Method and device for adjusting intervention actions for infectious diseases, storage medium, and electronic device

Abstract

The present disclosure is about a method and device, storage medium, and electronic device for adjusting intervention actions for infectious diseases, and relates to the field of big data processing technology. The method includes: obtaining social demographic data of a target area; wherein the social demographic data includes one or more of age data, family data, employment data, workplace data, school data, and commuting data; constructing an individual simulation model based on the social demographic data, and using the individual simulation model to calculate the regional state data after the current intervention action is performed in the target area; wherein the regional state data includes the number of newly confirmed cases per day, the cumulative number of confirmed cases, the number of vaccine injections, and multiple historical intervention actions; inputting the regional state data into a trained intervention action prediction model to obtain a predicted intervention level, and adjusting the current intervention action of the target area based on the predicted intervention level. The present disclosure improves the accuracy of the predicted intervention level.

Description

Background Art

[0002] In the existing intervention action adjustment method for infectious diseases, it is impossible to predict the intervention level in combination with the regional status data of a certain area, and thus it is impossible to adjust the current intervention action in time according to the predicted intervention level.

[0003] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure, and therefore may include information that does not constitute the prior art known to ordinary technicians in the field. Summary of the invention

[0004] The purpose of the present disclosure is to provide a method for adjusting intervention actions for infectious diseases, an apparatus for adjusting intervention actions for infectious diseases, a computer-readable storage medium, and an electronic device, thereby at least to a certain extent overcoming the problem of being unable to predict the intervention level due to the limitations and defects of related technologies.

[0005] According to one aspect of the present disclosure, a method for adjusting an intervention action for an infectious disease is provided, comprising:

[0006] Acquiring sociodemographic data of the target area; wherein the sociodemographic data includes one or more of age data, family data, employment data, workplace data, school data, and commuting data;

[0007] An individual simulation model is constructed according to the sociodemographic data, and the individual simulation model is used to calculate the regional status data after the current intervention action is performed in the target area; wherein the regional status data includes the number of newly confirmed cases per day, the cumulative number of confirmed cases, the number of vaccine injections, and multiple historical intervention actions;

[0008] The regional status data is input into the trained intervention action prediction model to obtain a predicted intervention level, and the current intervention action of the target area is adjusted based on the predicted intervention level; wherein the trained intervention action prediction model is obtained by training the intervention action prediction model to be trained using a reinforcement learning algorithm.

[0009] In an exemplary embodiment of the present disclosure, adjusting the current intervention action of the target area based on the predicted intervention level includes:

[0010] Lower the current intervention action of the target area by one level; or

[0011] Maintain the current intervention action of the target area; or

[0012] The current intervention action of the target area is adjusted up by one level;

[0013] Among them, the intervention action is used to indicate whether to perform at least one of the following action items: case isolation, wearing masks, working from home, closing schools and community lockdown.

[0014] In an exemplary embodiment of the present disclosure, a reinforcement learning algorithm is used to train the intervention action prediction model to be trained to obtain a trained intervention action prediction model, including:

[0015] constructing a target reward function according to the region state data, and inputting the region state data into the intervention action prediction model to be trained to obtain a first output result;

[0016] A target loss function is constructed according to the target reward function and the first output result, and the parameters of the intervention action prediction model to be trained are adjusted based on the target loss function to obtain the trained intervention action prediction model.

[0017] In an exemplary embodiment of the present disclosure, in the process of training the intervention action prediction model to be trained, the training data samples used are obtained by iteratively repeating the following method:

[0018] Inputting the regional status data at the current time node into the intervention action prediction model to be trained to obtain the current intervention level;

[0019] Adjusting the current intervention action based on the current intervention level to obtain the intervention action at the next time node;

[0020] The individual simulation model is used to calculate the regional state data of the next time node after the intervention action of the next time node is performed in the target area, so as to input the regional state data of the next time node into the intervention action prediction model.

[0021] In an exemplary embodiment of the present disclosure, constructing a target reward function according to the area state data includes:

[0022] Constructing an epidemic severity function of the infectious disease according to the number of newly confirmed cases per day and the maximum number of confirmed cases that the target area can accommodate;

[0023] Constructing an action strictness function according to the current intervention action at the current time node and the historical intervention action at the previous time node corresponding to the current time node;

[0024] The target reward function is constructed according to the epidemic severity function and the action strictness function.

[0025] In an exemplary embodiment of the present disclosure, an action strictness function is constructed according to a current intervention action at a current time node and a historical intervention action at a previous time node corresponding to the current time node, including:

[0026] Determine the action maximum value from the current intervention action at the current time node and the historical intervention action at the previous time node;

[0027] Calculating a first ratio between the current intervention action and the action maximum, and calculating a first difference between the current intervention action and the historical intervention action;

[0028] The action strictness function is obtained by performing a weighted summation on the inverse of the first ratio and the inverse of the first difference.

[0029] In an exemplary embodiment of the present disclosure, constructing an individual simulation model according to the sociodemographic data includes:

[0030] Divide the sociodemographic data into regions according to geographical administrative regions to obtain multiple communities, and construct multiple network layers of different dimensions based on the communities and preset social scenarios;

[0031] The infection probability of each network layer of different dimensions is calculated, and the individual simulation model is constructed according to the infection probability of each network layer and the epidemic dynamics propagation model.

[0032] According to one aspect of the present disclosure, there is provided an intervention action adjustment device for infectious diseases, comprising:

[0033] A data acquisition module, used to acquire social demographic data of a target area; wherein the social demographic data includes one or more of age data, family data, employment data, workplace data, school data, and commuting data;

[0034] A regional status data calculation module, used to construct an individual simulation model based on the sociodemographic data, and use the individual simulation model to calculate the regional status data after the current intervention action is performed in the target area; wherein the regional status data includes the number of newly confirmed cases per day, the cumulative number of confirmed cases, the number of vaccine injections, and multiple historical intervention actions;

[0035] An intervention action adjustment module is used to input the area status data into a trained intervention action prediction model to obtain a predicted intervention level, and adjust the current intervention action of the target area based on the predicted intervention level; wherein the trained intervention action prediction model is obtained by training the intervention action prediction model to be trained using a reinforcement learning algorithm.

[0036] According to one aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the method for adjusting an intervention action for an infectious disease described in any one of the above is implemented.

[0037] According to one aspect of the present disclosure, there is provided an electronic device, including:

[0038] Processor; and

[0039] A memory, configured to store executable instructions of the processor;

[0040] Wherein, the processor is configured to execute any one of the above-mentioned methods for adjusting intervention actions for infectious diseases by executing the executable instructions.

