Risk analysis method and risk analysis system

Abstract

To mitigate risks during emergencies, clarify the actions that should be taken during normal times. [Solution] In the risk analysis method, the risk analysis system defines multiple normal-time objective functions, which are objective functions that represent the objective to be achieved under normal circumstances, by using a predetermined formula as the input of a predetermined function that outputs a predetermined value for an input, and including multiple explanatory factors obtained by multiplying each explanatory factor by a first contribution as a first-order term or factor. The risk analysis system then defines an emergency objective function, which is an objective function that represents the objective to be achieved in order to reduce risk, as a function that outputs a value using a predetermined formula as the input, which includes multiple normal-time objective functions obtained by multiplying each normal-time objective function by a second contribution as a first-order term or factor. The risk analysis system then outputs the second contribution in the emergency objective function by machine learning of a second sample data related to the emergency objective function, and extracts a preferred normal-time objective function that should be prioritized for optimization from the normal-time objective functions based on the second contribution.

Description

Technical Field

[0001] The present invention relates to a risk analysis method and a risk analysis system.

Background Art

[0007] According to the present invention, for example, it is possible to clarify what should be done during normal times in order to reduce risks in emergencies. [Brief explanation of the drawing]

[0008] [Figure 1A] A diagram showing the configuration of the normal-time highest priority countermeasure selection system according to Embodiment 1. [Figure 1B] A diagram showing the hardware configuration of the normal-time highest priority countermeasure selection system according to Embodiment 1. [Figure 2] A flowchart illustrating the normal highest priority countermeasure selection process according to Embodiment 1. [Figure 3A] A diagram showing the screen for selecting the highest priority countermeasure during normal operation according to Embodiment 1. [Figure 3B] A diagram showing the screen for selecting the highest priority countermeasure during normal operation according to Embodiment 1. [Figure 3C] A diagram showing the screen for selecting the highest priority countermeasure during normal operation according to Embodiment 1. [Figure 3D] A diagram showing the screen for selecting the highest priority countermeasure during normal operation according to Embodiment 1. [Figure 3E] A diagram showing the screen for selecting the highest priority countermeasure during normal operation according to Embodiment 1. [Figure 4] A diagram showing the normal-mode objective function group, the emergency-mode objective function, and the explanatory factors according to Embodiment 2. [Modes for carrying out the invention]

[0009] Embodiments of the present invention will be described in detail below with reference to the drawings.

[0010] [Embodiment 1] Embodiment 1 aims to explore and present measures that people should take during normal times before an infectious disease outbreak in order to prevent the risk of infected individuals developing severe symptoms during an infectious disease outbreak.

[0011] In Embodiment 1, the emergency objective function G(x) to be minimized is defined as "the probability of severe illness from an infectious disease during an infectious disease epidemic." In Embodiment 1, the normal-time objective function Fi(x) to be minimized, based on explanatory factors Ej(x) which represent the health status of each subject, is defined as "the probability of developing diabetes," "the probability of developing hypertension," etc., which represent the disease prevalence of each subject. However, i=1,2,...,N and j=1,2,...,m. The explanatory factors Ej(x) for each subject are "HbA1c value ≥ 6.5," "systolic blood pressure value ≥ 140," etc., and are based on information such as health checkups, questionnaire responses, and diagnostic prescriptions for each subject.

[0012] (Configuration of the normal-time highest priority countermeasure selection system 1 according to Embodiment 1) Figure 1A shows the configuration of the normal-time highest priority countermeasure selection system 1 according to Embodiment 1. The normal-time highest priority countermeasure selection system 1 is a computer having a processor 11, memory 12, storage unit 13, input / output unit 14, and communication unit 15. The normal-time highest priority countermeasure selection system 1 is an example of a risk analysis system that performs risk analysis to reduce risks that occur in emergencies.

