JP7531331B2 - Prediction device, prediction method, and program - Google Patents

Description

The present invention relates to a prediction device, a prediction method, and a program.

When an infectious disease spreads, the current situation and future predictions are made in order to take measures. For example, Patent Document 1 discloses a method for grasping the infection situation within a hospital. Patent Document 2 discloses an infectious disease information display system that presents information useful for verifying the source and route of infection based on schools, commercial facilities, hospitals, transportation facilities, and other potential infection routes contained in map information of the location where an infectious disease patient has occurred. Patent Document 3 discloses an infectious disease prediction system that predicts the number of infectious disease patients based on reports of the number of patients who visited outpatient clinics with infectious diseases.

Japanese Patent Application Publication No. 7-79929 International Publication No. 2002/061647 JP 2011-128935 A

Countermeasures against infectious diseases require the arrangement of various resources, such as hospital beds, testing equipment, and transportation to transport patients to hospital beds. To do so, it is necessary to predict the demand for resources necessary for infectious disease countermeasures, but no method for doing so has been provided. Furthermore, Patent Documents 1 to 3 do not disclose a method for predicting the demand for resources necessary for infectious disease countermeasures.

Therefore, the purpose of this disclosure is to provide a prediction device, a prediction method, and a program that can solve the above-mentioned problems.

According to one aspect of the present disclosure, a prediction device calculates x for a plurality of combinations of a and b based on a relational equation between sensitivity indicating a proportion of true positives among people who have been tested and are determined to be positive, specificity indicating a proportion of true negatives among people who have been tested and are not actually infected, an apparent infection rate indicating a probability of people who have been tested being determined to be positive, and a true infection rate, where true positives are defined as people who have been tested and are determined to be negative, and true negatives are defined as people who have been tested and are not actually infected, and calculates x for a plurality of combinations of a and b based on x = (r - B) / (a - B) (where B = 1 - b, x is the true infection rate, a is the sensitivity, b is the specificity, and r is the apparent infection rate), and multiplies each of the calculated plurality of x by the number of people who have been tested to estimate a plurality of true infected numbers, and a test accuracy estimation unit that estimates the sensitivity and the specificity by multiplying the probability that symptoms of the infectious disease will progress to symptoms of the next level, which has been analyzed in advance based on clinical data, by the probability that symptoms of the infectious disease will progress to symptoms of the next level, and calculating an estimate of the number of infected people that will progress to the next level of symptoms, and calculating the number of hospital beds required by the number of infected people related to the estimated number of infected people after the required number of days, based on the estimated number of infected people and the number of days required for symptoms to progress to the next level, which has been analyzed in advance based on the clinical data, and comparing the number of hospital beds calculated for each of the multiple x's with an actual value of the number of hospital beds actually required after the required number of days, and specifying the true infection rate that is the x that is closest to the actual value , and and an infection status estimation unit that estimates the infection status of the prediction target area based on the predicted value of the number of test recipients ; and a simulator that simulates the progression of symptoms to the next level after the required number of days for each of the true positive individuals, false positive individuals indicating people who have been determined to be positive and are not actually infected, the true negative individuals, and false negative individuals indicating people who have been determined to be negative and are actually infected, and allocates the number and type of resources necessary for maintaining sanitation functions to the people whose symptoms have progressed to the next level; and a demand prediction unit that predicts a demand for the resources based on the estimated result of the infection status and the simulator , wherein the infection status indicates the number of true positive individuals, the number of false positive individuals, the number of true negative individuals, and the number of false negative individuals, and when the infection status estimation unit estimates the infection status based on the true infection rate and the number of test recipients, the number of true positive individuals is multiplied by the number of test recipients to calculate the number of true positive individuals. the number of true positives is subtracted from the number of people judged to be positive in the test to calculate the number of false positives; the number of people judged to be negative in the test is multiplied by the specificity to calculate the number of true negatives; the sum of the number of true positives, the number of false positives, and the number of true negatives is subtracted from the number of people taking the test to calculate the number of false negatives; when estimating the infection status based on the true infection rate and the predicted value of the number of people taking the test, the true infection rate is multiplied by the predicted value of the number of people taking the test to calculate the number of true positives in the future; the number of true positives is subtracted from the predicted value of the number of people judged to be positive in the test to calculate the number of false positives in the future; the predicted value of the number of people judged to be negative in the test is multiplied by the specificity to calculate the number of true negatives in the future; and the sum of the future number of true positives, the future number of false positives, and the future number of true negatives is subtracted from the predicted value of the number of people taking the test to calculate the number of false negatives in the future.

According to one aspect of the present disclosure, there is provided a prediction method executed by a computer, the prediction method comprising the steps of: calculating x for a plurality of combinations of a and b based on a relational equation x=(r-B)/(a-B) (where B=1-b, x is the true infection rate, a is the sensitivity, b is the specificity, and r is the apparent infection rate) between sensitivity indicating a proportion of true positives among those who have been tested and are actually infected, specificity indicating a proportion of true negatives among those who have not actually been infected, apparent infection rate indicating a probability of those who have been tested being determined to be positive, and true infection rate, where true positives are those who have been tested and are actually infected among those who have been tested and true negatives are those who have been tested and are true negatives; and multiplying each of the calculated plurality of x by the number of people who have been tested to obtain a true infection rate. a step of estimating a number of infected persons, multiplying each of the estimated true numbers of infected persons by a probability that symptoms of the infectious disease will progress to symptoms of the next level, which is analyzed in advance based on clinical data, to calculate an estimate of the number of infected persons that will progress to symptoms of the next level, calculating the number of hospital beds required by the number of infected persons related to the estimated number of infected persons after the required number of days, based on the estimated number of infected persons and the number of days required for symptoms to progress to the next level, which is analyzed in advance based on the clinical data, comparing the number of hospital beds calculated for each of the multiple x's with an actual value of the number of hospital beds actually required after the required number of days, and specifying the true infection rate that is x when it is closest to the actual value, thereby estimating the sensitivity and the specificity; or a predicted value of the number of examinees; a simulator that simulates the progression of symptoms to the next level after the required number of days for each of the true positives, false positives indicating people who have been determined to be positive and are not actually infected, the true negatives, and false negatives indicating people who have been determined to be negative and are actually infected, and allocates the number and type of resources required to maintain sanitation functions to the people whose symptoms have progressed to the next level; and a step of predicting the demand for resources based on the estimated result of the infection status, wherein the infection status indicates the number of true positives, the number of false positives, the number of true negatives, and the number of false negatives; and in the step of estimating the infection status , when the infection status is estimated based on the true infection rate and the number of examinees, the number of true positives is calculated by multiplying the true infection rate by the number of examinees; The number of true positives is subtracted from the number of people judged to be positive in the test to calculate the number of false positives, the number of people judged to be negative in the test is multiplied by the specificity to calculate the number of true negatives, and the sum of the number of true positives, false positives, and true negatives from the number of people taking the test to calculate the number of false negatives. When estimating the infection status based on the true infection rate and the predicted value of the number of people taking the test, the true infection rate is multiplied by the predicted value of the number of people taking the test to calculate the number of future true positives, the number of true positives is subtracted from the predicted value of the number of people judged to be positive in the test to calculate the number of future false positives, the predicted value of the number of people judged to be negative in the test is multiplied by the specificity to calculate the number of future true negatives, and the sum of the future number of true positives, future number of false positives, and future number of true negatives is subtracted from the predicted value of the number of people taking the test to calculate the number of future false negatives.

