A method for predicting the scale of respiratory infectious diseases and medical resource demand based on propagation dynamics

Abstract

The present application relates to a kind of respiratory infectious disease scale and medical resource demand prediction method based on propagation dynamics, the present application uses next generation matrix method to carry out preliminary estimation to effective contact rate, and then uses Markov chain monte carlo algorithm to correct fitting parameter, improve the accuracy of model prediction.Most epidemiological parameters cannot obtain an accurate value, and are mostly an interval, therefore the present application carries out sampling in interval, and calculates 95% confidence interval, improve the reliability of prediction.Increase the prediction of medical resource demand such as the number of hospital beds, ICU bed number, whether the reserve of medical resource can be effectively evaluated, and resource can be timely called according to the prediction result if necessary.The present application realizes the prediction of infection scale and medical resource demand under different intervention measures, provides basis for taking appropriate intervention measures.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical resource demand prediction, in particular to a respiratory infectious disease scale and medical resource demand prediction method based on propagation dynamics. BACKGROUND

[0002] Most infectious diseases are discovered through symptomatic patients. At present, many medical institutions have achieved extensive use of electronic medical record systems. These systems collect a large amount of patient data, including medical history, chief complaint, diagnosis, treatment and laboratory results, and other information. These data provide a wealth of basis for infectious disease outbreak prediction. Some regional and national public health agencies use electronic medical record data for infectious disease monitoring and outbreak prediction. Therefore, based on the mature hospital information management system, by monitoring and analyzing the related symptoms that may be caused by infectious diseases, it can be used for the prediction of infectious disease outbreaks. This involves three links of electronic medical record information extraction, data cleaning, statistics and infectious disease outbreak prediction. At present, the extraction technology of electronic medical record and the monitoring and early warning technology based on symptoms are separate applications, and the application range is too vague and broad. SUMMARY

[0003] The technical problem to be solved by the present application is to provide a respiratory infectious disease scale and medical resource demand prediction method based on propagation dynamics.

[0004] The specific scheme of the present application is as follows:

[0005] A respiratory infectious disease scale and medical resource demand prediction method based on propagation dynamics, comprising the following steps:

[0006] Step 1, collecting daily new / cumulative cases of target respiratory infectious diseases, epidemiological parameters and demographic data of the city / region where the target respiratory infectious diseases occur in the past period as input data;

[0007] Step 2, improving the classical SEIR model, dividing the infected population into two categories: symptomatic infection I s and asymptomatic infection I a ; the symptomatic infected people can further develop into ordinary hospitalized population H s and ICU hospitalized population H c ; the removed population is divided into death removal D and recovery removal R, a total of 8 populations, to obtain an improved classical SEIR model;

[0008] Step 3, inputting the epidemiological parameters of step 1 into the improved classical SEIR model, using the next generation matrix method to calculate the relationship between the basic reproduction number R0 and the effective contact rate β, and using the known R0 to preliminarily estimate the value of the effective contact rate β;

[0009] The Markov chain Monte Carlo algorithm is used to fit and correct the value of the preliminary estimated β by using the daily new / cumulative cases of the target respiratory infectious disease, and the value of β is finally obtained, and then the SEIR prediction model is obtained;

[0010] Step 4, according to the needs of the intervention measures, set the coefficient of β in the SEIR prediction model, the range of the coefficient of β is 0-1, and the set prediction model is obtained;

[0011] Step 5, collect the daily new / cumulative cases of the target respiratory infectious disease in a recent period of time, epidemiological parameters and demographic data of the city / region where the target respiratory infectious disease occurs, and input them into the set prediction model;

[0012] Step 6, the set prediction model outputs the daily case number, hospital bed number, ICU bed number and death number in a future period of time;

[0013] Step 7, according to the prediction results obtained in the last step, the medical resources are dispatched and allocated.

[0014] Further, the epidemiological parameters of step 1 include: basic reproduction number, transmission coefficient of latent period population relative to infected persons, proportion of symptomatic infected persons, death proportion, hospitalization proportion, ICU proportion, latent period, hospitalization days and ICU hospitalization days.

[0015] Further, the data collected for each epidemiological parameter is an interval value, and the input data is obtained by sampling in the interval, and the 95% confidence interval is calculated.

[0016] Further, in step 2, the improved classical SEIR model is calculated by the following differential equations to calculate the change amount of symptomatic infection I s , asymptomatic infection I a , ordinary hospitalized population H s , ICU hospitalized population H c , death removal D and recovery removal R with time t:

[0017]

[0018]

[0019]

[0020]

[0021]

[0022]

[0023]

[0024]

[0025] In the formula, the meaning of each parameter is listed as follows:

[0026]

[0027]

[0028] The beneficial effects of the present application are:

[0029] (1) The next generation matrix method is used to preliminarily estimate the effective contact rate, and then the Markov Chain Monte Carlo (MCMC) algorithm is used to correct the fitting of the parameters, thereby improving the accuracy of the model prediction.

