Operational Policy Planning Method

Abstract

To plan a management program in a time series satisfying a target according to a management target value.SOLUTION: A management program planning method makes a computer perform processing for setting a management target value corresponding to an output to a model which outputs an output value of an analysis by simulation with a control variable being a time-series parameter corresponding to a management program as an input, and outputting the time-series parameter at which a management target value can be obtained by using the output value which is obtained by inputting the prescribed time-series parameter as the control variable to the model, and the target value, and also by minimizing an error between the output value and the target value by employing a prescribed optimization technique.SELECTED DRAWING: Figure 3

Description

【Technical Field】 【0001】 The present invention relates to a method for planning operation policies. 【Background Art】 【0002】 Techniques for predicting the spread of infectious diseases and presenting infectious disease countermeasures that generate information related to suppressing the prevalence of infectious diseases are known (for example, Patent Document 1). In Patent Document 1, among a plurality of regions, the infection rate is low (βL) in some regions where infectious disease countermeasures are implemented, and the infection rate is high (βH) in other regions where infectious disease countermeasures are not implemented, and the spread of infectious diseases is simulated. Therefore, information related to suppressing the prevalence of infectious diseases, such as whether the prevalence of infectious diseases can be suppressed by the infectious disease countermeasures, how many regions should be subject to infectious disease countermeasures in a certain strategy, and which strategy can minimize the critical countermeasure region ratio pmin, can be generated. 【0003】 Also, techniques that enable micro-level prediction of infectious diseases specific to a region are known (for example, Patent Document 2). In Patent Document 2, the number of first patients who visited a medical institution during the first period and were diagnosed with an infectious disease, and the number of second patients who visited a medical institution during a second period different from the first period and were diagnosed with an infectious disease are input. Then, based on the number of first and second patients, a prediction of the prevalence of infectious diseases is made. 【0004】 Also, techniques that enable management of preventive measures according to the prevalence of infectious diseases such as influenza are known (for example, Patent Document 3). In Patent Document 3, based on information related to the prevalence of infectious diseases, for an organization including a plurality of users, schedule information is generated, and according to the prevalence of infectious diseases, the group to which the users in the organization belong is designated as the distribution destination, and the schedule information is distributed. 【0005】 Furthermore, techniques for predicting infectious disease outbreaks are known (for example, Patent Document 4). Patent Document 4 performs processes such as selecting probability peaks in a probability peak sequence, and determines the type of turning point, including an upward turning point which is the starting point of an epidemic season and a downward turning point which is the ending point of an epidemic season, for each epidemic season / non-epidemic season. [Prior art documents] [Patent Documents] 【0006】 [Patent Document 1] Japanese Patent Publication No. 2015-038708 [Patent Document 2] Japanese Patent Publication No. 2016-081321 [Patent Document 3] Japanese Patent Publication No. 2021-064137 [Patent Document 4] Japanese Patent Publication No. 2020-527787 [Overview of the Initiative] [Problems that the invention aims to solve] 【0007】 Taking conventional infectious disease control technologies as examples, there are those related to predicting infectious disease outbreaks and those related to both prediction and countermeasures. However, conventional technologies perform simulations based on parameters such as a certain infection rate defined under specific conditions or the number of patients over a certain period to predict the trend of transmission, or to optimize time-invariant parameters related to countermeasures. Therefore, the parameter values ​​are assumed to be fixed. 【0008】 Therefore, conventional technologies are difficult to use for planning infectious disease control measures over time, and cannot be used for planning operational measures over time. 【0009】 For example, the management departments of companies and organizations need to plan operational measures, including the duration of those measures, based on a target number of infections that is acceptable from the perspective of preventing and controlling the spread of infectious diseases. For instance, they need to consider how long contact restrictions in rooms and meeting rooms should be kept in check and when they can be restored, requiring the implementation of contact reduction measures over a timeline. Conventional technologies cannot be used for planning infectious disease control measures over such timelines. Furthermore, even when applying conventional technologies to fields other than infectious disease control, they cannot be used for planning operational measures over a timeline. 【0010】 Taking the above facts into consideration, the present invention aims to plan operational measures in a time series that satisfy the target according to the target values ​​of the operation. [Means for solving the problem] 【0011】 To achieve the above objective, the operational policy planning method of the present invention involves setting an operational target value corresponding to the output of a model that takes control variables, which are time-series parameters corresponding to operational policies, as input and outputs output values ​​from a simulation analysis. The computer then applies a predetermined optimization method to minimize the error between the output value and the target value by inputting predetermined time-series parameters as control variables into the model and using the output value obtained with respect to the target value, thereby outputting time-series parameters that realize the operational target value. This allows the computer to plan operational policies in a time series that satisfy the target according to the operational target value. 【0012】 Furthermore, in the operational strategy planning method, the model is configured to output the output values ​​for each time period, and target values ​​are set for each time period. Using the output values ​​obtained for each time period and the target values ​​for each time period, the time-series parameters can be output for each time period. This allows for the output of time-series parameters for each time period and enables the planning of operational strategies that satisfy the targets for each time period. 【0013】 Furthermore, in the operational strategy planning method, the model is configured to output the output values ​​for each operational area, and target values ​​are set for each area. Using the output values ​​obtained for each area and the target values ​​for each area, the time-series parameters can be output for each area. This allows for the output of time-series parameters for each area and the planning of operational strategies that satisfy the targets for each area. 