According to one aspect of the present disclosure, the program causes a computer to calculate x for a plurality of combinations of a and b based on a relational equation x=(r-B)/(a-B) (where B=1-b, x is the true infection rate, a is the sensitivity, b is the specificity , and r is the apparent infection rate) between sensitivity indicating the proportion of true positives among those who have been tested and are actually infected, specificity indicating the proportion of true negatives among those who have not actually been infected, apparent infection rate indicating the probability that those who have been tested will be determined to be positive, and true infection rate, where true positives are those who have been tested and are determined to be negative are those who are actually infected, and true negatives are those who have been tested and are determined to be negative, and calculates a plurality of x for each of the calculated x by the number of people who have been tested to estimate the number of true infected people, a step of multiplying the estimated true number of infected persons by a probability that symptoms of the infectious disease will progress to symptoms of the next level, which is analyzed in advance based on clinical data, to calculate an estimate of the number of infected persons that will progress to symptoms of the next level, and calculating the number of hospital beds required by the number of infected persons related to the estimated number of infected persons after the required number of days, based on the estimated number of infected persons and the number of days required for symptoms to progress to the next level, which is analyzed in advance based on the clinical data, and comparing the number of hospital beds calculated for each of the multiple x's with an actual value of the number of hospital beds actually required after the required number of days, and specifying the true infection rate that is x when it is closest to the actual value, thereby estimating the sensitivity and the specificity; a simulator that simulates the progression of symptoms to the next level after the required number of days for each of the true positives, false positives indicating people who have been determined to be positive but are not actually infected, the true negatives, and false negatives indicating people who have been determined to be negative and are actually infected, and allocates the number and type of resources required to maintain sanitation functions to the people whose symptoms have progressed to the next level; and a step of predicting the demand for resources based on the estimated result of the infection status, wherein the infection status indicates the number of true positives, the number of false positives, the number of true negatives, and the number of false negatives, and in the step of estimating the infection status, when the infection status is estimated based on the true infection rate and the number of people who have been tested, the number of true positives is calculated by multiplying the true infection rate by the number of people who have been tested, and the number of people who have been tested who have been positive is calculated by multiplying the true infection rate by the number of people who have been tested. the number of true positives is subtracted from the number of people judged to be positive in the test to calculate the number of false positives; the number of people judged to be negative in the test is multiplied by the specificity to calculate the number of true negatives; the sum of the number of true positives, the number of false positives, and the number of true negatives is subtracted from the number of people who underwent the test to calculate the number of false negatives; when estimating the infection status based on the true infection rate and the predicted value of the number of people who underwent the test, the true infection rate is multiplied by the predicted value of the number of people who underwent the test to calculate the number of future true positives; the number of true positives is subtracted from the predicted value of the number of people who will be judged to be positive in the test to calculate the number of future false positives; the predicted value of the number of people who will be judged to be negative in the test is multiplied by the specificity to calculate the number of future true negatives; and the sum of the future number of true positives, the future number of false positives, and the future number of true negatives is subtracted from the predicted value of the number of people who will be judged to be negative in the test to calculate the number of future false negatives.

This disclosure makes it possible to predict the demand for resources required to maintain hygiene functions.

FIG. 1 is a block diagram illustrating an example of a prediction device according to an embodiment. FIG. 13 is a diagram illustrating an example of clinical data according to an embodiment. FIG. 1 is a diagram showing an example of progression of symptoms in one embodiment. FIG. 1 is a first diagram illustrating a method for estimating a community infection rate in one embodiment. FIG. 2 is a second diagram illustrating a method for estimating a community infection rate in one embodiment. FIG. 3 is a third diagram illustrating a method for estimating a community infection rate in one embodiment. FIG. 4 is a fourth diagram explaining a method for estimating a community infection rate in one embodiment. FIG. 13 is a diagram illustrating an example of a hygiene function maintenance resource in one embodiment. FIG. 13 is a diagram illustrating an example of the movement of an asymptomatic person in one embodiment. FIG. 13 is a diagram illustrating an example of the movement of a person with mild symptoms according to an embodiment. FIG. 13 is a diagram illustrating an example of the movement of a person with moderate symptoms according to an embodiment. FIG. 11 is a diagram illustrating an example of the movement of a seriously ill patient according to an embodiment. FIG. 13 is a diagram illustrating an example of the movement of a seriously ill patient according to an embodiment. FIG. 13 is a diagram illustrating an example of a simulation result according to an embodiment. 11 is a flowchart illustrating an example of a demand amount prediction process according to an embodiment. FIG. 2 illustrates an example of a hardware configuration of a prediction device according to an embodiment.

<Embodiment>
Hereinafter, a method for predicting the demand for a sanitary function maintenance resource according to one embodiment of the present disclosure will be described with reference to FIGS.
(composition)
FIG. 1 is a block diagram illustrating an example of a prediction device according to an embodiment.
The prediction device 10 predicts the demand for resources required to maintain sanitation functions in infection control measures (hereinafter referred to as sanitation function maintenance resources). Examples of sanitation function maintenance resources include medical resources such as hospitals, hospital beds, medical equipment, testing equipment, and medical personnel, and transportation means such as ambulances for transporting patients and vehicles for transporting medical resources.

As shown in FIG. 1, the prediction device 10 includes a data acquisition unit 11, a test accuracy estimation unit 12, an infection status estimation unit 13, a demand prediction unit 14, an output unit 15, and a memory unit 16.

The data acquisition unit 11 acquires various data necessary for predicting the demand for sanitary function maintenance resources. For example, the data acquisition unit 11 acquires clinical data on infectious disease treatment, data on the results of clinical tests for infectious diseases, bed occupancy data, and survey data on people who have come into close contact with people who have tested positive.

The test accuracy estimation unit 12 estimates the accuracy of clinical tests such as PCR (polymerase chain reaction) tests and antibody tests. Clinical tests generally contain errors, and for example, a positive result may not actually indicate that a person is infected, and a negative result may actually indicate that a person is infected. A positive result and a person who is actually infected is called a true positive, a positive result and a person who is actually not infected is called a false positive, a negative result and a person who is actually not infected is called a true negative, and a negative result and a person who is actually infected is called a false negative. In addition, the proportion of true positives (true positive rate) among the group of people who have been tested and are actually infected is called the sensitivity of the test, and the proportion of true negatives (true negative rate) among the group of people who are not actually infected is called the specificity. The test accuracy estimation unit 12 estimates the sensitivity and specificity, or at least the specificity, regarding the accuracy of the clinical test.

The infection status estimation unit 13 estimates (including future predictions) the infection status in the prediction target area based on the true infection rate based on the accuracy of the clinical test estimated by the test accuracy estimation unit 12 and the number of people undergoing clinical tests or the predicted value of the number of people undergoing clinical tests in the prediction target area (a specific region such as a prefecture, city, town, or village, or one or more specific hospitals). For example, the infection status estimation unit 13 estimates the current infection rate and number of infected people in the city, predicts the number of people who will come to undergo clinical tests in the future based on the relationship between the trend in the number of infected people in the city and the number of people who actually came to undergo clinical tests, and estimates the future number of infected people (the number of true positives, the number of false positives, the number of true negatives, and the number of false negatives) based on the sensitivity and specificity estimated by the test accuracy estimation unit 12.

The demand prediction unit 14 predicts the demand for sanitary function maintenance resources such as hospital beds, ambulances, and testing equipment based on the number of infected people estimated by the infection status estimation unit 13.
The output unit 15 outputs the specificity, the number of infected people in the city, and the demand for resources for maintaining sanitary functions.
The memory unit 16 stores various data necessary for predicting the demand for sanitary function maintenance resources.

(Estimation of test accuracy)
First, a method for estimating sensitivity and specificity, which indicate the accuracy of a clinical test, will be described.
The data acquisition unit 11 acquires clinical data of infectious disease treatment. The clinical data is performance data showing the characteristics of an infectious disease. An example of the clinical data is shown in FIG. 2A. The clinical data includes schedule information showing the progression of symptoms of a person who has been determined to be positive in a clinical test, for example, the level of symptoms after how many days. For example, as shown in FIG. 2B, an infectious disease progresses in the order of asymptomatic (0), mild (1), moderate (2), severe (3), and severe (4). The numbers in parentheses indicate the level of symptoms. The test accuracy estimation unit 12 calculates the probability P ₀₁ of progression from asymptomatic (0) to mild (1) and the average number of days D ₀₁ required to progress from asymptomatic (0) to mild (1) based on the schedule information showing the progression of symptoms included in the clinical data. For example, if the clinical data includes data on 10 asymptomatic people, 8 of whom progress from asymptomatic (0) to mild (1), and all 8 progress to mild after 2 days, the test accuracy estimation unit 12 calculates P ₀₁ = 80% and D ₀₁ = 2 days. Similarly, the test accuracy estimation unit 12 calculates the probability P ₁₂ of progression from mild (1) to moderate (2), the average number of days required for progression from mild (1) to moderate ( ₂ ), the probability P ₂₃ of progression from moderate (2) to severe (3), the average number of days required for progression from moderate ( ₂ ) to severe (3), the probability P ₃₄ of progression from severe (3) to severe (4), and the average number of days required for progression from severe (3) to severe ( ₄ ).

Next, the relationship between the true infection rate, sensitivity, and specificity will be explained. If the true infection rate, sensitivity, specificity, and apparent infection rate (probability of a positive result in a clinical test) are x, a, b, and r, respectively, the following holds true.
x・a+(1-x)・(1-b)=r...(1)
Solving equation (1) for x gives the following equation:
x=(r-B)/(a-B); B=1-b...(2)

In clinical tests for infectious diseases, there are errors in determining infected individuals as negative and errors in determining non-infected individuals as positive. For example, errors in determining non-infected individuals as positive in the early stages of an epidemic unnecessarily strain medical resources, but because clinical tests in the early stages of an epidemic contain many errors, they are likely to cause errors in the allocation of resources to maintain sanitary functions. In this embodiment, it is assumed that there are errors in clinical tests, and the errors are corrected to estimate the net number of infected individuals and predict the required resources to maintain sanitary functions. Specifically, if accurate sensitivity a and specificity b can be obtained from equation (2), the true infection rate x can be calculated using the positive rate r obtained from the results of the clinical tests. The test accuracy estimation unit 12 estimates sensitivity a and specificity b using the method described below, and estimates the true infection rate x required to predict the demand for resources to maintain sanitary functions using equation (2).

