mobile intervention equipment, and the individual mobile intervention equipment may be subordinate to the individual mobile intervention system. The individual mobile intervention system may be a regional multi-agent reinforcement learning control system (R-MARL) created based on ABM (Agent-Based Modeling). Figure 1 .

[0052] In order to cope with the complexity of individual intervention in large-scale cities, the regional multi-agent reinforcement learning control system divides cities that need epidemic control (cities to be controlled) into multiple spatiotemporal regions, where the spatiotemporal regions can be divided based on residential areas or streets.

[0053] In order to carry out epidemic control in each spatiotemporal region respectively, the epidemic-related data in each spatiotemporal region are obtained respectively.

[0054] Specifically, epidemic-related data can be the number of infections, the implementation of intervention measures, individual infection risks, etc.

[0055] Step S20, calculating the epidemic-related data of the corresponding time and space region through the reinforcement learning algorithm corresponding to each of the intelligent agents, and obtaining the target intervention measures of the corresponding time and space region;

[0056] It should be noted that the regional multi-agent reinforcement learning control system assigns an agent to each space-time region. Each agent learns how to take the most effective intervention measures in its own space-time region through the reinforcement learning algorithm. That is, each agent only needs to calculate the epidemic-related data at the space-time regional level and make decisions at the space-time regional level, without the need to customize control measures for individuals.

[0057] Therefore, by using the reinforcement learning algorithms corresponding to each of the intelligent agents, the epidemic-related data of the corresponding space-time region are calculated, and the target intervention measures of the corresponding space-time region can be obtained, which can reduce the calculation amount of the intelligent agent, improve the calculation efficiency, and thus improve the real-time nature of the intervention.

[0058] Specifically, the framework of the reinforcement learning algorithm of each agent can be Actor-Critic. Figure 3 ; Among them, Critic network V π The Actor Network is responsible for estimating the long-term reward value of the agent's actions, while the Actor Network is responsible for determining the best action to optimize the balance between epidemic control and socioeconomic costs. In order to capture time-related patterns and effectively assess the risk of epidemic spread.

[0059] Reference Figure 3 , Long Short-Term Memory (LSTM) network models are used in both the Actor network and the Critic network. The LSTM network model can encode the past sequence of states, helping the regional multi-agent reinforcement learning control system to dynamically adjust the control strategy. The hidden state of LSTM is output through two layers of multilayer perceptron (MLP), and the output results are converted into probability values ​​and state values ​​of different actions. Therefore, the use of the above reinforcement learning algorithm enables the regional multi-agent reinforcement learning control system to make more accurate decisions in a complex and dynamic epidemic environment and effectively suppress the spread of the epidemic.

[0060] Specifically, the specific implementation method of calculating the epidemic-related data of the corresponding spatiotemporal region through the reinforcement learning algorithm corresponding to each of the intelligent agents to obtain the target intervention measures of the corresponding spatiotemporal region may be:

[0061] Based on the epidemic-related data in the multiple space-time regions, global epidemic-related data are calculated; through the reinforcement learning algorithms corresponding to the respective intelligent agents, the epidemic-related data in the corresponding space-time regions and the global epidemic-related data are calculated to obtain target intervention measures for the corresponding space-time regions.

[0062] It should be noted that, refer to Figure 3,In order to improve the calculation accuracy of the target ,intervention measures, while considering the epidemic related data (local ,variables or regional states) in each spatiotemporal region, the ,global epidemic related data (global variables or global states) can also be ,combined to calculate the target intervention measures for the ,corresponding spatiotemporal region.

[0063] Specifically, local variables can be defined as s m,t =(H m,t ,ΔI m,t ,C m,t ,D m,t ), where H m,t Indicates the number of individuals with different health states, ΔI m,t represents the number of new infections per day, C m,t represents the number of various intervention measures, D m,t Indicates the number of individuals whose infection risk is greater than a preset threshold (e.g., 0.1).

[0064] Specifically, the global epidemic-related data can be calculated based on the epidemic-related data in the multiple space-time regions. The global variable is an important variable for comprehensively evaluating the state of the multi-agent reinforcement learning control system in the entire region at time step t. The global variable can be defined as the sum of the local variables of each space-time region.

