JP7579765B2 - Analysis device, analysis method, and analysis program - Google Patents

Description

This disclosure relates to an analysis device, an analysis method, and an analysis program.

Various simulations are used to evaluate the risk of a target event. Examples of events include diseases, infectious diseases, and other events that harm human health. Patent Document 1 describes a healthcare data analysis device that is configured by a computer having a calculation device that executes a predetermined process and a storage device connected to the calculation device, and that uses at least one item of input healthcare data to generate a set of N pieces of data similar to each data, and generates risk analysis data in which a value located at a predetermined ratio from the larger or smaller of another item of healthcare data in the generated set is used as a response variable; a risk prediction model generation unit that uses part or all of the input healthcare data as explanatory variables and generates a risk prediction model that predicts the response variable included in the risk analysis data by learning these data; and a risk prediction unit that inputs healthcare data into the risk prediction model and predicts the risk represented by the response variable.

JP 2020-135489 A

When analyzing the target event, a large number of cases of the target event are required to increase the accuracy of the analysis. However, if it takes time to collect data on a large number of cases, it will take time before a response can begin.

In order to solve the above problems, at least one embodiment of the present disclosure aims to provide an analysis device, an analysis method, and an analysis program that can quickly evaluate an evaluation target with high accuracy.

The present disclosure provides an analysis device that analyzes an evaluation target whose stage is set based on the severity of symptoms and whose stage changes, the analysis device including: a storage unit that stores individual data for a plurality of targets, including the cause of the symptoms, attribute information of the target, and the change in stage of the target; a statistical data creation unit that creates statistical data for each cause using the attribute information and the change in stage; and a prediction unit that creates evaluation statistical data from the individual data of the cause of the evaluation target, compares the evaluation statistical data with the statistical data, and identifies a prediction result from statistical data similar to the evaluation statistical data of the cause of the evaluation target.

The present disclosure provides an analysis method for analyzing an evaluation target whose stage is set based on the severity of symptoms and whose stage changes, the analysis method including a storage step for storing individual data for a plurality of targets, including the cause of the symptoms, attribute information of the target, and the change in stage of the target; a statistical data creation step for evaluating the attribute information and the change in stage for each cause and creating statistical data; and a prediction step for creating evaluation statistical data from the individual data of the cause of the evaluation target, comparing the evaluation statistical data with the statistical data, and identifying causes of the statistical data similar to the evaluation statistical data of the cause of the evaluation target.

The present disclosure provides an analysis method for analyzing an evaluation target whose stage is set based on the severity of symptoms and whose stage changes, and provides an analysis program that causes a computer to execute the following steps: a storage step for storing individual data for a plurality of targets, including the cause of the symptoms, attribute information of the target, and the change in stage of the target; a statistical data creation step for evaluating the attribute information and the change in stage for each cause and creating statistical data; and a prediction step for creating evaluation statistical data from the individual data of the cause of the evaluation target, comparing the evaluation statistical data with the statistical data, and identifying the cause of the statistical data that is similar to the evaluation statistical data of the cause of the evaluation target.

The above configuration has the effect of enabling the evaluation subject to be evaluated quickly and with high accuracy.

FIG. 1 is a block diagram illustrating an example of an analysis device. FIG. 2 is an explanatory diagram illustrating an example of individual data. FIG. 3 is an explanatory diagram illustrating an example of statistical data. FIG. 4 is a flowchart illustrating an example of a process for creating virtual statistical data by the analysis device. FIG. 5 is a flowchart illustrating an example of a prediction process performed by the analysis device. FIG. 6 is an explanatory diagram showing an example of the creation criteria of the statistical data. FIG. 7 is an explanatory diagram illustrating an example of statistical data. FIG. 8 is an explanatory diagram showing an example of a prediction method.

Below, an embodiment of the present disclosure will be described in detail with reference to the drawings. Note that the present invention is not limited to this embodiment. Furthermore, the components in the following embodiments include those that are easily replaceable by a person skilled in the art, or those that are substantially the same. Furthermore, the components described below can be combined as appropriate, and when there are multiple embodiments, the respective embodiments can also be combined.