【0014】 Furthermore, taking infectious disease control as an example, in the operational policy planning method, the model can be set as an infection simulation model, the time-series parameter can be set as the expected number of contacts with an infected person, and the operational target value can be set as the number of infected persons that the business establishment can tolerate in its operations. This allows the time-series parameter to be output in the infection simulation to achieve the operational target value for suppressing the number of infected persons, thereby enabling the planning of operational policies. [Effects of the Invention] 【0015】 According to the present invention, it is possible to plan operational measures in a time series that satisfy the target according to the target value of the operation. [Brief explanation of the drawing] 【0016】 [Figure 1] This figure shows an example of an infection simulation model. [Figure 2A] This is a graph showing the trend in the expected number of contacts. [Figure 2B] This is a graph of the simulation results. [Figure 3] This is a block diagram showing the configuration of the planning device according to this embodiment. [Figure 4A] This is a graph showing the trend in the expected number of contacts. [Figure 4B] This is a graph of the simulation results. [Figure 5] This is a flowchart showing the planning process in the planning device according to this embodiment. [Modes for carrying out the invention] 【0017】 [Embodiments of the Present Invention] Hereinafter, a method according to an embodiment of the present invention will be described with reference to the drawings. As described in the above problems, conventionally, operation measures have been considered by optimizing fixed input parameters. The method of this embodiment enables the use in planning operation measures on the time axis by setting time-series control variables as the optimization targets. Thereby, it becomes possible to consider operation measures taking into account the transition on the time axis on the simulator. Taking infection control measures as an example, the implementation plan of contact reduction measures such as limiting the number of people in living rooms and conference rooms corresponds to the operation measures. 【0018】 In this embodiment, as an example of operation measures, a case where it is set as an optimization problem to obtain the transition of the expected number of contacts that realizes the transition of the target number of infected persons will be described as an example. 【0019】 Figure 1 shows an example of an infection simulation model. This model outputs simulation results according to control variables of time-series parameters, and outputs the number of infected people according to the expected number of contacts with latent individuals (time-series parameter). The expected number of contacts here refers to the expected number of contacts with latent individuals. The left side of Figure 1 shows the time series progression of the number of contacts. The vertical axis is the expected number of contacts, and the horizontal axis is the time series. The right side shows the simulation results of the model. The vertical axis is the predicted number of infected people, and the horizontal axis is the time series. The dotted line shows the time series progression of the expected number of contacts when the expected number of contacts is kept constant (i.e., fixed as before) and no countermeasures are taken, while the solid line shows the case after countermeasures have been taken over time. In this embodiment, while setting target infection control measures (target values) so that the simulation results after countermeasures are obtained, the expected number of contacts assumed in the countermeasures is not constant, and the parameter changes according to the progression of the time series, so that a time-series parameter that satisfies the target is obtained. In other words, it is a time-series parameter optimized for the operational objective. The time-series parameter of the expected number of contacts that satisfies the target can be considered as a time-series operational measure. Furthermore, the scope of application of this embodiment is not limited to infectious disease infection prediction, but can also be applied to facility operation planning such as alleviating congestion in theme parks and commercial facilities by restricting entry by time of day, and mitigating traffic congestion through dynamic road pricing, and various effects can be expected. 【0020】 In this embodiment, we assume that we will perform optimization analysis of the model. Here, we will explain using the simulation results when the time series is kept constant as an example. Figure 2A is a graph of the change in the expected number of contacts, and Figure 2B is a graph of the simulation results. The solid line in Figure 2B is the output result when infection control measures are continuously implemented, and corresponds to the solid line in Figure 2A with measures in place. The dashed line in Figure 2B is the output result when infection control measures are not implemented, and corresponds to the dashed line in Figure 2A with no measures in place. Thus, when parameters are fixed and a constant set of measures is continued, the effect of the measures is obtained in the simulation results, but it is not desirable in terms of operational policies to always continue with the same set of measures. Therefore, time series parameters are used as control variables to enable the operation of operational policies to be carried out in a practical manner. 【0021】 Figure 3 is a block diagram showing the configuration of the planning device 100 according to this embodiment. As shown in Figure 3, the planning device 100 is configured to include a model creation unit 110, a setting unit 112, a model execution unit 114, and an optimization unit 116. The planning device 100 is implemented by a computer including a CPU (Central Processing Unit), a ROM (Read Only Memory) that stores programs for realizing each processing routine, a RAM (Random Access Memory) for temporarily storing data, memory as a storage means, and a network interface. Note that the planning device 100 may acquire a model from an external device in advance, or it may be configured without a model creation unit 110. 【0022】 The model creation unit 110 creates a model used for optimizing time-series parameters. The model takes control variables, which are time-series parameters, as input and outputs the output value of the simulation analysis. The model can be of any type as long as it can set target values ​​and time-series parameters as control variables. For example, a model can be used that is based on the model in Reference 1, with the necessary parts modified to incorporate control variables that are time-series parameters (expected number of contacts with the suspected individual) that have been time-seriesized. If the input to the model is predetermined to be a fixed parameter θ, then the time-series parameter θ t It should be treated as such. [References 1] Rahmandad, H., & Sterman, J. (2008). Heterogeneity and network structure in the dynamics of diffusion: Comparing agent-based and differential equation models. Management science, 54(5), 998-1014. 【0023】 Furthermore, for example, Patent Document 1 lists the SIS model, SIR model, and SEIR model as simulation models for simulating the transmission of infectious diseases, and these can be modified and applied. If we consider applying these to the model of this embodiment, since they are all systems of differential equations, we can create a model that can be created by performing discrete-time simulation using the Runge-Kutta method or the like, and then modifying it so that the parameters, which are time-series, can be set as control variables. 