The test accuracy estimation unit 12 sets various values for the sensitivity a and specificity b using the Monte Carlo method or the like, and generates a large number of calculation formulas (2) for the true infection rate x.
The test accuracy estimation unit 12 reads out the data of people who were determined to be positive on M1/D1 from the clinical data and divides the total number by the number of people examined on that day to calculate the apparent infection rate r. The test accuracy estimation unit 12 substitutes the calculated value for each r in the many generated formulas (2) to calculate the true infection rate x for each combination of sensitivity a and specificity b. The test accuracy estimation unit 12 multiplies the total number of people examined in the clinical test on M1/D1 by the true infection rate x to calculate an estimate of the true number of infected people for each of the many generated formulas (2).

The test accuracy estimation unit 12 multiplies the previously calculated _P01 , _P12 , _P23 , and _P34 by the true number of infected people for each formula (2) to calculate the number of people who will progress to the next symptom level for each formula (2) and for each symptom level. Next, the test accuracy estimation unit 12 applies _D01 , _D12 , _D23 , and _D34 to the calculated number of people for each symptom level to calculate the number of people who have developed the disease for each symptom level (0 to 4) for each number of days elapsed since the clinical test was performed.
For example, if the number of generated sensitivity a and specificity b patterns is 100, then 100 sets of subsequent infectious disease progression schedule information for people who were determined to be positive on a certain day (M1 month D1 day) have been generated by the processes up to now. The test accuracy estimation unit 12 performs similar processing for people who were tested on other days and determined to be positive, and who are included in the clinical data.

The data acquisition unit 11 also acquires bed data that indicates the change in the number of occupied hospital beds. The bed data includes, for example, the new and total number of occupied hospital beds and ICUs (intensive care units) per day by hospital. The bed data may include not only hospital beds, but also the number of beds in hotels that accommodate patients with mild symptoms.

The test accuracy estimation unit 12 compares the previously calculated schedule information of the progression of many infectious diseases with the hospital bed data and selects the progression data that matches best. For example, if it is determined that, as the symptom level of an infectious disease progresses from 0 to 4, level 0 is home treatment, level 1 is treatment in a hotel, level 2 is a private room bed in a general hospital, level 3 is a private room bed in a base hospital, and level 4 is treatment in an ICU bed in a base hospital, the test accuracy estimation unit 12 compares the number of people who have developed symptoms at levels 0 to 4 on a daily basis in the calculated schedule information of the progression of many infectious diseases with the actual change in the number of hospital beds indicated by the bed data acquired by the data acquisition unit 11 (the number of new beds in operation at hotels, general hospitals, and base hospitals), and selects the progression schedule information that is closest to the bed data. For example, in the bed data, the number of new beds in operation at the general hospital after two days is 2, and the number of new beds in operation at the base hospital is 2, and in the "progress schedule information 1", the number of people who reach level 2 symptoms after two days is 5, and the number of people who reach level 3 symptoms after two days is 5, and in the "progress schedule information 2", the number of people who reach level 2 symptoms after two days is 2, and the number of people who reach level 3 symptoms after two days is 2, and in the "progress schedule information 3", the number of people who reach level 2 symptoms after two days is 3, and the number of people who reach level 3 symptoms after two days is 1, in this case, the test accuracy estimation unit 12 selects the "progress schedule information 2". The test accuracy estimation unit 12 specifies the sensitivity a and specificity b corresponding to the selected progress schedule information. Generally, the sensitivity a and specificity b (especially the specificity b), which indicate the accuracy of a clinical test, are difficult values to specify, but the sensitivity a and specificity b can be specified by collating with the actual bed data. Since r can be calculated by aggregating daily test data, the true infection rate can be calculated from formula (2) by estimating accurate sensitivity a and specificity b by comparing it with hospital bed data. Note that sensitivity a and specificity b are considered to be constant for the same infectious disease.

As for the sensitivity a, it may be possible to estimate an accurate value as clinical data accumulates with the spread of the infection. In that case, a value estimated by an expert or the like from clinical data may be applied to the sensitivity a, and random numbers may be generated only for the specificity b to generate a large number of patterns, and an accurate specificity b may be estimated by the above process.

(Estimated number of infected people in the city)
Next, we will explain how to estimate the number of infected people in the city.
When a clinical test for an infectious disease shows a positive result and the infection is subsequently confirmed, an investigation is conducted of close contacts of the infected person. In this investigation, cases are often divided into two categories: those with a zero or very low infection rate, and those occurring in clusters. In this case, cases with a zero or very low infection rate are considered to be close to community infection, and the positive rate in such cases is estimated as the community infection rate (the proportion of positive cases among all those tested).

(Method 1)
The data acquisition unit 11 acquires the following data from close contact surveys conducted during the same period: the positivity rate r obtained as a result of clinical tests in cases where the infection rate was zero or very low, the city population N (people), the number of positive cases k (people) obtained as a result of clinical tests conducted on close contacts, and the sample extraction number n (people) (the number of people surveyed as close contacts).
The infection status estimation unit 13 estimates the number of infected people m in the city using the following equation (3).
r = ( _m C _k・_(N-m) C _(n-k) )/ _N C _m ...(3)
The infection status estimation unit 13 creates equation (3) for each sample in the close contact survey. The infection status estimation unit 13 solves equation (3) for m to calculate the variance of m for each sample in the close contact survey conducted during the same period. If the variance value is equal to or less than a threshold, a representative value of m (e.g., the average, median, mode, etc.) is identified as the number of infected people in the city. The infection status estimation unit 13 estimates the community infection rate by dividing the number of infected people m by the city population N.

(Method 2)
If the variance value is greater than or equal to the threshold, the infection rate in the community, r, is estimated using the following method:
First, the onset date and infection date of infected persons A to D are estimated. This is shown in Figure 3A. The black circle in Figure 3A is the onset date, and the white circle in Figure 3A is the infection date (estimated date). Next, the behavior of infected persons A to D is investigated, for example, where they stopped by on what date and time, and with whom they had close contact, and the results are summarized. Figure 3B shows a table summarizing the behavior of A's close contacts A-1 to A-n, B's close contacts B-1 to B-m, C's close contact C-1, etc. The stars in Figure 3B show the days when each close contact came into contact with one of infected persons A to D, and the crosses in Figure 3B show the days when the close contacts were investigated. Next, the contact situation with the infected person (whether the environment was conducive to infection) is organized by location and date and time. Figure 3C shows a table showing the contact situation with the infected person. The white squares in Figure 3C indicate the locations and days when infected people were present, and the black squares indicate the locations and days when clusters of infected people were confirmed. Figure 3D shows a diagram in which locations 1 to L are classified in terms of human density and the proportion of positive cases. Location 2 is the location where many infected people were present, and the remaining locations 1, 3, and L are locations where the number of infected people is below the threshold. The infection situation in locations where infected people were present but the number of infected people is below the threshold is considered to be close to the situation of community infection, and the community infection rate r is estimated based on the results of a survey of close contacts who came into contact with infected people in locations 1, 3, and L.

The data acquisition unit 11 acquires data for all close contacts, including whether or not they developed symptoms n days after coming into contact with an infected person.
The infection status estimation unit 13 creates the following formula for all close contacts.
R = P ₀ + (1-P ₀ ) × [P ₁ + (1-P ₁ )・P ₁ +...+(1-P ₁ ) ^n-1・P ₁ ]... (4)
Here, R is the probability that a close contact will be positive on the test date n days after contact m days ago, _P0 is the community infection rate m days ago, and _P1 is the infection rate per day. _P1 is assumed to be constant.
The infection status estimation unit 13 uses data including the presence or absence of onset of symptoms n days later acquired by the data acquisition unit 11 for all close infected persons who came into contact with an infected person at any of the locations 1, 3, and L m days ago, and creates equation (4) by substituting 1 for R if the person had onset n days later and 0 for R if the person did not have onset n days later, and determines P ₀ and P ₁ that fit the infection status (positive or not) from m days ago to n days later for all close infected persons who came into contact with any of the infected persons A to D m days ago using the least squares method or the like. When P ₀ and P ₁ are determined, the community infection rate m days ago (i.e., P ₀ ) can be estimated. The infection status estimation unit 13 performs the same process by changing the value of m to estimate the community infection rate for each day.