In this embodiment, the phenomenon to be analyzed is the tendency and characteristics of symptom fluctuations when a human is infected with a virus. In this embodiment, symptoms are classified into a plurality of stages based on the severity of the symptoms. In this embodiment, symptoms are classified into an infectious stage, a mild stage, and a severe stage. Examples of the characteristics of the virus (cause) include the risk of moving (transitioning) from each stage to another stage, specifically, the risk of infection, the risk of onset, and the risk of severe illness. Note that this embodiment uses an infectious disease as an example, but is not limited to this. The analysis target is not limited to humans, and can also be applied to, for example, breakdowns and abnormalities in industrial machinery. Note that the analysis target is one in which the severity of symptoms changes over time.

FIG. 1 is a block diagram showing an example of an analysis device. The analysis device 10 according to this embodiment acquires individual data for an analysis target, including information on the cause (type of infected virus), attribute information (attribute data), and stage transition (symptom change data), for individual patients (targets, target events), and analyzes the individual data. The analysis device 10 calculates the characteristics of each virus by performing analysis for each virus. The analysis device 10 also calculates the characteristics of a virtual virus (virtual statistical data) based on the acquired information on the characteristics of the virus (statistical data) and the set conditions. Furthermore, the analysis device 10 compares the individual data of the evaluation target infected with the virus to be evaluated with the statistical data and the virtual statistical data to identify a virus with similar characteristics. The analysis device 10 also estimates the characteristics of the virus to be evaluated based on the statistical data of the virus with similar characteristics.

The analysis device 10 includes an input unit 12, an output unit 14, a communication unit 16, a calculation unit 18, and a storage unit 20. The input unit 12 includes an input device such as a keyboard and mouse, a touch panel, or a microphone that collects speech from an operator, and outputs a signal corresponding to an operation performed by the operator on the input device to the calculation unit 18. The output unit 14 includes a display device such as a display, and displays a screen including various information such as a processing result or an image to be processed based on a display signal output from the calculation unit 18. The output unit 14 may also include a recording device that outputs data on a recording medium. The communication unit 16 transmits data using a communication interface. The communication unit 16 communicates with an external device and sends the various data and programs acquired to the storage unit 20 and stores them. The communication unit 16 may be connected to an external device via a wired communication line or a wireless communication line.

The calculation unit 18 includes integrated circuits (processors) such as a CPU (Central Processing Unit) and a GPU (Graphics Processing Unit), and memory that serves as a working area, and executes various programs using these hardware resources to perform various processes. Specifically, the calculation unit 18 reads out the programs stored in the storage unit 20, expands them in memory, and causes the processor to execute the instructions included in the programs expanded in memory, thereby performing various processes.

The calculation unit 18 includes a statistical processing unit 30, a virtual model creation unit 32, a learning unit 34, and a prediction unit 36. Note that the analysis device 10 may execute processing by connecting the statistical processing unit 30, the model creation unit 32, the learning unit 34, and the prediction unit 36 via a communication network, or may execute processing in a single processing device. The functions of each unit will be described later.

The memory unit 20 is made of a non-volatile memory device such as a magnetic memory device or a semiconductor memory device, and stores various programs and data. The memory unit 20 includes individual data 40, a template data 42, statistical data 44, a processing program 46, and a prediction program 48.