[0013] The processor 11 is a CPU (Central Processing Unit) or the like that realizes each functional unit by executing a program in cooperation with the memory 12. The processor 11 includes a normal objective function group definition unit 11a, an emergency objective function definition unit 11b, a priority normal objective function extraction unit 11c, a priority maximum explanatory factor extraction unit 11d, and a normal highest priority countermeasure output unit 11e.

[0014] The normal objective function group definition unit 11a defines a plurality of normal objective functions Fi(x) that are objective functions representing the objectives to be achieved during normal times. The normal objective function Fi(x) is a predetermined formula including, as a primary term or factor, a plurality of explanatory factors Ej(x) obtained by multiplying the input of a predetermined function (for example, a sigmoid function) that outputs a predetermined value with respect to the input by the contribution degree qij (the first contribution degree) related to each explanatory factor Ej(x). The predetermined formula here is, for example, qi1×E1(x)+···+qim×Em(x) of the following formulas (I-1) to (I-N).

[0015] The emergency objective function definition unit 11b defines an emergency objective function G(x) that is an objective function representing the objective to be achieved for reducing the risk (aggravation of an infectious disease during an infectious disease epidemic) occurring during an emergency. The emergency objective function G(x) is a function that outputs a value with, as an input, a predetermined formula including, as a primary term or factor, a plurality of normal objective functions Fi(x) obtained by multiplying the contribution degree wj (the second contribution degree) related to each normal objective function Fi(x). The predetermined formula here is, for example, w1×F1(x)+···+wN×FN(x) of the following formula (2).

[0016] The priority normal objective function extraction unit 11c outputs, by machine learning of the second sample data (for example, the normal objective function extraction learning data 13c) related to the emergency objective function G(x), the contribution degree wj in the emergency objective function G(x). Then, the priority normal objective function extraction unit 11c extracts a priority normal objective function Fs(x) to be preferentially optimized from the normal objective functions Fi(x) based on the contribution degree wj.

[0017] The priority maximum explanatory factor extraction unit 11d outputs the contribution degree qij in the normal priority objective function Fs(x) by machine learning using the first sample data related to the normal priority objective function Fs(x) (for example, the learning data 130d for extracting the priority maximum explanatory factor). Then, the priority maximum explanatory factor extraction unit 11d extracts explanatory factors corresponding to a predetermined number (N natural numbers from the largest to the Nth) of contribution degrees qij in descending order of the contribution degree qij.

[0018] The normal-time most-priority countermeasure output unit 11e manages by associating the explanatory factor Ej(x) with countermeasure candidates for reducing risks in the explanatory factor countermeasure candidate table 13e described later. The normal-time most-priority countermeasure output unit 11e acquires and outputs the countermeasures corresponding to the explanatory factors extracted by the priority maximum explanatory factor extraction unit 11d by referring to the explanatory factor countermeasure candidate table \\(13e\\).

[0019] The storage unit 13 is a storage device that stores programs and various data. The storage unit 13 stores the normal-time objective function definition data 13a, the emergency-time objective function definition data 13b, the normal-time objective function extraction learning data 13c, the priority maximum explanatory factor extraction learning data 13d, and the explanatory factor countermeasure candidate table 13e.

[0020] The input / output unit 14 includes an input device such as a keyboard and a mouse, and an output device such as a display. The normal-time most-priority countermeasure selection screen 13D described later is displayed on the display of the input / output unit 14. The communication unit <15> is a communication device when the normal-time most-priority countermeasure selection system 1 communicates with other computers.

[0021] (Hardware Configuration of the Normal-Time Most-Priority Countermeasure Selection System 1) FIG. 1B is a diagram showing the hardware configuration of the normal-time most-priority countermeasure selection system 1 according to Embodiment 1. The normal-time most-priority countermeasure selection system 1 is constructed on the server device 1A. The server device 1A has a CPU 11A, a memory 12A, a storage 13A, and an output I / F 14A. The CPU 11A realizes each functional unit of the normal-time most-priority countermeasure selection system 1 shown in FIG. 1A by executing a program in cooperation with the memory 12A.