(Calculating infection status for resource forecasting)
The sensitivity a, specificity b, and community infection rate were estimated by the above processing. The test accuracy estimation unit 12 and the infection status estimation unit 13 perform the above processing every day to estimate the latest sensitivity a, specificity b, and community infection rate. Next, the infection status estimation unit 13 calculates the number of false positives, false negatives, true positives, and true negatives required to predict the demand for sanitary function maintenance resources. For example, the infection status estimation unit 13 multiplies the total number of people who visited a hospital or the like for testing from m days ago to the present by the true infection rate x calculated by formula (2) to calculate the number of true positives. The infection status estimation unit 13 calculates the number of false positives by subtracting the number of true positives from the total number of people who were determined to be positive. The infection status estimation unit 13 calculates the number of true negatives by multiplying the total number of people who were determined to be negative by the specificity. The infection status estimation unit 13 calculates the number of false negatives by subtracting the sum of the number of true positives, the number of false positives, and the number of true negatives from the total number of people who underwent clinical testing. The demand prediction unit 14 predicts the demand for hygiene function maintenance resources for people who have undergone testing from m days ago to the present, based on the current number of true positives, false positives, true negatives, and false negatives.

In addition, because there is an incubation period and asymptomatic individuals, not all people infected with an infectious disease undergo clinical testing immediately after infection. In general, it is thought that people undergo clinical testing after a certain period of time (e.g., several days) has passed since infection. In other words, it is thought that the community infection rate on a certain day is related to the number of people who undergo clinical testing several days after that day. Based on this idea, the infection status estimation unit 13 multiplies the daily community infection rate by the city population N to calculate the number of people infected in the community on a daily basis, and may predict the number of people who will undergo clinical testing in the future from the relationship between the daily community infection rate or number of people infected in the community and the number of people who undergo clinical testing at hospitals, etc. on a daily basis.

For example, if it can be estimated from the data illustrated in FIG. 3A that people come for testing two days after infection on average and are found to be positive, the infection status estimation unit 13 may estimate the number of positive cases on the next day based on the number of community infections one day before (m=1), and estimate this number of positive cases as the number of people examined. For example, the infection status estimation unit 13 may calculate the correlation coefficient between the number of community infections on a certain day and the number of people who underwent clinical testing X days after that day, estimate the number of people examined on the next day based on the number of community infections X-1 days before, and calculate the number of true positives, false positives, true negatives, and false negatives on the next day in the same manner as above. The infection status estimation unit 13 may also estimate the future community infection rate, etc. from the daily community infection rate or the trend in the number of community infections, and predict the number of people examined in the future by applying the relationship between the past community infection rate and the number of people examined to the estimated future community infection rate. For example, if the community infection rate value for the previous week is approximately constant, the infection status estimation unit 13 predicts that the community infection rate value will remain at the same value for the next week, and predicts that the number of people undergoing testing will also remain the same as last week. The infection status estimation unit 13 may also predict the number of people undergoing testing in the future using a predicted value of the infectious disease epidemic by a public institution using the IBM method or the SIR method. For example, the infection status estimation unit 13 may calculate the correlation between the predicted value of the number of infected people in the past published by a public institution in the past and the actual value of the number of people undergoing clinical testing at that time, and predict the number of people undergoing testing in the future based on the correlation and the predicted value of the number of infected people in the future.

If the number of people undergoing testing can be estimated from the trend in community infection rates, the future number of true positives, false positives, true negatives, and false negatives can be calculated using equation (2). The demand prediction unit 14 predicts future demand for sanitation function maintenance resources based on the future number of true positives, false positives, true negatives, and false negatives.

(Prediction of demand for resources to maintain sanitation functions)
The demand prediction unit 14 includes a simulator that simulates the subsequent movement of people who have undergone clinical testing based on the status of the number of infected people (the number of true positives, the number of false positives, the number of true negatives, and the number of false negatives) estimated by the infection status estimation unit 13. For example, true positives and false negatives (i.e., people who are infected regardless of the test result) progress to the next level of symptoms with a probability of P ₀₁ to P ₃₄ calculated based on the clinical data (FIG. 2A), while the others improve. The simulator advances the symptoms of true positives and false negatives to the next level (more severe symptoms) with a probability of P ₀₁ to P ₃₄ , and moves the patient to a hospital accordingly. The simulator allocates resources to be used from among the preset sanitary function maintenance resources. The demand prediction unit 14 calculates the sanitary function maintenance resources (hospital beds and transportation means) required due to the movement of patients, and calculates the demand for sanitary function maintenance resources.

Figure 4 shows an example of a resource for maintaining sanitation functions. When an infectious disease outbreak occurs, people suspected of infection undergo clinical testing at infectious disease testing laboratories 20 set up in hospitals or outdoor tents, and depending on the results, they are assigned to one of the following locations: home 30, a hotel 31 for temporary isolation, a hotel 32 for people with mild symptoms, a general hospital 33 for people without infection, a general hospital 34 for people with moderate symptoms, or a base hospital 35 for people with severe symptoms.

Figure 5A shows an example of the movement of an asymptomatic person when undergoing clinical testing. For example, if a person is actually infected but the test result is negative (first row of data in Figure 5A), this person is temporarily returned from the testing center 20 to their home 30 because the test result is negative. However, since the person is actually infected, they will develop symptoms after a while and be admitted to a hotel 32 for people with mild symptoms. If their symptoms worsen thereafter, they will be transferred to a general hospital 34 for people with moderate symptoms, and if their symptoms worsen further, they will be transferred to a base hospital 35 for people with severe symptoms, and if their symptoms improve with treatment, they will be transferred to the general hospital 34 for people with moderate symptoms and the hotel 32 for people with mild symptoms.

The demand prediction unit 14 simulates the same movement of the asymptomatic false negatives as described above (first row in FIG. 5A), and calculates the number of testing centers, hospital beds, transportation means, and the like required for that. For example, the demand prediction unit 14 multiplies the number of false negatives among the current or future number of true positives, false positives, true negatives, and false negatives calculated by the infection status estimation unit 13 by a predetermined coefficient or a randomly set coefficient to set the number of asymptomatic false negatives, and calculates the movement of each false negative and the resources required at that time, assuming that the above-mentioned symptom changes occur for each of the set false negatives. For example, the demand prediction unit 14 simulates the movement of a certain false negative person, in which the person undergoes a clinical test at a testing center and then waits at the testing center until the results are available. In addition, a maximum number of tests that can be performed at a testing center in one day is set, and if the maximum number is exceeded, the person is simulated to be tested on the next day. For example, the demand prediction unit 14 simulates the movement of a false negative person to be tested, wait at a testing center until a negative result is obtained, and then return to the home 30. Next, the demand prediction unit 14 simulates the false negative person returning to the home 30 and developing mild symptoms in the evening X days later, and calculates that one ambulance 25 and one room (= one hospital bed) in a hotel 32 for mildly ill people will be required in the evening X days later. For example, when the ambulance 25 is being used to transport other patients, the demand prediction unit 14 calculates the waiting time until the ambulance 25 becomes available to transport the target false negative person. The demand prediction unit 14 also calculates the required quantity, whether or not it can be used, the waiting time until it can be used, the time or period of use, etc. for the sanitary function maintenance resources required due to subsequent changes in symptoms, and records them in the storage unit 16. For example, the memory unit 16 records information such as "For false negative subject A, one ambulance 25 arrives at home 30 at 16:00 on D2 of M2, travels 30 minutes from home 30 to hotel 32, and stays at hotel 32 for two days." The demand prediction unit 14 similarly simulates the movements of other false negative subjects, and records the sanitary function maintenance resources required as symptoms change in the memory unit 16 together with information on the time for which the sanitary function maintenance resources are required. For example, the demand prediction unit 14 may simulate the movements of asymptomatic false negative subject B, in which mild symptoms develop in the evening X-1 days after returning to home 30, and B recovers after a further two-week stay in hotel 32 for mild cases, and returns to home 30 without moving to general hospital 34 for moderate cases. That is, the simulator of the demand prediction unit 14 may be configured to perform a simulation by arbitrarily setting the number of days required for the symptoms to progress to, for example, 1 to 2 days shorter or longer based on the values of D ₀₁ to D ₃₄ calculated from the clinical data of FIG. 2A. The simulator of the demand prediction unit 14 may determine whether or not to proceed to the next symptom level based on the values of P ₀₁ to P ₃₄ calculated from the clinical data, and determine the movement of the subject. Regarding the process of symptom improvement, the demand prediction unit 14 simulates the symptoms to improve with a predetermined probability (which may be set randomly with an allowable range based on the predetermined probability). For example, the demand prediction unit 14 may simulate the symptoms to improve from moderate to mild with a probability of (1-P ₂₃ ). The demand prediction unit 14 may also calculate whether or not the symptoms can be used, and if the symptoms cannot be used, the waiting time, etc., according to the usage status of the hygiene function maintenance resource set in advance, and record them in the storage unit 16 together with the necessary resources. In addition, not only the beds but also the means of transportation may be set to allocate different resources according to the level of symptoms. For example, two ambulances 25 used for transporting mildly ill patients and three ambulances 25 used for transporting moderately or more ill patients may be distinguished, and the demand prediction unit 14 may operate to allocate an ambulance 25 according to the level of symptoms to the transportation of each patient. In addition, a vehicle other than the ambulance 25 (such as a taxi) may be set to be preferentially allocated to the transportation of mildly ill or asymptomatic patients, and the demand prediction unit 14 may operate to allocate a taxi or the like to the transportation of mildly ill or asymptomatic patients.