The data stored in the storage unit 20 include individual data 40, template data 42, and statistical data 44. FIG. 2 is an explanatory diagram showing an example of individual data. As shown in FIG. 2, the individual data 40 is data including personal identification data 102, cause identification data 103, attribute data 104 of each person, and symptom transition data 106. The individual data 40 is created for each infection of a human, that is, for each analysis target (case). The individual identification data 102 is data for identifying the individual data 40, and is assigned an ID or the like. The individual identification data 102 only needs to be able to identify the individual data, and does not need to be a name or the like. The cause identification data 103 is the type of virus (cause) with which the subject is infected. In this embodiment, the type of virus is used, but in the case of an infectious disease, bacteria may also be included in addition to viruses. The individual data 102 can include various causes that cause similar symptoms in the subject. The attribute data 104 is information indicating whether the subject of the individual data 40 has a factor that affects the analysis target. In this embodiment, the attribute data 104 is various attribute information that may affect the transition of symptoms of an infectious disease, such as underlying diseases, medical history, age, sex, height, weight, lifestyle, etc. The symptom transition data 106 includes the transition of symptoms of the subject of the individual data, that is, the timing (date) when each stage of the disease progressed from asymptomatic to mild to severe.

In this embodiment, each individual data is associated with attribute information and a stage (symptom) transition, but the attribute information and the stage transition may be separate data. For example, information on individual attribute data and data showing symptom transitions for each attribute information may be provided, and the information may not identify the symptom transitions of an individual. Furthermore, data showing stage transitions for each attribute information and data showing overall stage transitions may be used as individual data.

Template data 42 includes data such as templates for individual data, templates for analysis models, templates for statistical data, and templates for displaying analysis results. The templates for individual data include items shown in FIG. 2 and items for inputting data on symptom transitions, specifically, information on the transitions of symptoms and the dates and times when the changes occurred. Template data 42 also includes information for creating a model for hypothetical causes.

The statistical data 44 is data created by processing the individual data in the statistical processing unit 30. The statistical data 44 is data that shows the stage transition statistics for each attribute by performing statistical processing for each attribute of the individual data 40 on the transition model created by the model creation unit 32. The statistical data 44 is data created from items that can be used for comparison processing in the prediction unit 36. The statistical data 44 is data that shows the probability of symptom transition for each attribute based on a reference date (in this embodiment, the date of infection). The statistical data 44 includes data (actual statistical data) obtained from real causes created based on the individual data 40, and data (virtual statistical data) obtained from virtual causes created based on virtual individual data.

FIG. 3 is an explanatory diagram showing an example of statistical data 30. As shown in FIG. 3, statistical data 30 is data in which the risk of developing each of stages 110, 112, and 114 is calculated for each combination of cause (virus) and attribute (disease). Stage 110 is data on infection risk, that is, a statistical value of the risk of a person with each disease being infected with each virus. Stage 112 is data on onset risk, that is, a statistical value of the risk of a person with each disease and infected with each virus developing the disease. Stage 114 is data on aggravation risk, that is, a statistical value of the risk of a person with each disease and developing symptoms caused by each virus becoming severe. As shown in FIG. 3, in stage 112, information is summarized by virus type in groups 120, 122, and 124. Group 120 includes risk values of each disease 130, 132, and 134 causing stage 110 (infection) in the case of virus A. Group 122 includes risk values of each disease causing stage 110 (infection) in the case of virus B. Group 124 contains similar information in the case of virus M. Note that groups are provided for each stage, the number of groups being equal to the number of viruses. As described above, statistical data 30 is aggregated based on individual data and virtual individual data, and is data in which the risk value of causing a transition from each stage to another stage is aggregated for each disease, virus, and stage.

The programs stored in the storage unit 20 include a processing program 46 and a prediction program 48. In the analysis device 10 of this embodiment, the processing program 46 and the prediction program 48 form the analysis program. Note that in the case of an analysis that does not perform prediction processing, the processing program 46 forms the analysis program.

The processing program 46 is a program executed by the statistical processing unit 30, the model creation unit 32, and the learning unit 34. The processing program 46 executes a process of creating statistical data from individual data, a process of creating virtual individual data, a process of creating virtual statistical data from the virtual individual data, and a process of learning a method of estimating the characteristics of a cause based on the statistical data and the virtual statistical data.

The prediction program 48 is a program executed by the prediction unit 36. The prediction program 48 evaluates the characteristics of the cause of the input individual data based on the estimation method created by the processing program 46 and the individual data to be evaluated.