[0022] Server device (storage) 1A stores normal objective function definition data 13a, emergency objective function definition data 13b, training data for normal objective function extraction 13c, and training data for priority maximum explanatory factor extraction 13d in storage 13A. Server device (storage) 1A also stores explanatory factor countermeasure candidate table 13e and various hardware operation setting parameters in storage 13A.

[0023] A display device (monitor) 14A1 is connected to the server device 1A via an output interface 14A. The server device 1A displays the normal operation objective function Fi(x), the emergency operation objective function G(x), the preferred normal operation objective function Fs(x), the preferred maximum explanatory factor qst, and the normal operation highest priority countermeasure Mt on the display device (monitor) 14A1, as described below.

[0024] (Normally, the highest priority action selection process) Figure 2 is a flowchart showing the normal-time highest priority countermeasure selection process according to Embodiment 1. Figures 3A to 3E show the normal-time highest priority countermeasure selection screen 13D according to Embodiment 1. The normal-time highest priority countermeasure selection screen 13D includes, for example, a normal-time objective function group definition box D11, an emergency objective function definition box D12, a priority normal-time objective function extraction box D13, and a priority maximum explanatory factor extraction box D14, as shown in Figure 3A. The normal-time highest priority countermeasure selection screen 13D also includes, for example, a normal-time highest priority countermeasure extraction box D15 and a selection field D16, as shown in Figure 3A.

[0025] First, in step S11, the normal-time objective function group definition unit 11a defines the normal-time objective function Fi(x) (i=1,2,…,N) as shown in equations (1-1) to (1-N), and stores it in the storage unit 13 as normal-time objective function definition data 13a. The normal-time objective function F1(x) is the probability of developing diabetes, the normal-time objective function F2(x) is the probability of developing hypertension, and so on. Each normal-time objective function Fi(x) represents the relationship between the disease prevalence of each subject and each explanatory factor. F1(x)=f1(q11×E1(x)+···+q1m×Em(x))…(1-1) F2(x)=f1(q21×E1(x)+···+q2m×Em(x))…(1-2) ... FN(x)=f1(qN1×E1(x)+···+qNm×Em(x))…(1-N)

[0026] In equations (1-1) to (1-N), f1(*) is a function that outputs a factor, such as a sigmoid function, but other functions (such as various activation functions) may also be used. The explanatory factor Ej(x) (j=1,2,…,m) takes the value "1" if the subject is an explanatory factor and "0" if it is not. qij(i=1,2,…,N, j=1,2,…,m) is the contribution of the explanatory factor Ej(x) to the objective function Fi(x) under normal circumstances.

[0027] Furthermore, the substitution of explanatory factors Ej(x) into the normal objective function Fi(x) is not limited to linear substitution as shown in equations (1-1) to (1-N), but may also be done by substituting each explanatory factor Ej(x) multiplied by its contribution into a separable nonlinear form. The same applies to the normal objective function Fi(x) in the emergency objective function G(x).

[0028] As shown in Figure 3A, in this embodiment, when the normal objective function group definition box D11 is expanded, the normal objective function group definition input box D111 appears. Also, spin boxes D16a to D161j, ... appear in the selection field D16 for selecting the definitions of the normal objective function Fi(x) and explanatory factors Ej(x). For example, in Figure 3A, spin box D16a has F1(x) "probability of developing diabetes" selected. Also, for example, spin box D16d has E1(x) "HbA1c value ≥ 6.5" selected.

[0029] Next, in step S12, the emergency objective function definition unit 11b defines the emergency objective function G(x) as shown in equation (2) and stores it in the storage unit 13 as emergency objective function definition data 13b. The emergency objective function G(x) represents the relationship between the severity of infectious disease during an infectious disease epidemic and each normal objective function Fi(x). G(x)=f2(w1×F1(x)+···+wN×FN(x))…(2)

[0030] In equation (2), f2(*) is a function that outputs a factor, for example, a sigmoid function, but other functions (for example, various activation functions) may also be used. The normal objective function Fi(x) is defined as in equations (1-1) to (1-N). wi is the contribution of Fi(x) to G(x).