[0041] The present disclosure discloses a method for adjusting intervention actions for infectious diseases. On the one hand, an individual simulation model can be constructed according to the socio-demographic data of the target area, and the regional status data of the target area can be calculated using the individual simulation model; the regional status data is then input into the trained intervention action prediction model to obtain a predicted intervention level, and the current intervention action of the target area is adjusted based on the predicted intervention level, thereby solving the problem in the prior art that the intervention level cannot be predicted in combination with the regional status data of a certain area, and thus the current intervention action cannot be adjusted in time according to the predicted intervention level; on the other hand, in the process of predicting the intervention level, the number of newly confirmed cases per day, the cumulative number of confirmed cases, the number of vaccine injections, and the historical intervention actions are fully considered, thereby improving the accuracy of the predicted intervention level; on the other hand, since the current intervention action of the target area can be adjusted in time based on the predicted intervention level, the problem of aggravation of the infectious disease epidemic caused by the current intervention actions being too loose is avoided.

[0042] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The accompanying drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the specification are used to explain the principles of the present disclosure. Obviously, the accompanying drawings described below are only some embodiments of the present disclosure, and for ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without creative work.

[0044] Figure 1 The flowchart schematically shows a method for adjusting an intervention action for an infectious disease according to an exemplary embodiment of the present disclosure.

[0045] Figure 2An exemplary diagram schematically illustrates a composite structure of a multi-layer contact network of an individual simulation model according to an exemplary embodiment of the present disclosure.

[0046] Figure 3 An example diagram of a calculation scenario of regional status data according to an example embodiment of the present disclosure is schematically shown.

[0047] Figure 4 A flowchart of a method for inputting regional status data into a trained intervention action prediction model to obtain a predicted intervention level according to an exemplary embodiment of the present disclosure is schematically shown.

[0048] Figure 5 A structural example diagram of a Transformer model according to an example embodiment of the present disclosure is schematically shown.

[0049] Figure 6 A flowchart of a method for training an intervention action prediction model to be trained according to an example embodiment of the present disclosure is schematically shown.

[0050] Figure 7 A flowchart of a method for constructing an epidemic severity function of an infectious disease according to the daily number of newly confirmed cases and the maximum number of confirmed cases that the target area can accommodate according to an example embodiment of the present disclosure is schematically shown.

[0051] Figure 8 The flowchart schematically shows another method for adjusting intervention actions for infectious diseases according to an exemplary embodiment of the present disclosure.

[0052] Fig. 9 A block diagram of a device for adjusting an intervention action for an infectious disease according to an exemplary embodiment of the present disclosure is schematically shown.

[0053] Fig.10 An electronic device for implementing the above-mentioned infectious disease intervention action adjustment method according to an exemplary embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0054] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as being limited to the examples set forth herein; on the contrary, these embodiments are provided so that the present disclosure will be more comprehensive and complete, and the concepts of the example embodiments are fully conveyed to those skilled in the art. The described features, structures, or characteristics may be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present disclosure. However, those skilled in the art will appreciate that the technical solutions of the present disclosure may be practiced while omitting one or more of the specific details, or other methods, components, devices, steps, etc. may be adopted. In other cases, known technical solutions are not shown or described in detail to avoid obscuring various aspects of the present disclosure.

[0055] In addition, the accompanying drawings are only schematic illustrations of the present disclosure and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor devices and / or microcontroller devices.

[0056] Simulation and modeling refer to the process of digitizing physical laws abstracted from real-world objects to create a system model that can describe or simulate the structure or behavior of the object. This model is then used for virtual testing and early verification of the real world, thereby reducing the cost of direct testing.

[0057] By combining scientific research literature and actual data, simulation and modeling technology plays an increasingly important role in the field of epidemiology: for example, by modeling a certain infectious disease, its propagation law can be analyzed through simulation; by simulating epidemic prevention measures, the possible effects of various measures after implementation can be analyzed. Further use of actual epidemic data for modeling and correction can also predict and judge the trend of the epidemic in advance, and maintain normal life while intervening in the epidemic. Therefore, formulating a reasonable intervention strategy is a necessary choice.

[0058] Existing intervention strategies can be formulated in the following ways:

[0059] One is to manually formulate and adjust intervention actions according to the trend of the epidemic; however, this method has a large number of subjective factors, and the adjustment of intervention actions is not timely, which cannot achieve the optimal effect of controlling the epidemic;

[0060] Another approach is to use the infectious disease dynamics approach to model infectious diseases using a set of differential equations, and then predict intervention strategies. However, because the infectious disease dynamics model is too macroscopic, the method of replacing the sum of individual effects with average effects simplifies the real scene. In the real scene with coarse-grained characterization, it is difficult to closely simulate and analyze the implementation of specific intervention actions, and it cannot reflect the heterogeneity of the population, nor can it implement epidemic prevention measures for specific groups.

[0061] Another method is to use Agent-Based Modeling (ABM) to represent each individual in the simulation environment using a multi-agent approach, thereby simulating the spread of infectious diseases at a micro level. However, although this method can achieve micro-level modeling, its modeling method limits the scale of the simulation object and the implementation of epidemic prevention measures; and it does not implement epidemic prevention measures such as isolating confirmed patients and their close contacts and recommending the wearing of masks, which is not conducive to a comprehensive analysis of the effects of various epidemic prevention measures.

[0062] Based on this, this exemplary embodiment first provides an intervention action adjustment method for infectious diseases, which can be run on a server, server cluster or cloud server, etc. Of course, those skilled in the art can also run the method disclosed in this disclosure on other platforms as required, and this exemplary embodiment does not specifically limit this. Figure 1 As shown, the intervention action adjustment method for infectious diseases may include the following steps:

[0063] Step S110. Obtaining social demographic data of the target area; wherein the social demographic data includes one or more of age data, family data, employment data, workplace data, school data, and commuting data;

[0064] Step S120. Construct an individual simulation model based on the sociodemographic data, and use the individual simulation model to calculate the regional status data after the current intervention action is performed in the target area; wherein the regional status data includes the number of newly confirmed cases per day, the cumulative number of confirmed cases, the number of vaccine injections, and multiple historical intervention actions;

[0065] Step S130. Input the regional status data into the trained intervention action prediction model to obtain the predicted intervention level, and adjust the current intervention action of the target area based on the predicted intervention level; wherein the trained intervention action prediction model is obtained by training the intervention action prediction model to be trained using a reinforcement learning algorithm.