【0024】 The setting unit 112 sets target values ​​for operation corresponding to the model's output. The target values ​​can be set as data showing the trend of the number of infected people over time. The target values ​​can be set as the number of infected people that the facility can tolerate, or as values ​​obtained in advance from the relationship between the simulation results and the time-series parameters when countermeasures are in place, as shown in Figures 2A and 2B above. 【0025】 The model execution unit 114 outputs output values ​​to be used in the optimization unit 116, which will be described later. The processing of the model execution unit 114 is performed through the processing of the optimization unit 116. The model execution unit 114 outputs output values ​​obtained by inputting predetermined time series parameters as control variables to the model created by the model creation unit 110. The model execution unit 114 is given time series parameters that are updated during the processing of the optimization unit 116. The initial values ​​of the predetermined time series parameters can be, for example, time series parameters assuming no countermeasures are taken. 【0026】 The optimization unit 116 uses the output values ​​obtained from the model by the model execution unit 114 and the target values ​​set by the setting unit 112 to apply a predetermined optimization method to minimize the error between the output values ​​and the target values, thereby outputting time-series parameters that realize the target values ​​for operation. As an optimization method, for example, a modal iterative error correction method that minimizes the error between the target values ​​and the output values ​​of the simulation may be used. 【0027】 The display unit 118 displays, in graph form, the results of the time-series parameters that achieve the target values ​​of the operation output from the optimization unit 116, and the simulation results using those time-series parameters compared to the target values. An example of a graph is shown below. 【0028】 Figure 4A is a graph of time-series parameters that achieve the target values ​​of the operation optimized by the optimization unit 116. Looking at the optimized time-series parameters in Figure 4A, the optimization result is output in which the parameters change in stages from parameters corresponding to having countermeasures to parameters corresponding to not having countermeasures, in accordance with the progress of the time series. Figure 4B is a graph of the simulation result when the optimized time-series parameters are input to the model as control variables. It can be seen that the simulation result with the optimized time-series parameters converges to match the target value. Thus, it can be seen that the expected number of contacts trend that satisfies the target number of infected people trend has been searched. Furthermore, the searched expected number of contacts trend exceeds the expected number of contacts when no countermeasures are taken in the latter half, indicating that if relatively strong infection control measures are implemented in the first half, no special suppression is necessary in the latter half. In this way, the optimized time-series parameters allow us to understand what countermeasures should be taken over time. Note that if fixed parameters are assumed to be optimized, the target number of infected people trend can be achieved, but since the expected number of contacts is always fixed, it is not possible to determine what countermeasures should be taken over time. 【0029】 Next, the operation of the planning device 100 in this embodiment will be described. Figure 5 is a flowchart of the planning process in the planning device 100 according to this embodiment. The CPU reads programs and various data from ROM and executes them, thereby functioning as a part of the planning device 100 and performing the planning process. The planning process is an example of an operational policy planning method. 【0030】 In step S100, the CPU, acting as a model creation unit 110, creates a model to be used for optimizing time-series parameters. 【0031】 In step S102, the CPU, acting as the setting unit 112, sets target values ​​for operation corresponding to the model's output. 【0032】 In step S104, the CPU, acting as an optimization unit 116, uses the output values ​​obtained from the model and the set target values ​​to apply a predetermined optimization method to minimize the error between the output values ​​and the target values, thereby outputting time-series parameters that realize the target values ​​for operation. 【0033】 In step S106, the CPU, acting as a display unit 118, displays a graph showing the results of the time-series parameters that achieve the target values ​​for operation, and the simulation results using those time-series parameters compared to the target values. 【0034】 In the example of the embodiment described above, the model is set as an infection simulation model, the time-series parameter is set as the expected number of contacts with an incubator, and the operational target value is set as the number of infected persons that the business establishment can tolerate in its operations. 【0035】 As described above, the planning device 100 according to this embodiment makes it possible to plan operational measures in a time series that satisfy the objective. 【0036】 It should be noted that the present invention is not limited to the embodiments described above, and various modifications and applications are possible without departing from the spirit of the invention. The points to note when applying this embodiment are as follows. 【0037】 For example, the model creation unit 110 may create a model for each time period. In this case, the model for each time period is set to output output values ​​corresponding to that time period. Therefore, the setting unit 112 sets target values ​​for each time period. Furthermore, the optimization unit 116 can output time-series parameters for each time period using the output values ​​obtained for each time period and the target values ​​for each time period. 【0038】 Furthermore, the model creation unit 110 may create models for each area. Various ranges can be set for each area, depending on the range of the model's prediction target. If the prediction target range is a region, the range is a division of the region; if the range is within a building, it is a section of the building. In this case, the model for each area is set to output output values ​​for each area of ​​operation. Target values ​​can be set for each time period, and the time series parameters can be output for each time period using the output values ​​obtained for each time period and the target values ​​for each time period. This allows for the output of time series parameters for each area and the planning of operational measures that satisfy the targets for each area. 【0039】 Alternatively, the above embodiment example may be used as the overall time-series parameter, and time-series parameters for each time period and each area may be appropriately combined and weighted to obtain time-series parameters that achieve the target values ​​for operation, which are individually adjusted for each time period and each area. [Explanation of symbols] 【0040】 100 Planning device 110 Model Creation Department 112 Settings Section 114 Model Execution Unit 116 Optimization Unit 118 Display section