[0030] (2) Most epidemiological parameters cannot obtain an accurate value, and are mostly an interval, so the present application samples them in the interval and calculates the 95% confidence interval, thereby improving the reliability of the prediction.

[0031] (3) The prediction of the demand for medical resources such as the number of hospital beds and the number of ICU beds is increased, which can effectively evaluate whether the medical resource reserve is sufficient, and the resources can be timely adjusted according to the prediction results when necessary.

[0032] (4) The prediction of the infection scale and the demand for medical resources under different intervention measures is realized, which provides a basis for taking appropriate intervention measures. BRIEF DESCRIPTION OF DRAWINGS

[0033] Figure 1 The flow chart of the method of the present application is shown in the figure;

[0034] Figure 2 The cumulative case number from January 1 to January 22 in a certain city is shown in the figure;

[0035] Figure 3 The cumulative case number from January 22 to February 10 in a certain city is shown in the figure;

[0036] Figure 4 The prediction result of the demand for medical resources from January 22 to February 10 in a certain city is shown in the figure. DETAILED DESCRIPTION

[0037] The technical solutions of the present application will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative labor fall within the scope of protection of the present application.

[0038] A method for predicting the scale of respiratory infectious diseases and the demand for medical resources based on propagation dynamics, comprising the following steps:

[0039] Step 1. Data input

[0040] a. Input the daily increase or cumulative number of cases of the respiratory infectious disease

[0041] b. Input the epidemiological parameters of the respiratory infectious disease: parameters related to the transmission force (basic reproduction number, transmission coefficient of latent population relative to infected persons), parameters related to the outcome of the respiratory infectious disease (proportion of symptomatic infected persons, proportion of deaths, proportion of hospitalization, proportion of ICU, number of days in latent period, number of days in hospital, number of days in ICU, etc.), if the parameters are uncertain, an interval can be input

[0042] c. Demographic data of the city or region where the respiratory infectious disease occurs (total population or population of each age group)

[0043] Step 2. Establish an improved propagation dynamics model

[0044] The classical SEIR (Susceptible-Exposed-Infected-Removed) model divides the population into susceptible population (S), exposed population (E), infected population (I) and removed population (R). Based on previous experience and considering different interventions, the classical SEIR model is improved to predict the demand for medical resources. The infected population is divided into symptomatic infected population (I s ) and asymptomatic infected population (I a ); the symptomatic infected persons can further develop into ordinary hospitalized population (H s ) and ICU hospitalized population (H c ); the removed population is divided into death removed (D) and recovery removed (R), totaling 8 populations. The specific model framework is shown in Figure 1 .

[0045] The explanations of the parameters in the model are as follows: β is the effective contact rate, representing the probability of susceptible persons being infected after contacting infected persons; P represents the proportion of inter-compartment transformation, and γ is the rate of inter-compartment transformation. The product of P, γ and the number of persons in the previous compartment is the number of inter-compartment transformations per day, and the specific parameters are shown in Table 1.

[0046] Table 1. Setting of pathogenicity parameters of improved SEIR model

[0047]

[0048]

[0049] The differential equation set is constructed as follows:

[0050]

[0051]

[0052]

[0053]

[0054]

[0055]

[0056]

[0057]

[0058] Step 3. Parameter fitting

[0059] β is the effective contact rate, representing the probability of a susceptible individual being infected after contacting an infected individual, which is the most critical parameter in the transmission dynamics model. According to the improved model, the next generation matrix method is used to calculate the relationship between the basic reproduction number R0 and the effective contact rate β. The known R0 is used to calculate the effective contact rate β, and the specific calculation steps are as follows.

[0060] Extract the core linear equation:

[0061] ΔE = βS(I s + I a + δE) / N

[0062] ΔI s = P ES γ EI E-P SH Y SH I s

[0063] ΔI a = (1-P ES )γ EI E

[0064] Solve the Jacobian matrix:

[0065]

[0066] Solve the largest eigenvalue, which is equal to R0:

[0067]

[0068] Then the Markov Chain Monte Carlo (MCMC) algorithm is used to fit and correct it using the daily number of cases.

[0069] Step 4. Result prediction

[0070] a. Input the number of days you want to predict

[0071] b. Set the number of times of sampling for uncertain parameters, usually set to 1000 times, in order to get more stable results.

[0072] c. Adjust the effective contact rate β and set the isolation rate of infected persons to represent different interventions. The default state of the effective contact rate coefficient is 1, and the isolation rate is 0, which is used to simulate the development of the infection situation when no measures are taken.

[0073] d. Under the improved SEIR model, the results are constantly output, providing a basis for dealing with respiratory infectious diseases. The output results are the number of cases per day, the number of hospital beds per day, the number of ICU beds per day, the number of deaths per day, etc.

[0074] Example: This example takes the infection situation of a new respiratory infectious disease in a city from January 1 to February 10 of a certain year as an example, and establishes an improved transmission dynamics model. With the strict control measures taken on January 22 as the boundary, the daily cumulative case numbers from January 1 to January 22 are used for parameter fitting, the infection situation of a new respiratory infectious disease for 20 days from January 22 to February 10 is simulated by taking strict control measures, and the actual cumulative case numbers are compared.