The storage unit 20 may install the processing program 46 and the prediction program 48 by reading the processing program 46 and the prediction program 48 recorded on a recording medium, or may install the processing program 46 and the prediction program 48 by reading the processing program 46 and the prediction program 48 provided over a network.

The functions of each part of the calculation unit 18 will be described. Each part of the calculation unit 18 can be executed by executing a program stored in the storage unit 20. The statistical processing unit 30 reads out the individual data 40 and creates data that associates the risk of moving to each stage with each cause and attribute information. The risk may be calculated as a dimensionless number with 0 as the minimum and 1 as the maximum. The risk may be a single numerical value, or a function with time as a variable. In other words, a risk value for moving to the next stage may be set for each elapsed time from a reference point (the time when the stage was reached).

The virtual model creation unit 32 creates virtual individual data by setting virtual causes and setting the characteristics of the causes. The virtual model creation unit 32 creates virtual personal data by setting the risk of disease and symptom progression based on individual data 40 acquired from actual infectious diseases and performing random sampling using the Monte Carlo method based on the settings. In addition, the method of creating virtual personal data is not limited to random sampling, and various methods of generating data can be used. In addition, the virtual model creation unit 32 may create virtual statistical data without creating virtual individual data.

The learning unit 34 analyzes statistical data including real statistical data and virtual statistical data, and generates an algorithm for creating virtual individual data of a new cause to be evaluated from the individual data of the new cause to be evaluated. The algorithm is a processing rule for setting disease conditions of the virtual individual data by random sampling such as the Monte Carlo method, and further calculating changes in symptoms. The learning unit 34 may execute a process for setting criteria for creation based on the real statistical data and virtual statistical data, or a process for creating virtual individual data by machine learning.

The prediction unit 36 predicts the characteristics of the individual data of the new cause to be evaluated using the similarity determination process calculated by the learning unit 34. The prediction unit 36 identifies causes that have characteristics similar to those of the new cause to be evaluated. The prediction unit 36 predicts the relationship between the attribute information and the stage transition of the new cause to be evaluated based on the attribute information stored for the causes that have similar characteristics and the stage transition.

Next, the processing of the analysis device will be described with reference to Figures 4 and 5. Figure 4 is a flowchart showing an example of the processing of the analysis device. Figure 5 is a flowchart showing an example of the prediction processing of the analysis device.

An example of the process will be described with reference to FIG. 4. The process shown in FIG. 4 is realized by executing operations in each part of the analysis device 10. The analysis device 10 acquires individual data (step S12). Specifically, the analysis device 10 reads the individual data 40 from the storage unit 20. The analysis device 10 classifies the individual data by cause (step S14).

The analysis device 10 creates actual statistical data for each cause (step 16). For each disease, the analysis device 10 extracts individual data of people with the disease and extracts information on symptom transitions. Here, if the individual data is data on a person with multiple diseases, it is extracted for each of the multiple diseases. The analysis device 10 calculates a value for the risk of the symptom, that is, the possibility that the symptom will worsen, based on the statistical information on the symptom transitions for each disease. By performing risk calculations for each cause, symptom, and disease, the analysis device 10 calculates a risk value set for each cause, symptom, and disease.

The analysis device 10 sets the conditions of the virtual model (step S18). Specifically, the analysis device 10 sets a virtual virus and sets the characteristics of the virus. The characteristics of the virus can be created by setting a trend based on the individual data 40 and the statistical data 44 and adjusting the characteristics of the existing virus. The conditions of the virtual model define the characteristics for a disease, etc.

The analysis device 10 creates virtual individual data using the Monte Carlo method (step S20). The analysis device 10 uses the Monte Carlo method to randomly sample the presence or absence of a disease, and creates virtual individual data that estimates the transition of symptoms of an infected person based on the results and the conditions set in step S18. The analysis device 10 creates virtual statistical data based on the virtual individual data (step S22). The virtual statistical data can be created using the same process as for real statistical data.