[0031] As shown in Figure 3B, in this embodiment, when the emergency objective function definition box D12 is expanded, the emergency objective function definition input box D121 appears. In addition, spin boxes D16k, D16a~D16c, ... that display the definitions of the emergency objective function G(x) and the normal objective function Fi(x) appear in the selection field D16.

[0032] Next, in step S13, the preferred normal-time objective function extraction unit 11c extracts the main explanatory factors of the emergency objective function G(x) as the preferred normal-time objective function Fs(x). Specifically, the preferred normal-time objective function extraction unit 11c outputs contributions w1, ..., wN using machine learning. Then, as shown in equation (3), it extracts the preferred normal-time objective function Fs(x) corresponding to the index s that gives the largest contribution ws among the contributions w1, ..., wN. Fs(x)=f1(qs1×E1(x)+···+qsm×Em(x))…(3)

[0033] As shown in Figure 3C, in this embodiment, when the preferred normal objective function extraction box D13 is expanded, the preferred normal objective function display box D131 appears. Also, spin boxes D16k, D16a~D161c, ... that display the definitions of the emergency objective function G(x) and the normal objective function Fi(x) appear in the selection field D16. Also, a spin box D16l for selecting the name of the machine learning data file to be used in the machine learning in step S13 appears in the selection field D16. Also, a machine learning data display D16m that displays the training data 13c for normal objective function extraction corresponding to the name of the file selected in spin box D16l appears in the selection field D16. For example, in Figure 3C, the training data 13c for normal objective function extraction with the file name "test.csv" is selected in spin box D16l.

[0034] As shown in Figure 3C, in the training data 13c for extracting the normal-time objective function, for each subject's ID, the normal-time objective function Fi(x) that applies to each subject is set to "1", and the normal-time objective function Fi(x) that does not apply to each subject is set to "0". In addition, the emergency objective function G(x) is set to "1" if each subject falls under the category of "severe illness of infectious disease during an infectious disease epidemic", and to "0" if they do not. In step S13, the contributions w1, ..., wN in equation (2) are calculated by machine learning the training data 13c for extracting the normal-time objective function.

[0035] Next, in step S14, the priority maximum explanatory factor extraction unit 11d extracts the main explanatory factors of the priority normal objective function Fs(x) extracted in step S13 as the priority maximum explanatory factor qst. Specifically, the priority maximum explanatory factor extraction unit 11d outputs the contributions qs1, ..., qsm of equation (3) using machine learning. Then, the priority maximum explanatory factor extraction unit 11d extracts the priority maximum explanatory factor Et(x) corresponding to index t, which gives the priority maximum explanatory factor qst, the priority maximum explanatory factor with the largest contribution among the contributions qs1, ..., qsm.

[0036] As shown in Figure 3D, in this embodiment, when the preferred maximum explanatory factor extraction box D14 is expanded, the preferred maximum explanatory factor display box D141 appears. Also, a spin box D16z appears in the selection field D16, which displays the definition of the preferred normal objective function Fs(x). Also, spin boxes D16d~D16j, ... appear in the selection field D16, which display the definitions of the explanatory factors Ej(x). Also, a spin box D16o appears in the selection field D16 for selecting the name of the machine learning data file to be used in the machine learning in step S14. Also, a machine learning data display D16p appears in the selection field D16, which displays the preferred maximum explanatory factor extraction training data 13d corresponding to the name of the file selected in spin box D16o. For example, in Figure 3D, spin box D16o has selected the preferred maximum explanatory factor extraction training data 13d with the file name "test2.csv".

[0037] As shown in Figure 3D, in the training data 13d for extracting the preferred maximum explanatory factor, for each subject's ID, "1" is assigned to the explanatory factor Ej(x) to which the subject belongs, and "0" is assigned to the explanatory factor Ei(x) to which the subject does not belong. In addition, "1" is assigned if each subject corresponds to the preferred normal objective function Fs(x), and "0" if they do not. In step S14, the training data 13d for extracting the preferred maximum explanatory factor is subjected to machine learning, and the contributions qs1, ..., qsm in equation (3) are calculated.