[0065] Furthermore, the agent in each spatiotemporal region is m,t Make decisions and choose appropriate intervention measures; different levels of intervention measures can be used It indicates that they correspond to the three intervention measures of restricted movement, home isolation and complete isolation within the time and space area.

[0066] Step S30: Based on the target intervention measure, movement intervention is performed on individuals in the corresponding space-time area.

[0067] It can be understood that once the target intervention measures for each space-time region are determined, mobile intervention can be carried out on the individuals in the corresponding space-time region (for example, prompting the individual to carry out mobile intervention, or sending control information to the individual, etc.) to achieve control of the epidemic; the above method does not need to formulate intervention measures for each individual, thereby improving the timeliness of the intervention.

[0068] Specifically, the specific implementation method of performing mobile intervention on individuals in the corresponding spatiotemporal area based on the target intervention measure may be:

[0069] The number of infected individuals in each spatiotemporal area after the mobile intervention is obtained; based on the number of infected individuals corresponding to each of the spatiotemporal areas, the preset intervention cost, the preset balance weight between health and socio-economy, and the preset tendency weight, a reward value is calculated through a preset reward function, wherein the preset tendency weight is used to characterize the tendency to take intervention measures or the tendency to control costs; based on the reward value, the reinforcement learning algorithm is iteratively trained until the goal of minimizing the number of infected people and the intervention cost is achieved, so as to update the agent.

[0070] It should be noted that in order to improve the accuracy of the intervention measures formulated by the agent, feedback can be provided to the agent after each decision made by the agent, referring to Figure 3 , which makes it easier for the intelligent agent to continuously optimize the reinforcement learning algorithm and obtain an intelligent agent with better performance.

[0071] Specifically, the number of infected individuals in each space-time area after mobile intervention can be obtained, and the reward value can be calculated through a preset reward function based on the number of infected individuals corresponding to each space-time area, the preset intervention cost, the preset balance weight between health and socio-economy, and the preset tendency weight.

[0072] Among them, the preset reward function can be expressed as:

[0073]

[0074]

[0075]

[0076] Specifically, for the mth space-time region, represents the number of infected individuals living in the mth spatiotemporal region at time step t, which includes the infected individuals in the spatiotemporal region and individuals infected from the mth space-time region to other space-time regions (where j represents other space-time regions outside the mth space-time region); it should be noted that this number of infected individuals is used to measure the spread of the epidemic in each space-time region to provide feedback on the decision-making of the intelligent agent.

[0077] At the same time, the above preset reward function represents the preset intervention cost of the mth spatiotemporal region at time step t. The preset intervention cost can be obtained by weighted accumulation of intervention measures a in each spatiotemporal region. m,t that is, the preset intervention cost is the sum of the costs of the target intervention measures taken in the corresponding time and space area.

[0078] Moreover, the preset balance weight β in the above preset reward function is used to balance health and social economy, and the preset balance weight is determined based on the bearing capacity of the economic system and the bearing capacity of the medical system in the corresponding time and space region; the specific calculation method of the preset balance weight can be: Among them, ω Q represents the bearing capacity of the economic system in the corresponding time and space region, ω I Indicates the carrying capacity of the medical system in the corresponding time and space region.

[0079] In addition, the preset tendency weight α in the above preset reward function is used to characterize the tendency of the agent to take intervention measures. For example, if the α value is greater than the preset value, it is considered that the regional multi-agent reinforcement learning control system is more inclined to take intervention measures rather than control costs, so that the trained agent can prioritize controlling the risk of infection rather than controlling the cost of control.

[0080] Furthermore, based on the reward value, the reinforcement learning algorithm is iteratively trained until the goal of minimizing the number of infected people and the intervention cost is achieved, so as to update the intelligent agent. This continuous reinforcement learning method can ensure that the intelligent agent can determine the optimal intervention measures while minimizing the number of infected people and the intervention cost.

[0081] In this embodiment, through the regional multi-agent reinforcement learning system, the complex individual-level control problem is simplified to a regional-level control problem, which greatly reduces the computational complexity and enables efficient intervention in large-scale urban scenarios; by using local variables (such as the number of infected people, the intensity of intervention measures, etc.), the regional multi-agent reinforcement learning system can select the optimal intervention measures for each spatiotemporal region to achieve a balance between suppressing the spread of the epidemic and reducing socioeconomic costs; by combining spatiotemporal regional control with individual intervention, the real-time and computational complexity problems of individual mobility control in large-scale urban environments are solved, thereby improving real-time performance.