The analysis device 10 determines whether the termination condition has been met (step S24). The termination condition may be various conditions. For example, the termination condition may be whether or not a set number of virtual models have been analyzed. If the analysis device 10 determines that the termination condition has not been met (No in step S24), the analysis device 10 returns to step S18. The analysis device 10 executes the processes from step S18 to step S22 for a different virtual model, and creates different virtual statistical data.

When the analysis device 10 determines that the termination condition has been met (Yes in step S24), that is, when virtual statistical data has been created for the set virtual model, the analysis device 10 ends the process. Through the above process, the analysis device 10 creates statistical data including real statistical data and virtual statistical data.

Next, an analysis method using the data created in FIG. 4 will be described with reference to FIG. 5. FIG. 5 is a flow chart showing an example of a prediction process of the analysis device. The process shown in FIG. 5 is a process of analyzing individual data of the cause of the evaluation object using statistical data and predicting the characteristics of the cause of the evaluation object. In the process shown in FIG. 5, each step is executed by the prediction unit 36. The prediction unit 36 reads out the individual data of the evaluation object (step S32). The individual data of the object for which it is desired to estimate the progression of symptoms caused by the cause is obtained.

Next, the prediction unit 36 creates virtual individual data for the evaluation target (step S34). The prediction unit 36 inputs the individual data of the evaluation target to the learning unit 34, executes processing, and creates virtual individual data. This makes it possible to increase the individual data for analyzing the causes of the evaluation target.

The prediction unit 36 creates virtual statistical data for the evaluation target (step S36). The prediction unit 36 creates virtual statistical data for the evaluation target using the individual data for the evaluation target and the virtual individual data for the evaluation target.

The prediction unit 36 compares the statistical data and virtual statistical data for each cause with the virtual statistical data of the evaluation target, and identifies causes that produce data similar to the virtual statistical data of the evaluation target (step S38). The prediction unit 36 extracts information on the distribution of diseases contained in the virtual statistical data of the evaluation target, and calculates trends for the statistical data and virtual statistical data for each cause based on the extracted information on the distribution of diseases. The prediction unit 36 compares the statistical data and virtual statistical data for each cause that produce the same disease distribution with the virtual statistical data of the evaluation target, and identifies causes that produce similar trends.

The prediction unit 36 outputs a prediction result based on the statistical data or virtual statistical data of the identified cause (step S40). The prediction result is output from the output unit 14. Here, the prediction result also includes information on stage transitions for diseases that are not included in the virtual statistical data to be evaluated.

The analysis device 10 creates actual statistical data corresponding to the cause, and creates evaluation statistical data for the cause of the evaluation target based on the actual statistical data of the evaluation target. Furthermore, the analysis device 10 compares the value of the evaluation statistical data of the cause of the evaluation target with the value of each cause of the statistical data, and identifies causes of the statistical data that are similar to the evaluation statistical data of the cause of the evaluation target. In this way, the analysis device 10 creates actual statistical data corresponding to various causes, compares the evaluation statistical data of the evaluation target with the actual statistical data, and identifies statistical data that is similar to the cause of the evaluation target, thereby being able to appropriately evaluate the risk of the cause of the evaluation target.

The analysis device 10 preferably generates virtual statistical data for a virtual cause as in this embodiment. In this case, the analysis device 10 generates real statistical data corresponding to the cause and virtual statistical data for the virtual cause, and generates evaluation statistical data for the cause of the evaluation target based on the real statistical data of the evaluation target and the virtual statistical data of the evaluation target. The evaluation statistical data of this embodiment is virtual evaluation statistical data calculated including the real statistical data and the virtual statistical data. Furthermore, the analysis device 10 compares the value of the evaluation statistical data of the cause of the evaluation target with the value of each cause of the statistical data to identify a cause of the statistical data similar to the evaluation statistical data of the cause of the evaluation target. By comparing the statistical data for each cause and the virtual statistical data for each cause with the evaluation statistical data of the evaluation target, the analysis device 10 can appropriately evaluate the risk of the cause of the evaluation target even if there is little individual data of the cause of the evaluation target. By generating virtual statistical data for a virtual cause, the analysis device 10 can increase the number of objects to be compared with the virtual statistical data of the cause of the evaluation target, and can detect similar causes with high accuracy.