[0038] Next, in step S15, the normal-time highest priority countermeasure output unit 11e extracts the normal-time highest priority countermeasure Mt corresponding to the priority maximum explanatory factor Et(x) extracted in step S14 by referring to the explanatory factor countermeasure candidate table 13e, and displays it on the display of the input / output unit 14.

[0039] As shown in Figure 3E, in this embodiment, when the normal highest priority countermeasure extraction box D15 is expanded, the normal highest priority countermeasure display box D151 appears. Then, the explanatory factor countermeasure candidate table 13e is displayed in the explanatory factor countermeasure candidate table display D16q in the selection field D16. The explanatory factor countermeasure candidate table 13e stores the preferred maximum explanatory factor Et(x) and the normal highest priority countermeasure Mt in association with each other.

[0040] In Embodiment 1, for example, the preferred normal-time objective function Fs(x) that contributes most to the emergency objective function G(x) is assumed to be "the probability of developing diabetes." Also, for example, the preferred maximum explanatory factor Et(x) that contributes most to "the probability of developing diabetes" is assumed to be "drinking frequency ≥ 5 days per week." In this case, "substituting non-alcoholic beverages" to improve "drinking frequency" is presented as the highest priority normal-time measure Mt.

[0041] (Modified version of Embodiment 1) 1. Calculation model for extracting the preferred maximum explanatory factor Et(x) In Embodiment 1, in step S14 (Figure 2) of the normal highest priority countermeasure selection process, the contributions qs1, ..., qsm of equation (3) are output in common for all subject IDs, but they may be output individually for each subject ID. That is, the contribution qij may be output by machine learning for each sample of the training data 13d for extracting the preferred maximum explanatory factor. In other words, the calculation model may be changed for each subject ID when determining the preferred maximum explanatory factor qst for extracting the preferred maximum explanatory factor Et(x). By changing the calculation model for determining the preferred maximum explanatory factor qst for each subject ID in this way, it is possible to determine the preferred maximum explanatory factor Et(x) that is most appropriate for each subject, while taking into account the individual circumstances of each subject.

[0042] 2. Extraction of multiple priority maximum explanatory factors Et(x) In Embodiment 1, in step S14 (Figure 2) of the normal top priority countermeasure selection process, only one explanatory factor is extracted as the priority maximum explanatory factor Et(x). However, this is not limited to this, and multiple explanatory factors may be extracted. For example, for each subject's ID, unrealistic normal top priority countermeasures Mt may be excluded in advance, and from the remaining normal top priority countermeasures Mt, the priority maximum explanatory factor Et(x) corresponding to the top predetermined number of contributions qs1, ..., qsm may be extracted. For example, by applying the method for calculating importance (contribution) for each sample (feature vector) disclosed in a well-known document (Japanese Patent Publication No. 6912998), the priority normal objective function Fs(x) and the priority maximum explanatory factor Et(x) can be extracted for each subject's ID.

[0043] Furthermore, the magnitude of the contribution qij represents the importance of the corresponding primary maximum explanatory factor Et(x) and the countermeasures that correspond to that primary maximum explanatory factor Et(x). Therefore, by outputting the priority of countermeasures corresponding to explanatory factors Ej(x) according to the magnitude of the contribution qij along with the countermeasures, it becomes clear which of the multiple countermeasures should be implemented first.

[0044] (Effects of Embodiment 1) In the above-described embodiment 1, a second contribution to the emergency objective function is output by machine learning of a second sample data related to the emergency objective function, and a priority normal-time objective function that should be prioritized for optimization is extracted from the normal-time objective function based on the second contribution. This makes it clear which normal-time objective function should be prioritized for optimization in order to optimize the emergency objective function.