[0082] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above-mentioned embodiment 1 can refer to the above introduction, and will not be repeated later. Figure 4 , step S30 includes steps S01-S02:

[0083] Step S01, determining the proportion of intervened individuals in each of the time-space regions based on the target intervention measures corresponding to each of the time-space regions and the preset intervention number limits of different intervention measures;

[0084] In this embodiment, in order to improve the accuracy of intervention, individual risk assessment can be combined with intelligent control of spatiotemporal regions to dynamically optimize individual mobile intervention, aiming to solve the computational complexity and real-time problems of epidemic prevention and control in large-scale urban environments.

[0085] Specifically, after the intelligent agent determines the target intervention measures, the regional multi-agent reinforcement learning system can determine the proportion of intervened individuals in each of the space-time regions based on the target intervention measures corresponding to each of the space-time regions and the preset intervention number limits for different intervention measures.

[0086] Among them, the proportion of intervened individuals corresponding to different intervention measures can be defined as: and

[0087] Specifically, the proportion of intervened individuals corresponding to the target intervention measure may be determined based on the preset intervention number limits of different intervention measures. For example, the calculation method may be expressed as:

[0088]

[0089] Where κ=[κ (c) ,κ (q) ,κ (iso) ], κ represents the preset intervention number limit (capacity limit) of different intervention measures to ensure that the intervention measures will not exceed the resource capacity.

[0090] Based on the above calculation formula, we can know that the proportion of individuals that can be intervened is d m,t is the target intervention measure a corresponding to the spatiotemporal region m,t The term-by-term product (Hadamard product) of the capacity limit κ is used to convert the intervention measures selected by the spatiotemporal regional agent into the actual proportion of individuals that can be intervened.

[0091] The above approach can ensure that the regional multi-agent reinforcement learning control system can dynamically adjust the coverage of intervention measures during mobile intervention and will not exceed the actual available public resource capacity.

[0092] Step S02, based on the proportion of interveneable individuals in each of the spatiotemporal regions, determine the target individuals in each of the spatiotemporal regions, and perform mobile intervention on the target individuals in the corresponding spatiotemporal regions.

[0093] Furthermore, in order to accurately perform mobility intervention on some individuals in the spatiotemporal region, based on the target individuals in each of the spatiotemporal regions determined above, mobility intervention can be performed on the target individuals in the corresponding spatiotemporal region.

[0094] Specifically, the specific implementation of determining the target individuals in each of the spatiotemporal regions based on the proportion of interveneable individuals in each of the spatiotemporal regions may be:

[0095] Based on the places visited by confirmed cases in the first time period in the epidemic-related data of each of the space-time regions, the high-risk places in the first time period are determined; based on the epidemic-related data, the infection risk of the first individual who visits the high-risk places in the second time period is determined; based on the social attributes of the first individual, the social contact risk of the first individual is determined; based on the infection risk, social contact risk and the proportion of intervened individuals, the target individual is determined from the first individual corresponding to each of the space-time regions.

[0096] It should be noted that the individual infection risk dynamic assessment model (I-RAM) can be used to analyze the temporal and spatial behavior of individuals and dynamically assess the infection risk of each individual. Figure 5 .

[0097] Specifically, the individual infection risk dynamic assessment model is used to determine the first time period (|t-τ a ,t)) places visited by confirmed cases within the period, such as schools, offices, communities, etc.; for each place, the regional multi-agent reinforcement learning control system determines the number of infection cases in the place within the first time period based on the epidemic-related data, thereby determining the infection risk of the place, and finally determining the high-risk places. For example, if the number of infection cases visiting the place exceeds a preset number, the place is considered to be a high-risk place.

[0098]

[0099]

[0100]

[0101] in, is the indicator function used to determine the infection case in the past time window τ a Have you visited this place? Indicates the infection risk of the place.

[0102] Furthermore, the infection risk of the first individual who visits the high-risk place within the second time period is calculated through the individual infection risk dynamic assessment model combined with the individual's spatiotemporal behavior data in the epidemic-related data.