In addition, in infectious disease countermeasures, transportation means are needed not only for patients but also for negative pressure tents, chemicals and equipment used in testing, and specimens collected in clinical testing, and the simulator of the demand prediction unit 14 may simulate the movement of transporting these. For example, the demand prediction unit 14 is configured to be able to set a predetermined number of vehicles for transporting medical supplies in addition to ambulances 25, and the demand prediction unit 14 simulates the movement of transporting collected specimens to hospitals or research institutions after testing is performed at the testing center 20, and records the number of vehicles required at that time, the time the vehicles will be used, the waiting time for the vehicles, etc. in the memory unit 16.

The second line of Fig. 5A shows an example of the movement of an asymptomatic person whose test result is negative and who is actually negative (true negative). Similarly, the third line of Fig. 5A shows an example of the movement of an asymptomatic person whose test result is positive and who is actually positive (true positive), and the fourth line shows an example of the movement of an asymptomatic person whose test result is positive and who is actually negative (false positive).
The demand prediction unit 14 calculates the number of asymptomatic true positives, the number of asymptomatic false positives, and the number of asymptomatic true negatives based on the number of true positives, the number of false positives, and the number of true negatives calculated by the infection status estimation unit 13. The demand prediction unit 14 simulates the movements and the like exemplified in the second line of Fig. 5A for asymptomatic true negatives, the movements and the like exemplified in the third line of Fig. 5A for asymptomatic true positives, and the movements and the like exemplified in the fourth line of Fig. 5A for asymptomatic false positives, and records information such as the type and quantity of sanitary function maintenance resources required for each movement, the required time or period, and whether or not they can be used in the storage unit 16.

Figure 5B shows the movement of mildly ill people by infection status. For example, the second line of Figure 5B shows an example of the movement of a person who tested negative and was actually negative (true negative). In this case, even though the person tested negative, because they had mild symptoms (they had symptoms), they were transported to a general hospital 34 for moderately ill people and diagnosed, and because it was not a target infectious disease, they were admitted to a general hospital 33 for non-infected people.
The demand prediction unit 14 also calculates the number of true positive cases with mild symptoms, the number of false positive cases with mild symptoms, the number of true negative cases with mild symptoms, and the number of false negative cases with mild symptoms based on the number of true positive cases, false positive cases, true negative cases, and false negative cases calculated by the infection status estimation unit 13, and calculates information such as the type and quantity of sanitary function maintenance resources required, the time or period required, and whether they can be used, and records this information in the memory unit 16.

Figure 5C shows the movement of moderately ill people by infection status. For example, the third line of Figure 5C shows an example of the movement of a person who tested positive and is actually positive (true positive). In this case, when the test result (positive) is known, the person is transported to a general hospital 34 for moderately ill people, and progresses to a severe case with a probability of _P34 . If the person becomes severely ill, the person is transported to a base hospital 35 for severe cases. After that, as the symptoms improve, the person is transported to a general hospital 34 for moderate cases and a hotel 32 for mild cases.
The demand prediction unit 14 sets a part of the number of true positive cases calculated by the infection status estimation unit 13 as moderately ill cases, and simulates the movement of these moderately ill cases, for example, as shown in the third line of Fig. 5C. The demand prediction unit 14 calculates the sanitary function maintenance resources required for this movement, and records it in the storage unit 16.
Similarly, the demand prediction unit 14 calculates the number of moderate false positives, the number of moderate true negatives, and the number of moderate false negatives based on the number of false positives, true negatives, and false negatives calculated by the infection status estimation unit 13, and simulates, for example, the movements illustrated in Figure 5C to calculate information such as the type and quantity of sanitary function maintenance resources required, the required time or period, whether they can be used, etc., and records this information in the memory unit 16.

FIG. 5D shows the movement of seriously ill patients by infection status. For example, the fourth line of FIG. 5D shows an example of the movement of a person who has a positive test result but is not actually infected (false positive). In this case, when the test result (positive) is found, the patient has symptoms, so he or she is temporarily transported to a general hospital 34 for moderate cases and diagnosed. After that, symptoms due to other diseases, which are not the target infectious disease, may worsen. The simulator simulates the worsening of symptoms for diseases other than infectious diseases with a predetermined probability. False positive patients are transported to a base hospital 35 for severe cases and treated there. The severe symptoms improve with subsequent treatment, and the patient is moved to a general hospital 34 for moderate cases or a hotel 32 for mild cases. The simulator, for example, simulates this movement, and the demand prediction unit 14 records in the memory unit 16 the number of hospital beds (general hospital 34, base hospital 35, general hospital 34, hotel 32) required during this movement, the period during which the hospital beds will be used, the ambulance 25 required for transportation between hospitals and its period of use, etc. The demand prediction unit 14 may also simulate the movement of other false positives, such as transporting them to a hotel for mild cases 32 after treatment at a general hospital 34, based on the probability that the symptoms defined for a disease other than an infectious disease will progress to the next level.
Similarly, the demand prediction unit 14 calculates the number of severe false negatives, the number of severe true negatives, and the number of severe true positives based on the number of false negatives, true negatives, and true positives calculated by the infection status estimation unit 13, and simulates, for example, the corresponding infection status trends in Figure 5D to calculate information such as the type and quantity of sanitary function maintenance resources required, the required time or period, whether they can be used, etc., and records this in the memory unit 16.

Figure 5E shows the movement of serious cases by infection status. The demand prediction unit 14 calculates the number of serious false negative cases, the number of serious true negative cases, and the number of serious true positive cases based on the number of false negative cases, the number of true negative cases, and the number of serious true positive cases calculated by the infection status estimation unit 13, and, for example, simulates the movement illustrated in Figure 5E to calculate information such as the type and quantity of sanitary function maintenance resources required, the time or period required, and whether or not they can be used, and records this information in the memory unit 16.

As described above, the demand prediction unit 14 simulates the movement of subjects according to the infection status of true positive, false positive, true negative, and false negative, and calculates and records the sanitary function maintenance resources required during that time. The demand prediction unit 14 varies parameters such as the probability that the infection will progress to the next level and the number of days required for the infection to progress to the next level, and also varies the quantity of sanitary function maintenance resources, simulating multiple patterns for the same population (a certain number of false negatives, true negatives, true positives, and false positives calculated by the infection status estimation unit 13).

FIG. 6 shows an example of data recorded in the storage unit 16 after the simulation.
The "simulation pattern ID" in the table of FIG. 6 is an identifier of the simulation pattern. For one simulation pattern, the probability of infection progressing to the next level, the number of days required for infection to progress to the next level, the number of asymptomatic, mild, moderate, severe, and critical false positives, true negatives, and false negatives calculated by the infection status estimation unit 13, the number of testing centers 20, the number of ambulances 25, the number of beds in the hotel 31 for temporary isolation, the hotel 32 for mild symptoms, the general hospital 33 for non-infected people, the general hospital 34 for moderate symptoms, and the base hospital 35 for severe symptoms are set, and a simulation of the movement of test recipients (demand forecast simulation) is performed. The "test recipient ID" represents one test recipient, and the sanitary function maintenance resources required for each test recipient according to the change in symptoms and movement are recorded (in the "resource type", "quantity", and "time/period" columns). In the "infection status", it is recorded whether the test recipient is asymptomatic, mildly ill, moderately ill, severely ill, or critically ill, and whether the result is true positive, false positive, true negative, or false negative. In the "simulation result", the availability of resources in the simulation, the waiting time, etc. are recorded. For example, if it takes 10 minutes for the ambulance 25 to arrive, "waiting 10 minutes" is recorded (first line). In the "description", the movement of the test recipient regarding the resource is recorded.