In addition, in the above embodiment, the tendency of stage transition for the cause of the evaluation target is predicted for each attribute information, and the prediction result is output, so that the cause of the evaluation target can be predicted in more detail, but causes that are determined to be similar to the cause of the evaluation target may also be output.

The analysis device 10 may also create virtual evaluation individual data by interpolating the individual data of the cause of the evaluation target. Using the individual data and the evaluation individual data to create virtual statistical data for the cause of the evaluation target, it is possible to make the comparison process easier. The analysis device 10 may also create evaluation statistical data using only the individual data of the cause of the evaluation target, without creating virtual evaluation individual data.

The analysis device 10 uses statistical data including real statistical data corresponding to the cause and virtual statistical data for the virtual cause to create virtual individual data for the cause of the evaluation target from the individual data for the cause of the evaluation target, making it possible to create virtual statistical data that is easy to compare even when there is little individual data for the cause of the evaluation target. In addition, by creating virtual individual data for the cause of the evaluation target using statistical data corresponding to the cause and virtual statistical data for the virtual cause, it is possible to create highly accurate virtual individual data. Note that while it is preferable to use virtual individual data because this increases the accuracy of the comparison, statistical data may be created using only the individual data for the cause of the evaluation target, and comparison may be performed.

The analysis device 10 classifies the statistical data and virtual statistical data by cause, and further calculates the probability of each symptom progressing to a more serious symptom for each disease. This allows the risk for each disease to be used as an evaluation value, and when comparing statistical data, it is possible to evaluate diseases (attributes) that have an influence on the cause. This makes it possible to confirm factors that have an influence on the cause, and to predict with high accuracy the relationship between the disease and risk for the cause being evaluated, that is, the influence of attribute information.

The analysis device 10 can create a large number of virtual individual data by creating virtual individual data through random sampling such as the Monte Carlo method, and then creating virtual statistical data. The analysis device 10 of the above embodiment can obtain the above-mentioned effects by creating virtual statistical data for virtual causes in the virtual model creation unit 32, but if there is sufficient statistical data, i.e., real statistical data, for a large number of causes, the processing can be performed without creating virtual statistical data.

In addition, the analysis device 10 can analyze symptom transitions with greater accuracy by setting the risk value as a function of the time axis.

The analysis device 10 may also apply different risk values for different items to the same cause depending on the time period, based on the overall time axis, not on the time axis from onset. FIG. 6 is an explanatory diagram showing an example of the creation criteria for statistical data. FIG. 7 is an explanatory diagram showing an example of statistical data. In other words, multiple risk values may be set for one disease, with different risk values being applied to the disease depending on the overall time axis, that is, the period of onset. For example, the individual data generated in each of the periods 140, 142, and 144 shown in FIG. 6 may be classified as separate statistical data, and statistical data for period 140, period 142, and period 144 may be created for one cause. The analysis device 10 creates statistical data for the period 140, the period 142, and the period 144 for each cause, and as shown in FIG. 7, can create statistical data that calculates the occurrence probability for each cause, disease, and stage for the period 140, statistical data that calculates the occurrence probability for each cause, disease, and stage for the period 142, and statistical data that calculates the occurrence probability for each cause, disease, and stage for the period 144. In this case, since the statistical data is for one cause, the comparison is performed for each period when linking related information and comparing the statistical data. This makes it possible to take into account characteristics that change depending on the period. Note that when comparing periods, the starting point may be the timing when the cause occurred, or a specific date in the calendar (season), such as January 1st. This allows for more accurate analysis. For example, it is also possible to analyze seasonal trends including the overall time axis and seasonal factors.