[0066] In the above-mentioned infectious disease intervention action adjustment method, on the one hand, since an individual simulation model can be constructed according to the socio-demographic data of the target area, and the regional status data of the target area can be calculated using the individual simulation model; then the regional status data is input into the trained intervention action prediction model to obtain the predicted intervention level, and the current intervention action of the target area is adjusted based on the predicted intervention level, which solves the problem in the prior art that the intervention level cannot be predicted in combination with the regional status data of a certain area, and thus the current intervention action cannot be adjusted in time according to the predicted intervention level; on the other hand, since in the process of predicting the intervention level, the number of newly confirmed cases per day, the cumulative number of confirmed cases, the number of vaccine injections and historical intervention actions are fully considered, the accuracy of the predicted intervention level is improved; on the other hand, since the current intervention action of the target area can be adjusted in time based on the predicted intervention level, the problem of aggravation of the infectious disease epidemic caused by the current intervention actions being too loose is avoided.

[0067] Hereinafter, the intervention action adjustment method for infectious diseases according to the exemplary embodiment of the present disclosure will be explained and illustrated in detail with reference to the accompanying drawings.

[0068] First, the purpose of the invention of the example embodiment of the present disclosure is explained and illustrated. Specifically, the example embodiment of the present disclosure proposes a new modeling and simulation method for infectious diseases and their intervention actions, and uses real-world epidemic data for correction, ultimately achieving the purpose of making decision-making recommendations and effect simulation for epidemic intervention, which can be used for micro-level modeling of infectious disease transmission in larger groups; and modeling of epidemic prevention measures such as isolation of confirmed patients and their close contacts and recommendations for wearing masks; and then using real-world epidemic data to correct the model and give decision-making recommendations, so as to achieve the purpose of combining the micro-model with deep reinforcement learning technology to automatically learn the optimal combination of intervention actions; further, by using multi-agent-based simulation, it is also possible to analyze the epidemic prevention and control effects of various intervention actions from a micro level.

[0069] Secondly, in an intervention action adjustment method for an infectious disease according to an exemplary embodiment of the present disclosure:

[0070] In step S110, the socio-demographic data of the target area is obtained; wherein the socio-demographic data includes one or more of age data, family data, employment data, workplace data, school data and commuting data.

[0071] Specifically, the socio-demographic data of a target area can be obtained from the human resources database based on the smart city. The socio-demographic data can be, for example, basic socio-demographic indicators, which can include age data, family data, employment data, workplace data, school data, and commuting data of the area, etc.; among which, age data can be, for example, age structure (infants, children, teenagers, young people, middle-aged people, and elderly people), family data can be, for example, family size (the number of people in a family), employment data can be, for example, employment rate (the proportion of employed people in the total population), workplace data can be, for example, work group distribution (distribution of the number of people in each industry), school data can be, for example, school size distribution (the number of people in the school), commuting data can be, for example, specific commuting duration and commuting time, etc.

[0072] In step S120, an individual simulation model is constructed based on the socio-demographic data, and the individual simulation model is used to calculate the regional status data after the current intervention action is performed in the target area; wherein the regional status data includes the number of newly confirmed cases per day, the cumulative number of confirmed cases, the number of people vaccinated, and multiple historical intervention actions.

[0073] In this example embodiment, first, an individual simulation model is constructed based on the social demographic data. Specifically, it may include: first, dividing the social demographic data into regions according to geographical administrative regions to obtain multiple communities, and constructing multiple network layers of different dimensions based on each of the communities and preset social scenarios; second, calculating the infection probability of each network layer of different dimensions, and constructing the individual simulation model based on the infection probability of each network layer and the epidemic dynamics transmission model.

[0074] Specifically, the individual simulation model can be constructed according to the basic socio-demographic indicators (socio-demographic data) as follows: In the first step, the population is divided into D = 1, ..., k groups according to geographical administrative regions, and each region is divided into several communities C d , where C d =1, 2, …; then, according to the latest census statistics, each community is divided into households of different sizes in a certain proportion; among them, the family is the smallest and closest social contact unit in this method, and the distance between individuals is divided into levels. The contact rate decreases with the increase of the social contact unit. The specific order from small to large is family, family cluster, community, administrative district and city.

[0075] The second step is to divide people’s daily social patterns into two types in the individual simulation model: daytime and nighttime. Among them, the contact between people during the daytime occurs in the following three environments: community, school, and workplace. Figure 2 As shown, the individual simulation model adopts a composite structure of multi-layer contact networks to simulate different types of social activities in real scenarios. Each layer simulates a contact scenario, among which the three social activity networks of school, workplace and community are relatively independent and calculated synchronously.

[0076] Further, first, for the school network layer, most children go to school or participate in playgroups. Preschool children usually form playgroups or community kindergartens. Moreover, each community has mixed groups, representing two primary schools, one junior high school and one high school. The school network layer consists of n s The node number of each school is allocated according to the actual school size distribution and can include preschool education, elementary school, junior high school, and high school.

[0077] Secondly, for the workplace network layer, workplaces are assigned to the working-age workforce according to employment rates, and work destinations are assigned to each employed person according to actual commuting data; workers are assigned to communities within their destination areas to simulate community contacts during the day, and a working group of about 20 people represents their close contacts at the workplace; unemployed people stay in their home communities, and no workplace contacts occur during the day, except for contact with family members who are not working or going to school; and at night social contacts only occur in the community contact network, where everyone can contact their family, family cluster, family community, and other people in the family community. Similar to the family layer, the workplace layer consists of n w The network is composed of disconnected components, each component represents a workplace, and the number of nodes in each workplace follows the actual workplace size distribution. In addition, the individuals in each workplace are further subdivided into several work groups as the smallest contact unit in the workplace layer.

[0078] Then, for the community activity network layer, the nodes in the macro graph represent communities, and each node in the micro graph represents a person; the number of nodes for each family is determined by the actual family size distribution; and the association between age groups and each node can be set according to the actual age distribution within the family. Furthermore, referring to the distribution of education rate and employment rate by age, each node type determines whether it also participates in school or work layer activities.