[0045] Furthermore, in the above-described embodiment 1, the first contribution to the preferred normal-time objective function is output by machine learning of the first sample data related to the preferred normal-time objective function, and a predetermined number of explanatory factors corresponding to the first contribution are extracted in descending order of the first contribution. This makes it clear which explanatory factors should be prioritized for optimization in order to optimize both the emergency objective function and the preferred normal-time objective function.

[0046] Furthermore, in the above-described embodiment 1, explanatory factors and measures to reduce risk are managed in association, and measures corresponding to the explanatory factors extracted based on the preferred normal-time objective function are output. This clarifies the specific actions that should be taken to optimize the explanatory factors.

[0047] Furthermore, in the above-described embodiment 1, the first contribution is output for each sample of the first sample data by machine learning. This makes it possible to determine the most appropriate priority maximum explanatory factor for each sample, taking into account the individual circumstances of the sample.

[0048] Furthermore, in the above-described embodiment 1, along with the countermeasures, the priority order of countermeasures corresponding to a predetermined number of explanatory factors corresponding to the magnitude of the first contribution is output. This makes it clear which of the presented multiple countermeasures should be implemented first.

[0049] Furthermore, in Embodiment 1 described above, the risk is the severity of infectious disease during an infectious disease outbreak, and each explanatory factor is data related to the health status of each subject. Each normal-time objective function represents the relationship between each subject's disease prevalence and each explanatory factor, and the emergency objective function represents the relationship between the severity of infectious disease during an infectious disease outbreak and each normal-time objective function. This makes it possible to clarify the measures that people should take during normal times before an infectious disease outbreak occurs in order to achieve the objective of preventing the risk of severe illness in people infected with infectious diseases during an infectious disease outbreak.

[0050] [Embodiment 2] The following description of Embodiment 2 will focus on the differences from Embodiment 1, and redundant explanations will be omitted.

[0051] In Embodiment 2, the objective is to reduce the number of days of delay in delivery of industrial products by manufacturers in the event of an earthquake. To achieve this, measures that each supplier that supplies parts to manufacturers under normal circumstances before an earthquake occurs are explored and presented.

[0052] Figure 4 shows the set of normal-time objective functions, the emergency-time objective function, and the explanatory factors according to Embodiment 2.

[0053] In Embodiment 2, the emergency objective function G(x) to be minimized is defined as "the number of days of delivery delay (by the manufacturer) in the event of an earthquake." In Embodiment 2, the normal-time objective function Fi(x) to be minimized, based on the explanatory factor Ej(x) for each supplier, is defined as "the number of days of delivery delay for part A (by the supplier)," "the number of days of delivery delay for part B (by the supplier)," etc. The explanatory factor Ei(x) for each supplier is defined as "earthquake resistance of the factory building ≤ seismic intensity 5," "number of days to restore power to the factory ≥ 3 days," etc., and is information based on surveys, inspections, and answers to questions conducted with each supplier.

[0054] In Embodiment 2, instead of explanatory factor countermeasure candidate table 13e, explanatory factor countermeasure candidate table 13e2 is stored in the storage unit 13. In Embodiment 2, for example, suppose the priority normal-time objective function Fs(x) that contributes most to the emergency objective function G(x) is "the number of days of delivery delay for part A (when an earthquake occurs)". Also, suppose the priority maximum explanatory factor Et(x) that contributes most to "the number of days of delivery delay for part A (when an earthquake occurs)" is "the seismic resistance strength of the factory building (of the supplier of part A) ≤ seismic intensity 5". In this case, "seismic reinforcement work on the factory building (of the supplier of part A)" that improves "the seismic resistance strength of the factory building (of the supplier of part A)" is presented as the highest priority countermeasure Mt during normal times.

[0055] In other words, in Embodiment 2, the risk that arises in an emergency is the delay in product delivery due to the occurrence of an earthquake. Also in Embodiment 2, each explanatory factor Ej(x) is data related to the operation of each factory that produces the components that make up each product. Also in Embodiment 2, each normal-time objective function Fi(x) represents the relationship between the number of days of delay in delivery of each component supplied by each factory and the explanatory factor Ej(x). Also in Embodiment 2, the emergency objective function G(x) represents the relationship between the number of days of delay in product delivery due to the occurrence of an earthquake and each normal-time objective function Fi(x).