Statistical data may also be created using the timing of symptom stages, that is, the timing of stage transitions, as the time axis standard. For example, using the day of infection as the standard, separate statistical data may be created for days 1 to 5, days 6 to 10, and days 11 and beyond.

When comparing virtual statistical data for the cause to be evaluated, the prediction unit 36 may set an acceptable range (margin) for the risk value of each disease of the virtual statistical data for the cause to be evaluated, and evaluate the agreement of the values of the statistical data and the virtual statistical data. FIG. 8 is an explanatory diagram showing an example of a prediction method. As shown in FIG. 8, the prediction unit 36 sets an acceptable value 150 for each disease using the evaluation statistical data. The prediction unit 36 extracts real statistical data or virtual statistical data for one cause selected from the statistical data, and compares the risk value 152 of each disease of the extracted data with the acceptable range 150 of the cause to be evaluated. In the example shown in FIG. 8, there are many risk values 152 that are not included in the acceptable range 150. For this reason, the prediction unit 36 determines that the statistical data or virtual statistical data for the selected one cause are not similar.

The prediction unit 36 performs a comparison process for each cause, thereby enabling it to identify the most similar cause with high accuracy. By being able to identify similar causes, it is possible to estimate with high accuracy the characteristics of the effects on diseases whose characteristics are unknown with a small amount of individual data, enabling more appropriate responses to be taken. Furthermore, when separating statistical data on the time axis, the prediction unit 36 can make comparisons on each time axis, enabling it to respond appropriately even when characteristics change over time.

REFERENCE SIGNS LIST 10 Analysis device 12 Input unit 14 Output unit 16 Communication unit 18 Calculation unit 20 Storage unit 30 Statistical processing unit 32 Virtual model creation unit 34 Learning unit 36 Prediction unit 40 Individual data 42 Template data 44 Statistical data 46 Processing program 48 Prediction program 102 Personal identification data 103 Cause identification data 104 Attribute data 106 Symptom transition data 110, 112, 114 Stage 120, 122, 124 Group 130, 132, 134 Item

Claims (13)