[0079] Furthermore, the epidemic spread process can be represented by a discrete-time susceptible-exposed-infected-diagnosed-removed (SEIAR) model with a time step t (time unit is half a day). And, define S i,τ 、E i,τ ,I i,τ , A i,τ and R i,τis a state of the Bernoulli random variable individual i at time t, S i,τ +E i,τ +I i,τ +A i,τ +R i,τ =1;

[0080] Since the SEIAR disease dynamics can be described as a discrete Markov process in which the transition probability changes over time, and the susceptible (S) individual i can be infected by the infected (I) individual j and enter the exposed (E) state, its probability λ can be expressed by the following formula (1):

[0081]

[0082] Among them, θ l is the contact weight of layer l, l∈L=c,s,w, L is the set of layers, c, s, w represent the community, school and workplace respectively. θ l The value of also depends on the location of contact, the age of the individual, and the intervention actions performed at time t. For example, if the basic infection occurs at the community level, θ c It can be expressed by the following formula (2):

[0083]

[0084] Among them, M i ∈[0,1] is the protection rate of an individual wearing a mask. The smaller the protection rate, the more effective it is. as well as is an indicator variable. The former indicates whether i has contact with j at the community level, and the latter indicates whether the intervention action of suspension of work and school has an impact on individual i. as well as is a Bernoulli random variable, means that i is in contact with j at the community level, and vice versa This shows that the intervention action of suspension of work and school acts on individual i, and vice versa CR represents the community contact reduction rate, γ i represents the age group to which individual i belongs, cpcm[γ i ] represents the probability that individual i contacts an infected individual in the community. Among them, the contact probability is related to age. When individual j in layer l infects individual i at time t, all the stored information will help calculate the reproduction number and the final epidemic size. ω = 2 represents the causal multiplier, which means that when person i leaves school (school is closed) or works from home, the risk of infection at the community level increases.

[0085] Similarly, if contact infection occurs at the working layer, θ w It can be expressed by the following formula (3):

[0086]

[0087] Among them, cpw[γ i ] represents the probability that individual i comes into contact with infected person j in the working layer.

[0088] If the infection occurs at the school level, θ s It can be expressed by the following formula (4):

[0089]

[0090] Among them, cps[γ i ] represents the probability that individual i comes into contact with infected person j at the school level.

[0091] pri and prs are two individual attributes. pri represents the probability of an individual infecting others when infected, and prs represents the probability of the individual being infected by other infected persons. prs can be expressed by the following formula (5):

[0092] i.prs=(1-i.BVE s )*(1-VE s *EFF[γ i ]); Formula (5)

[0093] Among them, BVE s represents the baseline vaccine efficiency of susceptible individuals, VE s represents the vaccine efficiency of susceptible individuals, EFF[γ i ] represents the vaccine efficiency related to the age group; at the same time, for individual i, the probability of S→E can be expressed as: λ i =∑ j∈E λ i,j,τ .

[0094] So far, we can know that taking into account vaccines and masks, the new epidemic transmission dynamics can be expressed by the following formula:

[0095]

[0096]

[0097] Where μ is the probability of E→I, T1 and T2 are the time delays of I→A and A→R respectively.

[0098] Furthermore, the vaccine background setting can be carried out in the following ways: the algorithm adds considerations for vaccination, and pre-measures and reactive measures can be set according to the vaccination situation. The former refers to vaccination before the outbreak, and the latter refers to vaccination during the outbreak. When neither vaccination method is adopted, the model returns to the original situation without vaccination. At the same time, for the characteristics of the vaccine, the relevant parameters set in the model include vaccine number (when multiple vaccines exist at the same time and need to be distinguished, it is realized by numbering), the effectiveness of the vaccine for different groups, the proportion of vaccines in different types of people or whether they can be vaccinated (such as the elderly over 65 years old, minors under 18 years old, pregnant women, etc.), the number of injections required for the vaccine, and the vaccination cycle. In addition, for vaccine recipients, the model also sets corresponding parameters, including the proportion of vaccination in the appropriate age group, the priority of vaccination for different types of personnel (such as high-risk industries need to be vaccinated first, and people with congenital diseases or pregnant women are not vaccinated), and the effectiveness of the vaccine in different age groups (the default effectiveness is consistent); the relevant parameters can be completed as model input during the simulation initialization process, and some parameters can be changed according to conditions during the simulation process.

[0099] At this point, an individual simulation model can be obtained based on the above-mentioned new epidemic transmission dynamics model and the above-mentioned vaccine background settings; then the age structure, family size, employment rate, work group distribution, school size distribution and commuting data of the target area are input into the individual simulation model, and the daily number of newly confirmed cases, cumulative number of confirmed cases, number of vaccine injections and historical intervention actions in the target area can be obtained. For specific calculation scenario example diagrams, please refer to Figure 3 shown.

[0100] In step S130, the area status data is input into the trained intervention action prediction model to obtain a predicted intervention level, and the current intervention action of the target area is adjusted based on the predicted intervention level; wherein the trained intervention action prediction model is obtained by training the intervention action prediction model to be trained using a reinforcement learning algorithm.

[0101] In this example embodiment, first, the regional state data is input into the trained intervention action prediction model to obtain the predicted intervention level. Figure 4 As shown, the following steps may be included:

[0102] Step S410, encode the regional state data based on a preset Transformer model to obtain a first encoding vector; wherein the dimension of the first encoding vector is consistent with the dimension of the first input layer of the trained intervention action prediction model, and the preset Transformer model includes a first linear fusion layer, an approximate sparse self-attention layer, a splicing layer and a second linear fusion layer.

[0103] In this example embodiment, first, the structure of the Transformer model is explained and illustrated. Figure 5 As shown, the Transformer model may include an input layer 501, a first linear fusion layer (Linear) 502, an approximate sparse self-attention layer (ProbSparse Self-Attention) 503, a splicing layer (Contact) 504, a second linear fusion layer (Linear) 505, and an output layer 506; wherein the input layer, the first linear fusion layer, the approximate sparse self-attention layer, the splicing layer, the second linear fusion layer, and the output layer are connected in sequence.

[0104] Secondly, encoding the regional status data based on the preset Transformer model to obtain a first encoding vector, which may specifically include: first, using the first linear fusion layer to linearly fuse the daily number of newly confirmed cases, the cumulative number of confirmed cases, the number of vaccine injections, and the historical intervention actions to obtain the current input vector; secondly, using the approximate sparse self-attention layer to calculate the current input vector to obtain the first importance, second importance, third importance, and fourth importance of the daily number of newly confirmed cases, the cumulative number of confirmed cases, the number of vaccine injections, and the historical intervention actions in the current input vector; then, using the connection layer to connect the daily number of newly confirmed cases, the cumulative number of confirmed cases, the number of vaccine injections, and the historical intervention actions as well as the first importance, second importance, third importance, and fourth importance to obtain a first sub-vector, a second sub-vector, a third sub-vector, and a fourth sub-vector; finally, using the second linear fusion layer to linearly fuse the first sub-vector, the second sub-vector, the third sub-vector, and the fourth sub-vector to obtain the first encoding vector.