[0103] Specifically, the individual infection risk dynamic assessment model dynamically assesses the infection risk of each individual based on the infection risk of the high-risk places they have visited, and considers the time window τ p , to reflect the individual in the time window τ p The probability of contact with an infected case; specifically, the individual's infection risk can be expressed as:

[0104]

[0105]

[0106] Among them, W n,e (t) is an indicator function indicating whether the individual has visited a high-risk place at a specific time (e.g., the second time period), Indicates the individual's risk of infection.

[0107] In addition, the individual infection risk dynamic assessment model also determines the social contact risk of the first individual based on the social attributes of the first individual (for example, the occupation and age of the first individual); specifically, students have a higher frequency of daily contact than the elderly, so students' social contact risk is higher than that of the elderly.

[0108] Furthermore, based on the infection risk, social contact risk and the proportion of interveneable individuals, the target individual can be determined from the first individual corresponding to each of the time and space regions, so as to achieve precise intervention and improve the effectiveness of the intervention.

[0109] Specifically, the specific implementation of determining the target individual from the first individual corresponding to each of the spatiotemporal regions based on the infection risk, the social contact risk, and the proportion of interveneable individuals may be:

[0110] The first individuals are sorted from high to low according to the infection risk. When there are individuals with zero infection risk and / or individuals with the same infection risk among the first individuals, the individuals with zero infection risk and / or the individuals with the same infection risk are sorted from high to low according to the social contact risk to obtain a sorted second individual, wherein the higher the ranking, the higher the intervention priority; based on the proportion of interveneable individuals and the intervention priority, the target individual is determined from the second individuals.

[0111] Specifically, we can first determine the risk of infection based on individual All first individuals are sorted from high to low according to the infection risk. When there are individuals with zero infection risk and / or individuals with the same infection risk among the first individuals, they can be sorted according to their social contact risk. Sort from high to low to get the sorted second individual.

[0112] It can be understood that the higher the ranking of the second individual after sorting, the higher the intervention priority, and the higher the priority for strict isolation; the lower the ranking of the individual, the lower the intervention priority.

[0113] Controlling the proportion of individuals that can be intervened by the agent in the space-time region and In this case, the regional multi-agent reinforcement learning control system determines the target individuals in the second individual who need to be quarantined at home and completely isolated according to the above intervention priorities.

[0114] It can be understood that if, based on the proportion of intervened individuals, it is determined that 100 people can be quarantined at home and 50 people can be completely isolated, since people with higher intervention priorities are given higher priority for strict isolation, if the second individual includes 500 people, the top 50 people ranked in the second individual can be used as target individuals and completely isolated, and the top 100 people ranked from 50 to 150 in the second individual can be used as target individuals and completely isolated.

[0115] In this embodiment, the individual infection risk dynamic assessment model can be used to assess the infection risk of each individual in real time, adjust the intervention object in time, ensure the flexibility and timeliness of epidemic prevention and control, and use two time lookback windows to calculate the infection risk of high-risk places and individual infection risk respectively, accurately identify high-risk individuals, and provide data support for subsequent spatiotemporal regional and individual interventions, so as to give priority to high-risk individuals for intervention; combined with the individual's spatiotemporal behavior and social contact risk, provide accurate individual intervention measures, reduce unnecessary intervention costs, and balance epidemic control with social and economic sustainability; based on the individual's spatiotemporal behavior and social contact network, the individual's infection risk can be accurately calculated,

[0116] Furthermore, in order to verify the effectiveness of the above method, by collecting micro-family structure survey data and macro-demographic data in Shenzhen, a community-level synthetic population containing 1.7 million agents and their spatial unit attributes was generated at a sampling rate of 10%. Based on the real geographic, population, social network and epidemic transmission data of Shenzhen, the epidemic transmission risks in different time and space regions were simulated, and a discrete-time susceptible-latent-infectio us-removed (SLIR) model was set up according to COVID-19 to simulate the changes in individual health status; the experimental results show that the present application can effectively reduce the number of infections in different epidemic scenarios. Compared with traditional epidemic control strategies (such as unified isolation policies or global reinforcement learning strategies), the multi-agent reinforcement learning system of the present application effectively suppresses the spread of the epidemic through refined individual mobility control, successfully reduces unnecessary isolation and restricted movement, and enables social and economic activities to operate relatively normally during the epidemic prevention and control period. Compared with excessive isolation measures, the present application balances epidemic control and economic costs, and can effectively control the epidemic while reducing intervention costs. Refer to Figure 7 .