[0056] (Effects of Embodiment 2) In the above-described embodiment 2, the risk is the delay in product delivery during an earthquake, and each explanatory factor is data related to the operation of each factory that produces the components that make up each product. Each normal-time objective function represents the relationship between the number of days of delay in delivery of each component supplied by each factory and the explanatory factor, and the emergency objective function represents the relationship between the number of days of delay in product delivery during an earthquake and each normal-time objective function. This makes it possible to clarify the measures that each supplier that supplies components to manufacturers during normal times before an earthquake occurs in order to achieve the objective of reducing the number of days of delay in delivery of industrial products by manufacturers during an earthquake.

[0057] The embodiments described above are explained in detail for the purpose of clearly illustrating the present invention, and are not necessarily limited to those comprising all the described configurations. Furthermore, it is possible to replace parts of the configuration of one embodiment with those of another embodiment, and to add configurations from other embodiments to the configuration of one embodiment. Also, it is possible to add, delete, or replace parts of the configuration of each embodiment with those of other embodiments. Some or all of the above configurations, functions, processing units, processing means, etc., may be implemented in hardware, for example, by designing them as integrated circuits. Alternatively, the above configurations, functions, etc., may be implemented in software by having a processor interpret and execute programs that realize each function. Furthermore, information such as programs, tables, and files that realize each configuration can be stored in memory, a recording device such as a hard disk or SSD (Solid State Drive), or a recording medium such as an IC card, SD card, or DVD. [Explanation of Symbols]

[0058] 1: Normal highest priority countermeasure selection system, 11: Processor, 12: Memory, 13: Storage unit

Claims

1. A risk analysis method performed by a risk analysis system that conducts risk analysis to reduce risks that occur in emergencies, The processor of the aforementioned risk analysis system A predetermined function that outputs a predetermined value for an input is defined as a predetermined formula that includes a plurality of explanatory factors, obtained by multiplying the input of each explanatory factor by a first contribution for each explanatory factor, as a first-order term or factor, thereby defining a plurality of normal-time objective functions that represent the objective to be achieved under normal circumstances. An emergency objective function, which is an objective function that represents the objective to be achieved in order to reduce the aforementioned risk, is defined as a function that takes a predetermined formula as input and outputs a value, which includes a plurality of normal-time objective functions obtained by multiplying each of the normal-time objective functions by a second contribution, as a first-order term or factor. The second contribution to the emergency objective function is output by machine learning of the second sample data related to the emergency objective function, and a preferred normal objective function that should be optimized by prioritizing it over the normal objective function is extracted based on the second contribution. A risk analysis method characterized by having each of the following processes.

2. A risk analysis method according to claim 1, The aforementioned processor, The first contribution in the preferred normal-time objective function is output by machine learning of the first sample data related to the preferred normal-time objective function, and a predetermined number of explanatory factors corresponding to the first contribution are extracted in descending order of the first contribution. A risk analysis method characterized by having a processing step.

3. A risk analysis method according to claim 2, The aforementioned processor, The aforementioned explanatory factors and measures to reduce the aforementioned risks are managed in correspondence, Output the countermeasures corresponding to the extracted explanatory factors. A risk analysis method characterized by having each of the following processes.

4. A risk analysis method according to claim 2, The aforementioned processor, The first contribution is output for each sample of the first sample data using machine learning. A risk analysis method characterized by the following features.

5. A risk analysis method according to claim 3, The aforementioned processor, Along with the aforementioned countermeasures, the priority order of the countermeasures corresponding to the explanatory factors according to the magnitude of the first contribution of the predetermined number is output. A risk analysis method characterized by the following features.