[0105] It should be noted here that the number of newly confirmed cases and the cumulative number of confirmed cases are quite different from the number of people injected with the vaccine; therefore, in order to facilitate the extraction and calculation of features, an approximate sparse self-attention layer is used here, which can improve the accuracy of the first importance, the second importance, the third importance and the fourth importance, and ultimately achieve the purpose of improving the accuracy of the predicted intervention level; and, the dimension of the first coding vector is limited here to be consistent with the dimension of the first input layer of the trained intervention action prediction model, so that when the first coding vector is input into the trained intervention action prediction model, the model can accurately perform mapping processing according to the first coding vector, and then obtain the corresponding output result.

[0106] Step S420, using the trained intervention action prediction model to map the first coding vector to obtain an output result; wherein the output result includes three levels, the three levels are respectively keeping the current intervention action unchanged, increasing the current intervention action by one level, and decreasing the current intervention action by one level.

[0107] Specifically, the trained intervention action prediction model can be obtained by training the intervention action prediction model to be trained using a reinforcement learning algorithm. The intervention action prediction model to be trained can be, for example, a reinforcement learning model, which can receive the first encoding vector through a three-layer perceptron network with each hidden layer having 64 nodes, and obtain an output result after mapping the first encoding vector. The output result can include 3 nodes, which respectively correspond to the adjustment methods of the intervention action, specifically keeping the current intervention action level unchanged, increasing it by one level, and decreasing it by one level.

[0108] Step S430: taking the level with the highest probability in the output result as the predicted intervention level.

[0109] For example, the probabilities of keeping the current intervention action level unchanged, increasing one level, and decreasing one level are 0.8, 0.05, and 0.15 respectively, and it can be determined that the predicted intervention level is to keep the current intervention action level unchanged.

[0110] Furthermore, after obtaining the predicted intervention level, the current intervention action of the target area can be adjusted based on the predicted intervention level. The specific adjustment methods may include: lowering the current intervention action of the target area by one level; or keeping the current intervention action of the target area unchanged; or raising the current intervention action of the target area by one level; wherein the intervention action is used to indicate whether to perform at least one of the following action items: case isolation, wearing masks, working from home, closing schools and community lockdown.

[0111] That is to say, one or more of the above action items can be combined or used alone to obtain multiple intervention actions with different intervention levels. For example, the current intervention action can include the following four levels. The first level is: no case isolation, no need to wear a mask, no need to work from home, no need to close the school, and no need for community lockdown; the second level is: case isolation, must wear a mask, no need to work from home, no need to close the school, and no need for community lockdown; the third level is: case isolation, must wear a mask, must work from home, must close the school, and no community lockdown; the fourth level is: case isolation, must wear a mask, must work from home, must close the school, and must be community-locked. It should be noted here that other levels of intervention actions can also be constructed according to actual needs, and this example does not specifically limit this.

[0112] The following will explain and illustrate the settings and enabling of various intervention actions.

[0113] Work stoppage (home office): Work stoppage is achieved by setting a certain proportion of employees to work from home, that is, a certain proportion of practitioners do not participate in work-level contact activities during the day, but participate in activities in the community where they live. The start and end time of home office can be set.

[0114] Suspension of school (school closure): There are two suspension plans. The first is to suspend classes in schools in medium- and high-risk areas, while other schools will continue to operate normally. The second is to suspend classes in all schools in the area the day after a confirmed case appears. The start and duration of the suspension can be set.

[0115] Community closed management (community lockdown): By setting a certain proportion of communities to implement community closed management, the probability of contact between people in the community is reduced. The start and end time of community closed management can be set.

[0116] Isolation (case isolation): There are two types of isolation, one is home isolation and the other is centralized isolation. In the former, there is a virus transmission channel between family members, and family transmission may occur. In the latter, close contacts are isolated in a special isolation place and receive professional medical services, so that further family transmission will not occur. The model defaults to centralized isolation for close contacts (such as family members of the infected person, colleagues in the same office, etc.), and home isolation for secondary close contacts. The isolation of different types of close contacts can be set before the simulation starts, and the definition of close contacts and secondary close contacts can be changed to expand or reduce the number of people who need to be isolated. Similarly, the isolation duration is adjustable, and the default is 14 days.

[0117] Mask wearing (wearing a mask): The model is set to wear a mask, which is required in public places, that is, in places other than residences. After wearing a mask, according to the protection rate of the mask, the probability of virus transmission is also multiplied by a corresponding factor less than 1, indicating that the probability of virus transmission is reduced after wearing a mask. The relevant parameters include: mask wearing rate, mask protection rate, time to start wearing a mask, and time to relax mask wearing. For mask wearing, the default setting is that the simulation is effective at all times.

[0118] Based on the above-mentioned intervention action settings and enabled settings, the following four different levels of intervention actions can be set, as shown in the following Table 1:

[0119] Intervention Action Level CI WM WH SC CL 0 × × × × × 1 √ √ × × × 2 √ √ √ √ × 3 √ √ √ √ √

[0120] Among them, in the above Table 1, CI: Case Isolation, case isolation; WM: Wearing Masks, wearing masks; WH: Working from Home, working from home; SC: School Closures, school closures; CL: Community Lockdown, community lockdown.

[0121] The exemplary embodiment of the present disclosure also provides another method for adjusting intervention actions for infectious diseases. Figure 6 As shown, the infectious disease intervention action adjustment method may also include the following steps:

[0122] Step S610: construct a target reward function according to the region state data, and input the region state data into the intervention action prediction model to be trained to obtain a first output result.

[0123] In this example embodiment, first, a target reward function is constructed according to the area state data. Specifically, refer to Figure 7 As shown, the following steps may be included:

[0124] Step S710, constructing an epidemic severity function of the infectious disease according to the number of newly confirmed cases per day and the maximum number of confirmed cases that the target area can accommodate.

[0125] Step S720: construct an action strictness function according to the current intervention action at the current time node and the historical intervention action at the previous time node corresponding to the current time node.

[0126] Among them, constructing an action strictness function based on the current intervention action of the current time node and the historical intervention action of the previous time node corresponding to the current time node can include: first, determining the maximum action value from the current intervention action of the current time node and the historical intervention action of the previous time node; second, calculating a first ratio between the current intervention action and the maximum action value, and calculating a first difference between the current intervention action and the historical intervention action; finally, performing weighted summation on the inverse of the first ratio and the inverse of the first difference to obtain the action strictness function.

[0127] Step S730, constructing the target reward function according to the epidemic severity function and the action strictness function.