A decision maker who allocates resources for maintaining sanitary functions can execute a simulation by changing the quantity of each resource in various ways, and can understand whether the supply of laboratories, hospital beds, and transportation means such as ambulances is sufficient by referring to the simulation results shown in FIG. 6. For example, the decision maker can repeatedly change the quantities of laboratories, hospital beds, and transportation means and perform simulations until no "waiting" is recorded for any of the laboratories, hospital beds, and transportation means. If the optimal quantities of laboratories, hospital beds, and transportation means can be calculated by the simulation, the decision maker can take measures such as increasing the resources by comparing them with the current quantities of those resources. In addition, for example, if the infection spreads in the city and the number of positive cases increases, there is a possibility that the number of hospital beds will be insufficient. For example, if the shortage of hospital beds is a hotel 32 that accommodates patients with mild symptoms, the decision maker estimates the number of hospital beds that will be insufficient and the period for which they will be insufficient, and considers renting a hotel temporarily for this period. The decision maker can increase the number of beds in hotel 32 for mild cases by the number of beds owned by the newly rented hotel, and have the demand prediction unit 14 run a simulation based on a prediction of the infection situation during the period when there will be a shortage of beds.As a result, the decision maker can identify the number of beds in hotel 32 for mild cases when there are no more "waiting" beds in the table shown in FIG. 6 or when the number of records with "waiting" recorded falls below a threshold, and can then consider renting a number of hotels to cover the identified number of beds.
Similarly, if the number of positive cases increases and there is a shortage of ambulances 25, the decision maker considers borrowing from other municipalities for the period and number of vehicles that are in short supply, or using a private ambulance service only for those with mild symptoms. The decision maker repeatedly causes the demand prediction unit 14 to execute a simulation based on the prediction of the infection situation during the period when there is a shortage of ambulances, while increasing the number of ambulances 25, for example, by one at a time. As a result, the number of ambulances 25 when there is no "waiting" for the transportation means in the table illustrated in FIG. 6 or the number of records in which "waiting" is recorded is equal to or less than a threshold is specified, and it is possible to consider borrowing from other municipalities so that the specified number of ambulances can be covered, or using a private ambulance service for the transportation means used to move those with mild symptoms.

(Operation)
FIG. 7 is a flowchart illustrating an example of a demand amount prediction process according to an embodiment.
The test accuracy estimation unit 12 estimates the test accuracy (step S11). The test accuracy estimation unit 12 calculates the sensitivity a and the specificity b.
Next, the infection status estimation unit 13 estimates the community infection status (step S12). The infection status estimation unit 13 estimates the infection rate in the community.
The processes of steps S11 and S12 are executed, for example, daily.

Next, the infection status estimation unit 13 identifies the infection status to be the target of the demand prediction (step S13). For example, the infection status estimation unit 13 calculates the number of true positives, false positives, true negatives, and false negatives for the people who underwent the clinical test based on the number of people who actually underwent the clinical test, the sensitivity a and specificity b estimated in step S11, and formula (2).
Also, for example, the infection status estimation unit 13 calculates the correlation between the number of people who underwent clinical testing on a certain day and the community infection rate X days before that day, and estimates the number of people who will be examined X-1 days later based on the latest community infection rate (for example, the community infection rate on the previous day) and the calculated correlation. Also, for example, the infection status estimation unit 13 calculates the true infection rate based on the latest positive rate, sensitivity a, specificity b, and formula (2), and calculates the future number of true positives, false positives, true negatives, and false negatives from the true infection rate and the estimated number of people who will be examined in the future (for example, X-1 days later). The output unit 15 outputs the current or future number of true positives, false positives, true negatives, and false negatives for people who underwent clinical testing to a display device.

Next, the decision maker sets the parameters required for the demand prediction (step S14). For example, the decision maker inputs parameters such as the number of testing centers 20, ambulances 25, hotels 31 for temporary isolation, hotels 32 for mild cases, the number of hospital beds in general hospitals 33 for non-infected patients, general hospitals 34 for moderate cases, and base hospitals 35 for severe cases, the number of vehicles transporting medical supplies, and the number of true positives, false positives, true negatives, and false negatives to be targeted in the demand prediction into the prediction device 10. The demand prediction unit 14 acquires the input parameters, records them in the memory unit 16, and sets them as conditions for the simulation.

The decision maker may input other parameters, such as the probability that the infection will progress to the next level, the number of days it will take for the infection to progress to the next level, etc., but these may be set automatically by the demand prediction unit 14 based on P ₀₁ to P ₃₄ and D ₀₁ to D ₃₄ calculated by the test accuracy estimation unit 12.

Next, the demand prediction unit 14 simulates the movement of the test recipient (step S15). The demand prediction unit 14 records in the storage unit 16 information on the sanitary function maintenance resources that have become necessary during the simulation.
When various parameters are set in step S14, the demand prediction unit 14 may execute a plurality of patterns of simulations under those conditions. For example, when the simulation pattern changes, the demand prediction unit 14 changes the breakdown of asymptomatic, mild, moderate, severe, and critical symptoms in the number of true positives, or changes the progression pattern of symptoms (for example, changing moderate → severe → recovered to moderate → recovered), simulates the movement of test recipients in another pattern, and calculates the sanitary function maintenance resources required for that pattern.

Next, the output unit 15 outputs the simulation result (step S16). For example, the output unit 15 may refer to the information (FIG. 6) recorded in the storage unit 16 as a result of the simulation of a plurality of patterns performed by the demand prediction unit 14, and may count records of waiting times of a predetermined time or more by resource, and display these. For example, if 30 cases of waiting at a laboratory (carrying over to the next day, etc.) are recorded throughout all the simulation patterns, 10 cases of waiting for a hospital bed (waiting for one day or more) are recorded throughout all the simulation patterns, and 20 cases of shortage of transportation means (waiting for one hour or more) are recorded throughout all the simulation patterns, the output unit 15 outputs "30 cases of shortage of a laboratory, 10 cases of shortage of a hospital bed, 20 cases of shortage of transportation means" as the result of the demand prediction for the parameters set this time.
The decision maker repeatedly sets the parameters in step S14 (resets the quantity of sanitary function maintenance resources) and has the demand prediction unit 14 run a simulation (step S15) until a satisfactory result is obtained.

As described above, according to this embodiment, it is possible to estimate sensitivity a and specificity b based on clinical data and hospital bed data. This makes it possible to calculate the true infection rate. In addition, it is possible to estimate the community infection rate based on the test results of close contacts, and predict the future number of people undergoing testing from the trend in the community infection rate. In addition, it is possible to estimate the future infection situation (number of true positives, number of false positives, number of true negatives, number of false negatives) based on the future number of people undergoing testing and the true infection rate, and to predict the demand for resources to maintain hygiene functions based on the future infection situation and the movement of people for each infection situation.

FIG. 8 is a diagram illustrating an example of a hardware configuration of a prediction device according to an embodiment of the present disclosure.
The computer 900 includes a CPU 901 , a main memory device 902 , an auxiliary memory device 903 , an input/output interface 904 , and a communication interface 905 .
The prediction device 10 described above is implemented in a computer 900. The above-mentioned functions (data acquisition unit 11, test accuracy estimation unit 12, infection status estimation unit 13, demand prediction unit 14, output unit 15) are stored in the auxiliary storage device 903 in the form of a program. The CPU 901 reads the program from the auxiliary storage device 903, loads it in the main storage device 902, and executes the above-mentioned processing according to the program. The CPU 901 also secures a storage area in the main storage device 902 according to the program. The CPU 901 also secures a storage area in the auxiliary storage device 903 for storing data being processed according to the program.

A program for realizing all or part of the functions of the prediction device 10 may be recorded on a computer-readable recording medium, and the program recorded on the recording medium may be read into a computer system and executed to perform processing by each functional unit. The term "computer system" as used herein includes hardware such as an OS and peripheral devices. In addition, if a WWW system is used, the term "computer system" also includes a home page providing environment (or display environment). In addition, the term "computer-readable recording medium" refers to portable media such as CDs, DVDs, and USBs, and storage devices such as hard disks built into a computer system. In addition, if the program is distributed to a computer 900 via a communication line, the computer 900 that receives the program may expand the program in the main storage device 902 and execute the above processing. In addition, the above program may be for realizing part of the functions described above, or may be capable of realizing the functions described above in combination with a program already recorded in the computer system. In addition, the prediction device 10 may be composed of multiple computers 900. The storage unit 16 may be stored in an external storage device separate from the computer 900.

In addition, the components in the above-described embodiments may be replaced with well-known components as appropriate, without departing from the spirit of this disclosure. Furthermore, the technical scope of this disclosure is not limited to the above-described embodiments, and various modifications may be made without departing from the spirit of this disclosure.

<Additional Notes>
The prediction device 10, the prediction method, and the program described in each embodiment can be understood, for example, as follows.

(1) The prediction device 10 of the first aspect includes a test accuracy estimation unit 12 that estimates the accuracy of an infectious disease test, an infection status estimation unit 13 that estimates the infection status in a prediction target area based on a true infection rate (equation (2)) based on the accuracy of the test and the number of people who have undergone the test in the prediction target area (e.g., a specific prefecture, city, town, or village, or a specific hospital) (actual value to date) or a predicted value of the number of people who have undergone the test (a predicted value of future people who have undergone the test based on the relationship between the community infection rate and the actual number of people who have undergone the test, etc.), and a demand prediction unit 14 that predicts the demand for resources to maintain sanitation functions based on the number of infected people indicated by the infection status.
This makes it possible to estimate the true infection situation in the target area taking into account the accuracy of infectious disease testing, and to predict the demand for resources to maintain sanitation functions (testing centers, hospital beds, transportation such as ambulances) based on the number of infected people indicated by the true infection situation.