6. A risk analysis method according to claim 3, The aforementioned risk is the progression to severe illness during an infectious disease outbreak. Each of the aforementioned explanatory factors is data relating to the health status of each subject. Each of the aforementioned normal-condition objective functions represents the relationship between the disease prevalence of each subject and each of the aforementioned explanatory factors. The aforementioned emergency objective function represents the relationship between the severity of the infectious disease during the infectious disease outbreak and each of the aforementioned normal-time objective functions. A risk analysis method characterized by the following features.

7. A risk analysis method according to claim 3, The aforementioned risk is the delay in product delivery during an earthquake. Each of the aforementioned explanatory factors is data relating to the operations of each factory that produces the components constituting each of the aforementioned products. Each of the above-mentioned normal-time objective functions represents the relationship between the number of days of delay in delivery of each of the above-mentioned parts supplied by each of the above-mentioned factories and the explanatory factor, The aforementioned emergency objective function represents the relationship between the number of days of delay in the delivery of the product during the earthquake and each of the aforementioned normal-time objective functions. A risk analysis method characterized by the following features.

8. A risk analysis system that performs risk analysis to reduce the risks that occur in emergencies, The processor of the aforementioned risk analysis system is A predetermined function that outputs a predetermined value for an input is defined as a predetermined formula that includes a plurality of explanatory factors, obtained by multiplying the input of each explanatory factor by a first contribution for each explanatory factor, as a first-order term or factor, thereby defining a plurality of normal-time objective functions that represent the objective to be achieved under normal circumstances. An emergency objective function, which is an objective function that represents the objective to be achieved in order to reduce the aforementioned risk, is defined as a function that takes a predetermined formula as input and outputs a value, which includes a plurality of normal-time objective functions obtained by multiplying each of the normal-time objective functions by a second contribution, as a first-order term or factor. The second contribution to the emergency objective function is output by machine learning of the second sample data related to the emergency objective function, and a preferred normal objective function that should be optimized by prioritizing it over the normal objective function is extracted based on the second contribution. A risk analysis system characterized by the following features.

9. A countermeasure selection system according to claim 8, The aforementioned processor, The first contribution in the preferred normal-time objective function is output by machine learning of the first sample data related to the preferred normal-time objective function, and a predetermined number of explanatory factors corresponding to the first contribution are extracted in descending order of the first contribution. A risk analysis system characterized by the following features.

10. A risk analysis system according to claim 9, The aforementioned processor, The aforementioned explanatory factors and measures to reduce the aforementioned risks are managed in correspondence, Output the countermeasures corresponding to the extracted explanatory factors. A risk analysis system characterized by the following features.

11. A risk analysis system according to claim 9, The aforementioned processor, The first contribution is output for each sample of the first sample data using machine learning. A risk analysis system characterized by the following features.

12. A risk analysis system according to claim 10, The aforementioned processor, Along with the aforementioned countermeasures, the priority order of the countermeasures corresponding to the explanatory factors according to the magnitude of the first contribution of the predetermined number is output. A risk analysis system characterized by the following features.

13. A risk analysis system according to claim 10, The aforementioned risk is the progression to severe illness during an infectious disease outbreak. Each of the aforementioned explanatory factors is data relating to the health status of each subject. Each of the aforementioned normal-use objective functions represents the relationship between the prevalence of each disease in each subject and each of the aforementioned explanatory factors. The aforementioned emergency objective function represents the relationship between the severity of the infectious disease during the infectious disease outbreak and each of the aforementioned normal-time objective functions. A risk analysis system characterized by the following features.

14. A risk analysis system according to claim 10, The aforementioned risk is the delay in product delivery during an earthquake. Each of the aforementioned explanatory factors is data relating to the operations of each factory that produces the components constituting each of the aforementioned products. Each of the above-mentioned normal-time objective functions represents the relationship between the number of days of delay in delivery of each of the above-mentioned parts supplied by each of the above-mentioned factories and the explanatory factor, The aforementioned emergency objective function represents the relationship between the number of days of delay in the delivery of the product during the earthquake and each of the aforementioned normal-time objective functions. A risk analysis system characterized by the following features.

Images