[0128] Below, step S710-step S730 will be explained and illustrated. Specifically, the state data (States) information can calculate Re (epidemic severity function) and Rh (action strictness function) to obtain Reward (target reward function). Among them, the specific calculation formulas of Re (epidemic severity function), Rh (action strictness function) and Reward (target reward function) can be shown in the following formulas (11), (12) and (13):

[0129]

[0130] R t =R h (t)+R e (t); Formula (13)

[0131] Among them, n A (t) represents the number of new cases on the tth day under the current intervention action A, C is the maximum case capacity that can be allowed under the current medical conditions of the target area, and is a fixed parameter value; a(t) is the current intervention action at the current time node, and its value can be [0-3]; a(t-1) is the historical intervention action at the previous time node, and its value can be [0-3]; maxA is the maximum value of the intervention action, and its value is 3; ε, η, and w are constant parameters.

[0132] Furthermore, after obtaining the target reward function, the regional state data can be input into the intervention action prediction model to be trained to obtain the first output result. It should be noted here that the intervention action prediction model to be trained can include a Transformer network model + a DQN network model. The Transformer network model can be used to encode the regional state data, and the DQN network model can be used to map the encoded regional state data to obtain the first output result (Q-Value); of course, the intervention action prediction model to be trained can also be other models that can be used for classification, such as a logistic regression model or a support vector machine and a hidden Markov model, etc. This example does not specifically limit this.

[0133] It should be further explained here that, in the process of training the intervention action prediction model to be trained, the training data samples used are obtained by repeating the following iterations: first, the regional status data of the current time node is input into the intervention action prediction model to be trained to obtain the current intervention level; second, the current intervention action is adjusted based on the current intervention level to obtain the intervention action at the next time node; finally, the individual simulation model is used to calculate the regional status data of the next time node after the intervention action at the next time node is performed in the target area, so as to input the regional status data of the next time node into the intervention action prediction model. That is to say, the intervention action prediction model to be trained continuously outputs the current intervention level, and then adjusts the current intervention action based on the current intervention level, so as to obtain the intervention action at the next time node; further, the system continuously feeds back the intervention action at the next time node to the individual simulation model, and the individual simulation model, under the action of the intervention action at the next time node, continuously outputs the regional state data of the next time node after the target area executes the intervention action at the next time node, and then encodes the regional state data of the next time node through the Transformer network model, and maps the encoded regional state data of the next time node based on the DQN network model, and then obtains the intervention level of the next time node, thereby completing the entire iterative training process.

[0134] Step S620, constructing a target loss function according to the target reward function and the first output result, and adjusting the parameters of the intervention action prediction model to be trained based on the target loss function to obtain the trained intervention action prediction model.

[0135] Specifically, the target reward function Reward combined with the Q-value calculated by DQN can be used to determine the action of the next day (i.e., the intervention action); the parameters in the deep reinforcement learning module are learned and updated in a cycle, and the optimal intervention action can be generated after the training converges.

[0136] The following, combined Figure 8 The intervention action adjustment method for infectious diseases according to the exemplary embodiment of the present disclosure is further explained and illustrated. Figure 8 As shown, the intervention action adjustment method for infectious diseases may include the following steps:

[0137] Step S801, constructing an underlying individual simulation model, which may include commuting conditions, location distribution, and population structure, etc.; the location distribution includes work units, schools, and communities; the population structure includes census results (age structure, family size, employment rate, high-risk job employment ratio, etc.) and population vector data (which can be obtained by fitting administrative district statistical data);

[0138] Step S802, setting and verifying infectious disease related parameters included in the individual simulation model; wherein the infectious disease related parameters may include R0 and Beta set based on the historical infection situation in the local area, the virus in two based on existing research, and the contact probability of different groups based on age distribution and the type and closeness of the group;

[0139] Step S803, adding vaccination background and intervention design and implementation to the individual simulation model;

[0140] Step S804, calculating the regional state data using the constructed individual simulation model, and encoding the regional state data;

[0141] Step S805 , mapping the encoded regional status data to the intervention action prediction model to obtain a predicted intervention level, and adjusting the current intervention action based on the predicted intervention level.

[0142] The infectious disease intervention action adjustment method recorded in the example embodiment of the present disclosure, on the one hand, fills the gap in infectious disease simulation by conducting detailed micro-level modeling and simulation of the disease transmission process and prevention and control measures, which is conducive to a better understanding of the transmission laws of infectious diseases; on the other hand, it uses deep reinforcement learning technology to automatically learn the optimal intervention action solution.

[0143] The present disclosure also provides an intervention action adjustment device for infectious diseases. Fig. 9As shown, the intervention action adjustment device for infectious diseases may include a data acquisition module 910, a regional state data calculation module 920, and an intervention action adjustment module 930. Among them:

[0144] The data acquisition module 910 may be used to acquire social demographic data of the target area; wherein the social demographic data includes one or more of age data, family data, employment data, workplace data, school data, and commuting data;

[0145] The regional status data calculation module 920 can be used to construct an individual simulation model based on the social demographic data, and use the individual simulation model to calculate the regional status data after the current intervention action is performed in the target area; wherein the regional status data includes the number of newly confirmed cases per day, the cumulative number of confirmed cases, the number of vaccine injections, and multiple historical intervention actions;

[0146] The intervention action adjustment module 930 can be used to input the area status data into the trained intervention action prediction model to obtain the predicted intervention level, and adjust the current intervention action of the target area based on the predicted intervention level; wherein the trained intervention action prediction model is obtained by training the intervention action prediction model to be trained using a reinforcement learning algorithm.

[0147] In an exemplary embodiment of the present disclosure, the regional status data is input into an intervention level prediction model to obtain a predicted intervention level, including:

[0148] Encoding the regional status data based on a preset Transformer model to obtain a first encoding vector; wherein the dimension of the first encoding vector is consistent with the dimension of the first input layer of the intervention level prediction model;

[0149] Mapping the first coding vector using the intervention level prediction model to obtain an output result; wherein the output result includes three levels, the three levels are respectively keeping the current intervention action unchanged, increasing the current intervention action by one level, and decreasing the current intervention action by one level;

[0150] The level with the highest probability in the output results is used as the predicted intervention level.

[0151] In an exemplary embodiment of the present disclosure, the preset Transformer model includes a first linear fusion layer, an approximate sparse self-attention layer, a splicing layer, and a second linear fusion layer;

[0152] The region state data is encoded based on a preset Transformer model to obtain a first encoding vector, including:

[0153] Using the first linear fusion algorithm, linearly fuse the daily number of newly confirmed cases, the cumulative number of confirmed cases, the number of vaccine injections, and the historical intervention actions to obtain a current input vector;

[0154] The approximate sparse self-attention layer is used to calculate the current input vector to obtain the first importance, second importance, third importance and fourth importance of the daily number of newly confirmed cases, the cumulative number of confirmed cases, the number of vaccine injections and the historical intervention actions in the current input vector;

[0155] The connection layer is used to connect the daily number of newly confirmed cases, the cumulative number of confirmed cases, the number of vaccine injections, the historical intervention actions, and the first importance, the second importance, the third importance, and the fourth importance to obtain a first sub-vector, a second sub-vector, a third sub-vector, and a fourth sub-vector;

[0156] The first sub-vector, the second sub-vector, the third sub-vector and the fourth sub-vector are linearly fused using the second linear fusion layer to obtain the first encoding vector.