[0117] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the individual mobility intervention method of the present application. More simple transformations based on this technical concept are all within the protection scope of the present application.

[0118] This application also provides an individual mobile intervention device, please refer to Figure 7 , the individual mobile intervention device comprises:

[0119] An acquisition module 10 is used to acquire epidemic-related data in a plurality of spatiotemporal regions obtained by dividing the city to be controlled, each of the spatiotemporal regions corresponding to an intelligent agent;

[0120] A calculation module 20, for calculating the epidemic-related data of the corresponding time and space region through the reinforcement learning algorithm corresponding to each of the intelligent agents, and obtaining the target intervention measures of the corresponding time and space region;

[0121] The intervention module 30 is used to perform movement intervention on individuals in the corresponding space-time area based on the target intervention measures.

[0122] The individual mobile intervention device provided by the present application adopts the individual mobile intervention method in the above embodiment, which can solve the technical problem of poor real-time performance of individual mobile intervention. Compared with the prior art, the beneficial effects of the individual mobile intervention device provided by the present application are the same as the beneficial effects of the individual mobile intervention method provided by the above embodiment, and other technical features in the individual mobile intervention device are the same as the features disclosed in the above embodiment method, which will not be repeated here.

[0123] The present application provides an individual mobile intervention device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the individual mobile intervention method in the above-mentioned embodiment one.

[0124] Reference below Figure 8 , which shows a schematic diagram of the structure of an individual mobile intervention device suitable for implementing the embodiment of the present application. The individual mobile intervention device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, tablet computers, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 8 The individual mobile intervention device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0125] like Figure 8As shown, the individual mobile intervention device may include a processing device 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. Various programs and data required for the operation of the individual mobile intervention device are also stored in the RAM 1004. The processing device 1001, the ROM 1002, and the RAM 1004 are connected to each other via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following regional multi-agent reinforcement learning control systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices 1003 including, for example, a tape, a hard disk, etc.; and communication devices 1009. The communication device 1009 can allow the individual mobile intervention device to communicate with other devices wirelessly or by wire to exchange data. Although the figure shows individual mobile intervention devices with various regional multi-agent reinforcement learning control systems, it should be understood that it is not required to implement or have all the regional multi-agent reinforcement learning control systems shown. More or fewer regional multi-agent reinforcement learning control systems may be implemented or have instead.

[0126] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0127] The individual mobile intervention device provided by the present application adopts the individual mobile intervention method in the above embodiment, which can solve the technical problem of poor real-time performance of individual mobile intervention. Compared with the prior art, the beneficial effects of the individual mobile intervention device provided by the present application are the same as the beneficial effects of the individual mobile intervention method provided by the above embodiment, and other technical features in the individual mobile intervention device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0128] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0129] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0130] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, and the computer-readable program instructions are used to execute the individual movement intervention method in the above-mentioned embodiment.

[0131] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor regional multi-agent reinforcement learning control system, a regional multi-agent reinforcement learning control system or a device, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM: Random Access Memory), a read-only memory (ROM: Read Only Memory), an erasable programmable read-only memory (EPROM: Erasable Programmable Read Only Memory or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM: CD-Read Only Memory), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction-executing regional multi-agent reinforcement learning control system, a regional multi-agent reinforcement learning control system or a device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency: Radio Frequency), etc., or any suitable combination thereof.

[0132] The computer-readable storage medium may be included in the individual mobile intervention device; or may exist independently without being assembled into the individual mobile intervention device.

[0133] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by the individual mobile intervention device, the individual mobile intervention device executes the individual mobile intervention method.

[0134] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0135] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the regional multi-agent reinforcement learning control system, method and computer program product according to each embodiment of the present application. In this regard, each box in the flowchart or block diagram may represent a module, a program segment, or a portion of a code, which contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box may also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart, can be implemented with a dedicated hardware-based regional multi-agent reinforcement learning control system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0136] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0137] The readable storage medium provided by the present application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned individual mobile intervention method, and can solve the technical problem of poor real-time performance of individual mobile intervention. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided by the present application are the same as the beneficial effects of the individual mobile intervention method provided by the above-mentioned embodiment, and will not be repeated here.

[0138] The present application also provides a computer program product, comprising a computer program, which implements the steps of the individual movement intervention method as described above when the computer program is executed by a processor.