[0075] Data sources: The daily cumulative case numbers of a new respiratory infectious disease in a city from January 1 to February 10 of a certain year reported in published literature, and the epidemiological parameters of an early new respiratory infectious disease. In the actual use process, these parameters can be obtained through field epidemiological investigation and clinical data.

[0076] Analysis process description:

[0077] (1) Read the data, including daily cumulative case number data and epidemiological parameter data, see Tables 2-3:

[0078] Table 2 Daily cumulative and new case number data

[0079]

[0080] Table 3 Related epidemiological parameters

[0081]

[0082] (2) Construct an improved transmission dynamics model

[0083] (3) Using the daily cumulative cases from January 1 to January 22, the parameter fitting was performed, the effective contact rate was preliminarily estimated using the next-generation matrix method, and the MCMC algorithm was used for fitting correction. The fitting results are shown in Figure 2

[0084] (4) The simulation of strict control measures was performed to predict the infection of a new respiratory infectious disease from January 22 to February 10 for 20 days, and the actual cumulative cases were compared. The demand for hospital beds and ICU beds was predicted. The results are shown in Figures 3-4 .

[0085] Conclusion:

[0086] Previous studies have difficulty in solving the following problems: a. How to calculate a more accurate value of the key parameter effective contact rate b. The epidemiological parameters of respiratory infectious diseases are not fixed values. How to reduce the influence of their uncertainty on model prediction c. Prediction of the demand for hospital beds and ICU beds d. Comprehensive consideration of intervention measures while predicting the demand for medical resources. This method uses the next-generation matrix method combined with MCMC to more accurately estimate the effective contact rate, uses sampling to reduce the influence of the uncertainty of epidemiological parameters on the prediction results, and integrates the intervention measures and the demand for medical resources into the traditional transmission dynamics model. In the face of new emerging respiratory infectious diseases, the infection situation can be more quickly and efficiently evaluated, the intervention measures needed to be taken can be evaluated, and the medical resources can be reasonably allocated.

[0087] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for predicting the scale and medical resource demand of respiratory infectious diseases based on transmission dynamics, characterized in that, Includes the following steps: Step 1: Collect the daily new / cumulative number of cases of the target respiratory infectious disease, epidemiological parameters, and demographic data of the cities / regions where the target respiratory infectious disease occurs as input data over a period of time. Step 2: Improve the classic SEIR model by dividing the infected population into symptomatic infections. and asymptomatic infection Two categories: symptomatic infected individuals who further develop into general hospitalized patients. And ICU inpatients; the removed population was divided into death-removed population (D) and recovery-removed population (R), totaling 8 populations, resulting in an improved classic SEIR model; Step 3: Input the epidemiological parameters from Step 1 into the improved classical SEIR model, and use the next-generation matrix method to calculate the relationship between the basic reproduction number R0 and the effective contact rate β. Use the known R0 to make a preliminary estimate of the value of the effective contact rate β. The Markov chain Monte Carlo algorithm is used to fit and correct the value of β by using the daily new / cumulative number of cases of the target respiratory infectious disease, and finally obtain the value of β, thus obtaining the SEIR prediction model; Step 4: According to the needs of the intervention measures, set the coefficient of β in the SEIR prediction model. The coefficient of β ranges from 0 to 1, and the prediction model is set up. Step 5: Collect the daily new / cumulative number of cases of the target respiratory infectious disease, epidemiological parameters, and demographic data of the cities / regions where the target respiratory infectious disease occurs over a recent period and input them into the pre-set prediction model; Step 6: The set-up prediction model outputs the daily number of cases, number of hospital beds, number of ICU beds, and number of deaths for a future period of time; Step 7: Allocate and schedule medical resources based on the prediction results obtained in the previous step.

2. The method for predicting the scale and medical resource demand of respiratory infectious diseases based on transmission dynamics according to claim 1, characterized in that, The epidemiological parameters in step 1 include: basic reproduction number, transmission coefficient of the incubation period population relative to the infected population, proportion of symptomatic infected individuals, mortality rate, hospitalization rate, ICU rate, incubation period days, hospitalization days, and ICU hospitalization days.

3. The method for predicting the scale and medical resource demand of respiratory infectious diseases based on transmission dynamics according to claim 2, characterized in that, The data collected for each epidemiological parameter are interval values. Input data are obtained by sampling within the interval and calculating the 95% confidence interval.

4. The method for predicting the scale and medical resource demand of respiratory infectious diseases based on transmission dynamics according to claim 1, characterized in that, In step 2, the improved classical SEIR model calculates symptomatic infection using the following differential equation. Asymptomatic infection General hospitalized population ICU inpatients The changes in death removal (D) and recovery removal (R) over time t: ； ； ； ； ； ； ； ； In the formula, the meanings of each parameter are as follows:

Images