[0157] In an exemplary embodiment of the present disclosure, adjusting the current intervention action of the target area based on the predicted intervention level includes:

[0158] Lower the current intervention action of the target area by one level; or

[0159] Maintain the current intervention action of the target area; or

[0160] The current intervention action of the target area is adjusted up by one level;

[0161] Among them, the intervention action is used to indicate whether to perform at least one of the following action items: case isolation, wearing masks, working from home, closing schools and community lockdown.

[0162] In an exemplary embodiment of the present disclosure, the infectious disease intervention action adjustment device further includes:

[0163] A reward function construction module may be used to construct a target reward function according to the region state data, and input the region state data into the intervention action prediction model to be trained to obtain a first output result;

[0164] The network model training module can be used to construct a target loss function based on the target reward function and the first output result, and adjust the parameters of the intervention action prediction model to be trained based on the target loss function to obtain the trained intervention action prediction model.

[0165] In an exemplary embodiment of the present disclosure, in the process of training the intervention action prediction model to be trained, the training data samples used are obtained by iteratively repeating the following method: inputting the regional status data of the current time node into the intervention action prediction model to be trained to obtain the current intervention level; adjusting the current intervention action based on the current intervention level to obtain the intervention action at the next time node; using the individual simulation model to calculate the regional status data of the next time node after the intervention action at the next time node is performed in the target area, so as to input the regional status data of the next time node into the intervention action prediction model.

[0166] In an exemplary embodiment of the present disclosure, constructing a target reward function according to the area state data includes:

[0167] Constructing an epidemic severity function of the infectious disease according to the number of newly confirmed cases per day and the maximum number of confirmed cases that the target area can accommodate;

[0168] Constructing an action strictness function according to the current intervention action at the current time node and the historical intervention action at the previous time node corresponding to the current time node;

[0169] The target reward function is constructed according to the epidemic severity function and the action strictness function.

[0170] In an exemplary embodiment of the present disclosure, an action strictness function is constructed according to a current intervention action at a current time node and a historical intervention action at a previous time node corresponding to the current time node, including:

[0171] Determine the action maximum value from the current intervention action at the current time node and the historical intervention action at the previous time node;

[0172] Calculating a first ratio between the current intervention action and the action maximum, and calculating a first difference between the current intervention action and the historical intervention action;

[0173] The action strictness function is obtained by performing a weighted summation on the inverse of the first ratio and the inverse of the first difference.

[0174] In an exemplary embodiment of the present disclosure, constructing an individual simulation model according to the sociodemographic data includes:

[0175] Divide the sociodemographic data into regions according to geographical administrative regions to obtain multiple communities, and construct multiple network layers of different dimensions based on the communities and preset social scenarios;

[0176] The infection probability of each network layer of different dimensions is calculated, and the individual simulation model is constructed according to the infection probability of each network layer and the epidemic dynamics propagation model.

[0177] The specific details of each module in the above-mentioned infectious disease intervention action adjustment device have been described in detail in the corresponding infectious disease intervention action adjustment method, so they will not be repeated here.

[0178] It should be noted that, although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into multiple modules or units to be embodied.

[0179] In addition, although the steps of the method in the present disclosure are described in a specific order in the drawings, this does not require or imply that the steps must be performed in this specific order, or that all the steps shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps, etc.

[0180] In an exemplary embodiment of the present disclosure, an electronic device capable of implementing the above method is also provided.

[0181] Those skilled in the art will appreciate that various aspects of the present disclosure may be implemented as systems, methods or program products. Therefore, various aspects of the present disclosure may be specifically implemented in the following forms, namely: complete hardware implementation, complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software, which may be collectively referred to herein as "circuits", "modules" or "systems".

[0182] Refer to the following Fig.10 The electronic device 1000 according to this embodiment of the present disclosure is described. Fig.10 The electronic device 1000 shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0183] like Fig.10 As shown, the electronic device 1000 is in the form of a general computing device. The components of the electronic device 1000 may include, but are not limited to: the at least one processing unit 1010, the at least one storage unit 1020, a bus 1030 connecting different system components (including the storage unit 1020 and the processing unit 1010), and a display unit 1040.

[0184] The storage unit stores program codes, which can be executed by the processing unit 1010, so that the processing unit 1010 performs the steps according to various exemplary embodiments of the present disclosure described in the above “Exemplary Method” section of this specification. For example, the processing unit 1010 can perform the following steps: Figure 1 Step S110 shown in: obtaining socio-demographic data of a target area; wherein the socio-demographic data includes one or more of age data, family data, employment data, workplace data, school data, and commuting data; step S120: constructing an individual simulation model based on the socio-demographic data, and using the individual simulation model to calculate the regional status data after the current intervention action is performed in the target area; wherein the regional status data includes the number of newly confirmed cases per day, the cumulative number of confirmed cases, the number of vaccine injections, and a variety of historical intervention actions; step S130: inputting the regional status data into a trained intervention action prediction model to obtain a predicted intervention level, and adjusting the current intervention action of the target area based on the predicted intervention level; wherein the trained intervention action prediction model is obtained by training the intervention action prediction model to be trained using a reinforcement learning algorithm.

[0185] The storage unit 1020 may include a readable medium in the form of a volatile storage unit, such as a random access storage unit (RAM) 10201 and / or a cache storage unit 10202 , and may further include a read-only storage unit (ROM) 10203 .

[0186] The storage unit 1020 may also include a program / utility 10204 having a set (at least one) of program modules 10205, such program modules 10205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0187] Bus 1030 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0188] The electronic device 1000 may also communicate with one or more external devices 1100 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 1000, and / or may communicate with any device that enables the electronic device 1000 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface 1050. Furthermore, the electronic device 1000 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via a network adapter 1060. As shown, the network adapter 1060 communicates with other modules of the electronic device 1000 via a bus 1030. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 1000, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0189] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the implementation of the present disclosure.

[0190] In an exemplary embodiment of the present disclosure, a computer-readable storage medium is also provided, on which a program product capable of implementing the above method of the present specification is stored. In some possible implementations, various aspects of the present disclosure may also be implemented in the form of a program product, which includes a program code, and when the program product is run on a terminal device, the program code is used to enable the terminal device to execute the steps according to various exemplary implementations of the present disclosure described in the above "Exemplary Method" section of the present specification.