Claims

[Claim 1] For a model that takes control variables, which are time-series parameters corresponding to operational measures, as input and outputs values ​​from simulation analysis, set operational target values ​​corresponding to the output. By inputting predetermined time-series parameters as control variables to the aforementioned model, the output values ​​obtained and the target values ​​are used, and a predetermined optimization method is applied to minimize the error between the output values ​​and the target values, thereby outputting time-series parameters that realize the target values ​​for operation. A method for planning operational measures where processing is performed by a computer. [Claim 2] The aforementioned model is configured to output the aforementioned output value for each time period. Set the aforementioned target values ​​for each time period, Using the output values ​​obtained for each time period and the target values ​​for each time period, Output the aforementioned time-series parameters for each time period. The operational policy planning method described in claim 1. [Claim 3] The aforementioned model is configured to output the aforementioned output value for each area of ​​operation. Set the aforementioned target values ​​for each area. Using the output value obtained for each area and the target value for each area, Output the aforementioned time-series parameters for each area. The operational policy planning method described in claim 1. [Claim 4] The aforementioned model is set up as a model for infection simulation. The aforementioned time-series parameter is the expected number of contacts with the latent person. The operational plan method according to claim 1, wherein the target value for the aforementioned operations is the number of infected persons determined as the limit that the business establishment can tolerate in its operations.

Images