(2) A prediction device 10 according to a second aspect is the prediction device 10 of (1), wherein the infection status estimation unit 13 calculates the number of true positives who are determined to be positive and are actually infected, the number of false positives who are determined to be positive and are not actually infected, the number of false negatives who are determined to be negative and are actually infected, and the number of true negatives who are determined to be negative and are actually not infected; and the demand prediction unit 14 calculates a first quantity of the resources required by the true positives based on changes in symptoms of the true positives and the number of the true positives, a second quantity of the resources required by the false positives based on changes in symptoms of the false positives and the number of the false positives, a third quantity of the resources required by the true negatives based on changes in symptoms of the true negatives and the number of the true negatives, and a fourth quantity of the resources required by the false negatives based on changes in symptoms of the false negatives and the number of the false negatives, and predicts the demand for resources based on the first quantity, the second quantity, the third quantity, and the fourth quantity.
Those who undergo an infectious disease test can be classified into true positive, false positive, false negative, and true negative, and the amount of sanitary function maintenance resources required varies depending on the classification. According to the second aspect, the demand for sanitary function maintenance resources is predicted for each true positive, false positive, false negative, and true negative, making it possible to accurately predict the demand.

(3) A prediction device 10 according to a third aspect is the prediction device 10 of (1) to (2), wherein the test accuracy estimation unit 12 generates a plurality of combinations of sensitivity and specificity for an infection rate calculation formula (formula (2)) expressed using the sensitivity and specificity of the test, calculates the infection rate corresponding to the plurality of combinations, calculates an estimate of the demand for the resources according to the progression of the infection for each of the plurality of combinations based on the plurality of infection rates, the number of subjects who will undergo the test, and infection rate progression schedule information, selects from the calculated estimates of demand the estimate of demand that is closest to the actual number of resources in operation required for the subjects who will undergo the test, identifies the combination corresponding to the estimate of demand, and determines the sensitivity and specificity associated with the identified combination as the accuracy of the test.
This allows us to estimate the sensitivity and specificity, and then estimate the true infection rate using equation (2).

(4) The prediction device 10 according to a fourth aspect is the prediction device 10 of (1) to (3), in which the resource is a hospital bed.
This makes it possible to predict the demand for hospital beds.

(5) The prediction device 10 according to the fifth aspect is the prediction device 10 of any one of (1) to (4), which is a means of transport (an ambulance 25, a vehicle for medical supplies) for transporting people or things (such as specimens) necessary for measures against the infectious disease.
This makes it possible to predict the demand for transportation such as ambulances.

(6) The prediction device 10 according to a sixth aspect is the prediction device 10 of (1) to (5), which is a facility (inspection station 20) that performs the inspection.
This allows for a forecast of laboratory demand.

(7) The prediction device 10 according to the seventh aspect is the prediction device 10 of (1) to (6), wherein the infection status estimation unit 13 estimates the infection rate in the city based on the number of positive cases in the test among close contacts of a test subject who was determined to be positive in the test, who had contact in an environment where infection is unlikely to occur.
By estimating the community infection rate, it becomes possible to analyze the relationship between the trend in the community infection rate and the number of people undergoing testing. In addition, the number of people undergoing testing in the future can be estimated from the recent trend in the community infection rate, and the future demand for resources to maintain hygiene functions can be predicted based on the future number of people undergoing testing.

(8) The prediction device 10 of the eighth aspect is the prediction device 10 of (1) to (7), wherein the infection status estimation unit 13 predicts the future number of people who will undergo the test based on the infection rate in the city, and the demand prediction unit 14 predicts the future demand for resources based on the infection status corresponding to the future number of people who will undergo the test.
This makes it possible to predict future demand for resources to maintain sanitation functions.

(9) The prediction method according to the ninth aspect includes a step of estimating the accuracy of infectious disease testing, a step of estimating the infection status of the prediction target area based on the true infection rate based on the accuracy of the testing and the number of people undergoing the testing in the prediction target area or a predicted value of the number of people undergoing the testing, and a step of predicting the demand for resources to maintain sanitation functions based on the number of infected people indicated by the infection status.

(10) The program according to the tenth aspect causes a computer to execute the steps of estimating the accuracy of infectious disease testing, estimating the infection status of a prediction target area based on a true infection rate based on the accuracy of the testing and the number of people undergoing the testing in the prediction target area or a predicted value of the number of people undergoing the testing, and predicting the demand for resources to maintain hygiene functions based on the number of infected people indicated by the infection status.

10: Prediction device 11: Data acquisition unit 12: Test accuracy estimation unit 13: Infection status estimation unit 14: Demand prediction unit 15: Output unit 16: Memory unit 20: Testing station 25: Ambulance 30: Home 31: Hotel for temporary isolation 32: Hotel for mild cases 33: General hospital for non-infected persons 34: General hospital for moderate cases 35: Base hospital for severe cases 900: Computer 901: CPU
902: Main storage device 903: Auxiliary storage device 904: Input/output interface 905: Communication interface

Claims (10)