[0191] According to the program product for implementing the above method in the embodiment of the present disclosure, it can adopt a portable compact disk read-only memory (CD-ROM) and include program code, and can be run on a terminal device, such as a personal computer. However, the program product of the present disclosure is not limited thereto. In this document, a readable storage medium can be any tangible medium containing or storing a program, which can be used by or in combination with an instruction execution system, an apparatus or a device.

[0192] The program product may use any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0193] Computer readable signal media may include data signals propagated in baseband or as part of a carrier wave, in which readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Readable signal media may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0194] The program code embodied on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the foregoing.

[0195] Program code for performing the operations of the present disclosure may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., through the Internet using an Internet service provider).

[0196] In addition, the above-mentioned figures are only schematic illustrations of the processes included in the method according to the exemplary embodiments of the present disclosure, and are not intended to be limiting. It is easy to understand that the processes shown in the above-mentioned figures do not indicate or limit the time sequence of these processes. In addition, it is also easy to understand that these processes can be performed synchronously or asynchronously, for example, in multiple modules.

[0197] Other embodiments of the present disclosure will be readily apparent to those skilled in the art after considering the specification and practicing the inventions invented herein. This application is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not invented by the present disclosure. The specification and examples are to be considered merely exemplary, and the true scope and spirit of the present disclosure are indicated by the claims.

Claims

1. A method for adjusting intervention actions for infectious diseases, characterized in that: include: Obtaining sociodemographic data of the target area from a human resources database based on a smart city; wherein the sociodemographic data includes one or more of age data, family data, employment data, workplace data, school data, and commuting data; An individual simulation model is constructed according to the sociodemographic data, and the individual simulation model is used to calculate the regional status data after the current intervention action is performed in the target area; wherein the regional status data includes the number of newly confirmed cases per day, the cumulative number of confirmed cases, the number of vaccine injections, and multiple historical intervention actions; The regional state data is encoded based on a preset Transformer model to obtain a first encoding vector; the first encoding vector is mapped using a trained intervention action prediction model to obtain an output result; the level with the highest probability in the output result is used as the predicted intervention level, and the current intervention action of the target area is adjusted based on the predicted intervention level; wherein the trained intervention action prediction model is obtained by training the intervention action prediction model to be trained using a reinforcement learning algorithm.

2. The infectious disease intervention action adjustment method according to claim 1, characterized in that: Adjusting a current intervention action of the target area based on the predicted intervention level includes: Lower the current intervention action of the target area by one level; or Maintain the current intervention action of the target area; or The current intervention action of the target area is adjusted up by one level; Among them, the intervention action is used to indicate whether to perform at least one of the following action items: case isolation, wearing masks, working from home, closing schools and community lockdown.

3. The infectious disease intervention action adjustment method according to claim 1, characterized in that: The intervention action prediction model to be trained is trained using a reinforcement learning algorithm to obtain a trained intervention action prediction model, including: constructing a target reward function according to the region state data, and inputting the region state data into the intervention action prediction model to be trained to obtain a first output result; A target loss function is constructed according to the target reward function and the first output result, and the parameters of the intervention action prediction model to be trained are adjusted based on the target loss function to obtain the trained intervention action prediction model.

4. The method for adjusting the intervention action of infectious diseases according to claim 3, characterized in that: In the process of training the intervention action prediction model to be trained, the training data samples used are obtained by repeating the following iterative method: Inputting the regional status data at the current time node into the intervention action prediction model to be trained to obtain the current intervention level; Adjusting the current intervention action based on the current intervention level to obtain the intervention action at the next time node; The individual simulation model is used to calculate the regional state data of the next time node after the intervention action of the next time node is performed in the target area, so as to input the regional state data of the next time node into the intervention action prediction model.

5. The method for adjusting the intervention action of infectious diseases according to claim 3, characterized in that: Constructing a target reward function according to the regional state data, including: Constructing an epidemic severity function of the infectious disease according to the number of newly confirmed cases per day and the maximum number of confirmed cases that the target area can accommodate; Constructing an action strictness function according to the current intervention action at the current time node and the historical intervention action at the previous time node corresponding to the current time node; The target reward function is constructed according to the epidemic severity function and the action strictness function.

6. The method for adjusting the intervention action of infectious diseases according to claim 5, characterized in that: According to the current intervention action at the current time node and the historical intervention action at the previous time node corresponding to the current time node, an action strictness function is constructed, including: Determine the action maximum value from the current intervention action at the current time node and the historical intervention action at the previous time node; Calculating a first ratio between the current intervention action and the action maximum, and calculating a first difference between the current intervention action and the historical intervention action; The action strictness function is obtained by performing a weighted summation on the inverse of the first ratio and the inverse of the first difference.

7. The infectious disease intervention action adjustment method according to claim 1, characterized in that: Constructing an individual simulation model based on the socio-demographic data, including: Divide the sociodemographic data into regions according to geographical administrative regions to obtain multiple communities, and construct multiple network layers of different dimensions based on the communities and preset social scenarios; The infection probability of each network layer of different dimensions is calculated, and the individual simulation model is constructed according to the infection probability of each network layer and the epidemic dynamics propagation model.

8. An intervention action adjustment device for infectious diseases, characterized in that: include: A data acquisition module, used to acquire social demographic data of a target area from a human resources database based on a smart city; wherein the social demographic data includes one or more of age data, family data, employment data, workplace data, school data, and commuting data; A regional status data calculation module, used to construct an individual simulation model based on the sociodemographic data, and use the individual simulation model to calculate the regional status data after the current intervention action is performed in the target area; wherein the regional status data includes the number of newly confirmed cases per day, the cumulative number of confirmed cases, the number of vaccine injections, and multiple historical intervention actions; An intervention action adjustment module is used to encode the regional state data based on a preset Transformer model to obtain a first encoding vector; use the trained intervention action prediction model to map the first encoding vector to obtain an output result; use the level with the highest probability in the output result as the predicted intervention level, and adjust the current intervention action of the target area based on the predicted intervention level; wherein the trained intervention action prediction model is obtained by training the intervention action prediction model to be trained using a reinforcement learning algorithm.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for adjusting intervention actions for infectious diseases according to any one of claims 1 to 7 is implemented.

10. An electronic device, characterized in that: include: processor; as well as A memory, configured to store executable instructions of the processor; Wherein, the processor is configured to execute the intervention action adjustment method for an infectious disease as described in any one of claims 1-7 by executing the executable instructions.

Images