[0139] The computer program product provided by this application can solve the technical problem of poor real-time performance of individual mobile intervention. Compared with the prior art, the beneficial effects of the computer program product provided by this application are the same as the beneficial effects of the individual mobile intervention method provided by the above embodiment, which will not be repeated here.

[0140] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. An individual mobility intervention method, characterized in that: The method includes: Obtaining epidemic-related data in multiple spatiotemporal regions obtained by dividing the city to be controlled, each of the spatiotemporal regions corresponds to an intelligent agent; By using the reinforcement learning algorithms corresponding to the intelligent agents, the epidemic-related data of the corresponding time and space regions are calculated to obtain target intervention measures for the corresponding time and space regions; Based on the target intervention measures, mobile intervention is performed on individuals in the corresponding space-time area.

2. The method according to claim 1, characterized in that The step of calculating the epidemic-related data of the corresponding time and space region through the reinforcement learning algorithm corresponding to each of the intelligent agents to obtain the target intervention measures of the corresponding time and space region includes: Calculating global epidemic-related data based on the epidemic-related data in the multiple spatiotemporal regions; Through the reinforcement learning algorithms corresponding to the intelligent agents, the epidemic-related data of the corresponding time and space regions and the global epidemic-related data are calculated to obtain the target intervention measures of the corresponding time and space regions.

3. The method according to claim 1, characterized in that After the step of performing movement intervention on individuals in the corresponding spatiotemporal area based on the target intervention measure, the method further comprises: Obtain the number of infected individuals in each spatiotemporal area after mobile intervention; The reward value is calculated by a preset reward function based on the number of infected individuals corresponding to each of the spatiotemporal regions, the preset intervention cost, the preset balance weight between health and socio-economy, and the preset tendency weight, wherein the preset tendency weight is used to characterize the tendency to take intervention measures or the tendency to control costs; Based on the reward value, the reinforcement learning algorithm is iteratively trained until the goal of minimizing the number of infected people and the intervention cost is reached to update the agent.

4. The method according to claim 3, characterized in that The preset intervention cost is the sum of the costs of the target intervention measures taken in the corresponding time and space region; the preset balance weight is determined based on the bearing capacity of the economic system and the bearing capacity of the medical system in the corresponding time and space region.

5. The method according to claim 1, characterized in that The step of performing movement intervention on individuals in the corresponding spatiotemporal area based on the target intervention measure includes: Determining the proportion of individuals who can be intervened in each of the time-space regions based on the target intervention measures corresponding to each of the time-space regions and the preset intervention number limits of different intervention measures; Based on the proportion of interveneable individuals in each of the space-time regions, target individuals in each of the space-time regions are determined, and mobile intervention is performed on the target individuals in the corresponding space-time regions.

6. The method according to claim 5, characterized in that The step of determining the target individuals in each of the spatiotemporal regions based on the proportion of interveneable individuals in each of the spatiotemporal regions comprises: Determine high-risk places within the first time period based on places visited by confirmed cases within the first time period in the epidemic-related data of each of the spatiotemporal regions; determining, based on the epidemic-related data, an infection risk of a first individual who visited the high-risk location within a second time period; Determining a social contact risk of the first individual based on the social attribute of the first individual; Based on the infection risk, social contact risk and the proportion of intervened individuals, a target individual is determined from the first individual corresponding to each of the spatiotemporal areas.

7. The method according to claim 6, characterized in that The step of determining the target individual from the first individual corresponding to each of the spatiotemporal regions based on the infection risk, the social contact risk, and the proportion of interveneable individuals comprises: The first individuals are sorted from high to low according to the infection risk, and when there are individuals with zero infection risk and / or individuals with the same infection risk among the first individuals, the individuals with zero infection risk and / or individuals with the same infection risk are sorted from high to low according to the social contact risk to obtain a sorted second individual, wherein the higher the ranking, the higher the intervention priority; Based on the proportion of interveneable individuals and the intervention priority, a target individual is determined from the second individuals.

8. An individual mobile intervention device, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the individual movement intervention method according to any one of claims 1 to 7.

9. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the individual movement intervention method according to any one of claims 1 to 7 are implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program, which, when executed by a processor, implements the steps of the individual movement intervention method according to any one of claims 1 to 7.

Images