Regarding the accuracy of an infectious disease test, if a person who is actually infected among those who have been tested and determined to be positive is defined as a true positive, and a person who is not actually infected among those who have been tested and determined to be negative is defined as a true negative, then the following formula is used to calculate x for a plurality of combinations of a and b (where B=1-b, x is the true infection rate, a is the sensitivity, b is the specificity, and r is the apparent infection rate): sensitivity indicates the proportion of true positives among those who have been tested and are actually infected; specificity indicates the proportion of true negatives among those who have not actually been infected; apparent infection rate indicates the probability that those who have been tested will be determined to be positive; and true infection rate. a test accuracy estimation unit that estimates a plurality of true numbers of infected persons by multiplying the number of people who have undergone the test, and calculates an estimate of the number of infected persons that will progress to symptoms of the next level by multiplying each estimated true number of infected persons by the probability that symptoms of the infectious disease will progress to symptoms of the next level, previously analyzed based on clinical data, and calculates the number of hospital beds required by the number of infected persons related to the estimated number of infected persons after the required number of days, based on the estimated number of infected persons and the number of days required for symptoms to progress to the next level, previously analyzed based on the clinical data, and compares the number of hospital beds calculated for each of the multiple x's with an actual value of the number of hospital beds actually required after the required number of days, and identifies the true infection rate that is the x when the true infection rate is closest to the actual value, thereby estimating the sensitivity and the specificity;
an infection status estimation unit that estimates the infection status of the prediction target area based on the sensitivity estimated by the test accuracy estimation unit, the specificity estimated by the test accuracy estimation unit, an apparent infection rate indicating the probability of a positive result in the test, the true infection rate specified by the test accuracy estimation unit, and the number of test recipients in the prediction target area or a predicted value of the number of test recipients;
a simulator that simulates the progression of symptoms to the next level after the required number of days for each of the true positive individuals, false positive individuals who are determined to be positive but are not actually infected, the true negative individuals, and false negative individuals who are determined to be negative and are actually infected, and allocates the number and type of resources required to maintain hygiene functions to the individuals whose symptoms have progressed to the next level, and predicts the demand for resources based on the infection status estimation result and the simulator;
Equipped with
The infection status indicates the number of true positives, the number of false positives, the number of true negatives, and the number of false negatives,
The infection status estimation unit is
When estimating the infection status based on the true infection rate and the number of people who took the test, multiply the true infection rate by the number of people who took the test to calculate the number of true positives, subtract the number of true positives from the number of people who were determined to be positive in the test to calculate the number of false positives, multiply the number of people who were determined to be negative in the test by the specificity to calculate the number of true negatives, and subtract the sum of the number of true positives, false positives, and true negatives from the number of people who took the test to calculate the number of false negatives.
When estimating the infection status based on the true infection rate and the predicted value of the number of people who will be tested, the true infection rate is multiplied by the predicted value of the number of people who will be tested to calculate the number of future true positives, the number of future false positives is calculated by subtracting the number of true positives from the predicted value of the number of people who will be determined to be positive in the test, the number of future true negatives is calculated by multiplying the predicted value of the number of people who will be determined to be negative in the test by the specificity, and the sum of the future number of true positives, the future number of false positives, and the future number of true negatives is calculated from the predicted value of the number of people who will be tested.
Prediction device. The simulator allocates the resources by simulating the movements of the true positives, the false positives, the true negatives, and the false negatives according to symptoms of asymptomatic, mild, moderate, severe, and critical, and
The demand prediction unit calculates the number of true positive cases, the number of false positive cases, the number of true negative cases, and the number of false negative cases for each symptom based on the number of true positive cases, the number of false positive cases, the number of true negative cases, and the number of false negative cases calculated by the infection status estimation unit,
calculating, for each symptom, a first quantity which is the quantity of the resources required by the true positive persons, a second quantity which is the quantity of the resources required by the false positive persons, a third quantity which is the quantity of the resources required by the true negative persons, and a fourth quantity which is the quantity of the resources required by the false negative persons, based on the calculated numbers of true positive persons, false positive persons, true negative persons, and false negative persons for each symptom and the simulator, and predicting the demand for the resources based on the first quantity, the second quantity, the third quantity, and the fourth quantity calculated for each symptom;
The prediction device according to claim 1 . The inspection accuracy estimation unit is
Generate a plurality of combinations of the sensitivity and the specificity for the relational expression, and calculate the true infection rate according to the plurality of combinations;
calculating an estimate of the demand for resources according to the progress of the infection for each of the combinations based on the true infection rates, the number of people undergoing the test, information on a schedule of the progress of the infection, and predetermined information defining the resources required according to the progress of the infection;
Among the calculated estimated values of the demand, an estimated value of the demand that is closest to the actual number of resources required for the test subjects is selected, the combination corresponding to the estimated value of the demand is identified, and the sensitivity and specificity related to the identified combination are regarded as the accuracy of the test.
The prediction device according to claim 1 or 2. The resource is a hospital bed.
The prediction device according to claim 1 . The resource is a means of transport for transporting people and goods necessary for countermeasures against the infectious disease.
The prediction device according to claim 1 . The resource is equipment for performing the inspection.
The prediction device according to any one of claims 1 to 5. The infection status estimation unit creates the following formula based on the positivity rate r obtained as a result of each of the multiple tests in which the ratio of people who were actually infected among close contacts of the test recipients who were determined to be positive in the test was equal to or less than a threshold value , the population N of the city , the number k of positive cases as a result of clinical tests conducted on the close contacts, and the number n of people who were investigated as close contacts,
r=( _m C _k・_(N-m) C _(n-k) )/ _N C _m
solving the equation for m, calculating the variance of m obtained by solving the equation for each of the tests, and if the value of the variance is equal to or less than a threshold, identifying any one of the average value, median value, or mode of m as the number of infected people in the city, and dividing the number of infected people m by the population N of the city to estimate the infection rate in the city ;
The prediction device according to any one of claims 1 to 6. The infection status estimation unit predicts the number of people who will be examined in the future based on the infection rate in the city,
the demand prediction unit predicts the future demand for the resources based on the infection status according to the number of people who will be examined in the future and the simulator;
The prediction device according to claim 7 . 1. A computer-implemented prediction method comprising:
Regarding the accuracy of an infectious disease test, if a person who has been tested and determined to be positive is actually infected and a person who has been tested and determined to be negative is actually not infected and a true negative is determined, the sensitivity indicating the proportion of the true positives among the people who have been tested and are actually infected, the specificity indicating the proportion of the true negatives among the people who have not actually been infected, the apparent infection rate indicating the probability that the people who have been tested will be determined to be positive, and the true infection rate are expressed as x = (r - B) / (a - B) (where B = 1 - b, x is the true infection rate, a is the sensitivity, b is the specificity, and r is the apparent infection rate), and the x for a plurality of combinations of a and b is calculated, and for each of the calculated plurality of x's, by the number of people who underwent the test to estimate a plurality of true numbers of infected people, and multiply each of the estimated true numbers of infected people by the probability that symptoms of the infectious disease will progress to symptoms of the next level, previously analyzed based on clinical data, to calculate an estimate of the number of infected people who will progress to the next level of symptoms, and calculate the number of hospital beds required by the number of infected people related to the estimated number of infected people after the required number of days based on the estimated number of infected people and the number of days required for symptoms to progress to the next level, previously analyzed based on the clinical data, and compare the number of hospital beds calculated for each of the multiple x's with an actual value of the number of hospital beds actually required after the required number of days, and identify the true infection rate that is the x when it is closest to the actual value, thereby estimating the sensitivity and the specificity;
A step of estimating the infection status of the prediction target area based on the estimated sensitivity, the estimated specificity, an apparent infection rate indicating the probability of a positive result in the test, the specified true infection rate, and the number of people undergoing the test in the prediction target area or a predicted value of the number of people undergoing the test;
a step of simulating the progression of symptoms to the next level after the required number of days for each of the true positives, false positives indicating people who are judged as positive but are not actually infected, the true negatives, and false negatives indicating people who are judged as negative and are actually infected, and allocating the number and type of resources necessary to maintain hygiene functions to the people whose symptoms have progressed to the next level, and predicting the demand for resources based on the estimated result of the infection status;
having
The infection status indicates the number of true positives, the number of false positives, the number of true negatives, and the number of false negatives,
In the step of estimating the infection status,
When estimating the infection status based on the true infection rate and the number of people who took the test, multiply the true infection rate by the number of people who took the test to calculate the number of true positives, subtract the number of true positives from the number of people who were determined to be positive in the test to calculate the number of false positives, multiply the number of people who were determined to be negative in the test by the specificity to calculate the number of true negatives, and subtract the sum of the number of true positives, false positives, and true negatives from the number of people who took the test to calculate the number of false negatives.
When estimating the infection status based on the true infection rate and the predicted value of the number of people who will be tested, the true infection rate is multiplied by the predicted value of the number of people who will be tested to calculate the number of future true positives, the number of future false positives is calculated by subtracting the number of true positives from the predicted value of the number of people who will be determined to be positive in the test, the number of future true negatives is calculated by multiplying the predicted value of the number of people who will be determined to be negative in the test by the specificity, and the sum of the future number of true positives, the future number of false positives, and the future number of true negatives is calculated from the predicted value of the number of people who will be tested.
Forecasting methods. On the computer,
Regarding the accuracy of an infectious disease test, if a person who has been tested and determined to be positive is actually infected and a person who has been tested and determined to be negative is actually not infected and a true negative is determined, the sensitivity indicating the proportion of the true positives among the people who have been tested and are actually infected, the specificity indicating the proportion of the true negatives among the people who have not actually been infected, the apparent infection rate indicating the probability that the people who have been tested will be determined to be positive, and the true infection rate are expressed as x = (r - B) / (a - B) (where B = 1 - b, x is the true infection rate, a is the sensitivity, b is the specificity, and r is the apparent infection rate), and the x for a plurality of combinations of a and b is calculated, and for each of the calculated plurality of x's, by the number of people who underwent the test to estimate a plurality of true numbers of infected people, and multiply each of the estimated true numbers of infected people by the probability that symptoms of the infectious disease will progress to symptoms of the next level, previously analyzed based on clinical data, to calculate an estimate of the number of infected people who will progress to the next level of symptoms, and calculate the number of hospital beds required by the number of infected people related to the estimated number of infected people after the required number of days based on the estimated number of infected people and the number of days required for symptoms to progress to the next level, previously analyzed based on the clinical data, and compare the number of hospital beds calculated for each of the multiple x's with an actual value of the number of hospital beds actually required after the required number of days, and identify the true infection rate that is the x when it is closest to the actual value, thereby estimating the sensitivity and the specificity;
A step of estimating the infection status of the prediction target area based on the estimated sensitivity, the estimated specificity, an apparent infection rate indicating the probability of a positive result in the test, the specified true infection rate, and the number of people undergoing the test in the prediction target area or a predicted value of the number of people undergoing the test;
a step of simulating the progression of symptoms to the next level after the required number of days for each of the true positives, false positives indicating people who are judged as positive but are not actually infected, the true negatives, and false negatives indicating people who are judged as negative and are actually infected, and allocating the number and type of resources necessary to maintain hygiene functions to the people whose symptoms have progressed to the next level, and predicting the demand for resources based on the estimated result of the infection status;
having
The infection status indicates the number of true positives, the number of false positives, the number of true negatives, and the number of false negatives,
In the step of estimating the infection status,
When estimating the infection status based on the true infection rate and the number of people who took the test, multiply the true infection rate by the number of people who took the test to calculate the number of true positives, subtract the number of true positives from the number of people who were determined to be positive in the test to calculate the number of false positives, multiply the number of people who were determined to be negative in the test by the specificity to calculate the number of true negatives, and subtract the sum of the number of true positives, false positives, and true negatives from the number of people who took the test to calculate the number of false negatives.
When estimating the infection status based on the true infection rate and the predicted value of the number of people who will be tested, a process of multiplying the true infection rate by the predicted value of the number of people who will be tested to calculate the future number of true positives, subtracting the number of true positives from the predicted value of the number of people who will be determined to be positive in the test to calculate the future number of false positives, multiplying the predicted value of the number of people who will be determined to be negative in the test by the specificity to calculate the future number of true negatives, and subtracting the sum of the future number of true positives, future number of false positives, and future number of true negatives from the predicted value of the number of people who will be tested to calculate the future number of false negatives;
A program that executes the following.

Images