An analysis device for analyzing an evaluation subject in which a stage is set based on the severity of a symptom and the stage changes, comprising:
A storage unit that stores individual data for a plurality of subjects, the individual data including a cause of the symptom, attribute information of the subject, and a change in stage of the subject;
a statistical data creation unit that creates statistical data using attribute information and stage changes for each cause of the symptom ;
Creating evaluation statistics from individual data on the causes of the symptoms to be evaluated;
comparing the evaluation statistical data with the statistical data of a cause of a symptom other than the evaluation target , identifying a cause of the symptom having a similar trend of stage transition in the evaluation statistical data of the cause of the symptom of the evaluation target, and extracting statistical data of the cause of the symptom having the similar identified evaluation statistical data;
and a prediction unit that identifies a prediction result including information of stage transitions not included in the statistical data of a cause of the symptom to be evaluated, based on a trend of stage transitions in the statistical data of a cause of the symptom to which the identified evaluation statistical data is similar. the statistical data creation unit includes a virtual model creation unit that sets a trend of stage transition of a virtual cause based on the statistical data and the individual data , creates a plurality of virtual individual data including the virtual cause, attribute information, and stage fluctuations based on a result of random sampling of the presence or absence of symptoms of the virtual cause and the trend of the virtual cause, and creates virtual statistical data of the virtual cause from the virtual individual data ;
The analysis device according to claim 1 , wherein the prediction unit includes the virtual statistical data in the statistical data. The analysis device according to claim 1 or 2, wherein the prediction unit predicts a relationship between attribute information regarding the cause of the symptom of the evaluation subject and a transition of stage change based on statistical data of the identified cause of the symptom . The analysis device according to claim 1 , wherein the statistical data is data in which a probability of occurrence of a transition from each stage to another stage is set for each piece of attribute information . The individual data includes information on a change in stage over time from a reference point in time,
The statistical data creation unit divides a time axis into a plurality of time periods and creates statistical data for each time period;
3 . The analysis device according to claim 1 , wherein the prediction unit also uses information on the time axis as a comparison target to identify a cause of statistical data similar to evaluation statistical data of a cause of the symptom of the evaluation target. The analysis device according to claim 5, wherein the reference time is the time when the stage changes. The analysis device according to claim 5, wherein the reference time is a set date. 8. The analysis device according to claim 1, wherein the prediction unit interpolates individual data of the cause of the symptom of the object to be evaluated to create virtual individual data of the cause of the symptom of the object to be evaluated, creates evaluation statistical data of the cause of the symptom of the object to be evaluated based on the individual data of the cause of the symptom of the object to be evaluated and the virtual individual data of the cause of the symptom of the object to be evaluated, and compares a value of the virtual statistical data of the cause of the symptom of the object to be evaluated with a value of each cause of the symptom in the statistical data to identify a cause of the symptom of the object to be evaluated that is similar to the evaluation statistical data of the cause of the symptom of the object to be evaluated. 3. The analysis device according to claim 2, wherein the prediction unit sets a range of risk values of stage fluctuations of virtual statistical data of the causes of the symptoms of the evaluation target, and compares the range of risk values of the virtual statistical data of the causes of the symptoms of the evaluation target with the risk values of each cause of the symptoms of the statistical data. the prediction unit generates a plurality of virtual individual data of the cause of the symptom of the evaluation subject by random sampling using the individual data of the cause of the evaluation subject and the statistical data;
10. The analysis device according to claim 8, further comprising: a processor configured to generate virtual statistical data on the cause of the symptom of the subject to be evaluated, using the individual data on the cause of the symptom of the subject to be evaluated and the virtual individual data. the subject is a human;
The analysis device according to claim 1 , wherein the cause is an infectious disease caused by a virus. A method for analyzing an evaluation subject in which a stage is set based on the severity of symptoms and the stage fluctuates, comprising:
A storage step in which the analysis device stores individual data for a plurality of subjects, the individual data including the cause of the symptom, attribute information of the subject, and a change in stage of the subject;
a statistical data creation step in which the analysis device creates statistical data in which attribute information and a change in stage are associated with each cause of the symptom;
the analysis device creates evaluation statistical data from individual data on the cause of the symptom to be evaluated, compares the evaluation statistical data with the statistical data on the cause of symptoms other than the target to be evaluated , identifies a cause of the symptom having a similar trend in stage transition of the evaluation statistical data to the evaluation statistical data on the cause of the symptom to be evaluated, and extracts statistical data on the cause of the symptom having the similar identified evaluation statistical data;
and a prediction step of identifying a prediction result including information of stage transitions not included in the statistical data of the cause of the symptom to be evaluated, based on the trend of stage transitions in the statistical data of the cause of the symptom to which the identified evaluation statistical data is similar. An analysis program for analyzing an evaluation subject whose stage is set based on the severity of symptoms and whose stage changes,
a storage step of storing individual data for a plurality of subjects, the individual data including a cause of the symptom, attribute information of the subject, and a change in stage of the subject;
a statistical data creation step of creating statistical data in which attribute information and a change in stage are associated with each cause of the symptom ;
creating evaluation statistical data from individual data of the cause of the symptom to be evaluated, comparing the evaluation statistical data with the statistical data of the cause of the symptom other than the evaluation target , identifying a cause of the symptom having a similar stage transition trend to the evaluation statistical data of the cause of the symptom to be evaluated, and extracting statistical data of the cause of the symptom having the similar identified evaluation statistical data;
and a prediction step of identifying a prediction result including information of stage transitions not included in the statistical data of the cause of the symptom to be evaluated, based on the trend of stage transitions in the statistical data of the cause of the symptom to be evaluated that is similar to the identified evaluation statistical data.

Images