Epidemic situation prediction method and device, computer equipment and storage medium
Through the combination of multi-channel adaptive Fourier decomposition and Transformer model, the limitations of existing epidemic prediction methods in dealing with nonlinear and non-stationary data are solved, and higher epidemic prediction accuracy and flexibility are achieved.
Patent Information
- Application Number
- CN202411828201.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-12
- Publication Date
- 2025-05-13
AI Technical Summary
Existing epidemic prediction methods show limitations when dealing with complex nonlinear and non-stationary data, making it difficult to accurately capture the dynamic changes of the epidemic.
Multi-channel adaptive Fourier decomposition is used to decompose multi-channel epidemic data, extract epidemic characteristics, and input these characteristics into the epidemic prediction model built on the Transformer network for prediction.
Through the combination of multi-channel adaptive Fourier decomposition and Transformer model, the nonlinear and non-stationary characteristics of the data can be better captured, improving the accuracy and flexibility of epidemic prediction.
Smart Images
Figure CN119993550A_ABST
Abstract
Description
Technical Field
[0001] The present application belongs to the technical field of epidemic prediction, and in particular, relates to an epidemic prediction method, apparatus, computer equipment and storage medium. Background Art
[0002] Current epidemic prediction methods mainly rely on traditional statistical models, machine learning models, and time series analysis methods. However, these models or analysis methods show limitations when dealing with complex nonlinear and non-stationary data, and it is difficult to accurately capture the dynamic changes of the epidemic. Traditional statistical models such as ARIMA and SEIR models assume that the data is linear and stationary, and therefore cannot cope with the nonlinear characteristics of epidemic spread. Machine learning and deep learning models, such as support vector machines and LSTM, although they can process nonlinear data, have difficulty effectively capturing the correlation between channels when processing multi-channel data, and deep learning models may encounter the problem of gradient vanishing when processing long sequences. In addition, time series analysis methods such as wavelet transform and empirical mode decomposition, although they can capture the short-term and long-term changes of non-stationary signals, usually rely on fixed basis functions, lack adaptability, and have deficiencies in the joint analysis of multi-channel signals.
[0003] Due to the above-mentioned deficiencies or defects in traditional statistical models, machine learning models and time series analysis methods, the epidemic prediction methods based on these models or analysis methods cannot achieve better prediction results when implemented. Summary of the invention
[0004] The present application provides an epidemic prediction method, apparatus, computer device and storage medium, which aim to address the above-mentioned deficiencies or defects of traditional statistical models, machine learning models and time series analysis methods.
[0005] In a first aspect, the present application provides a method for predicting an epidemic situation. In some embodiments, the method includes:
[0006] Obtain multi-channel epidemic data for a specified area;
[0007] Preprocessing the multi-channel epidemic data; the preprocessing includes standardization processing;
[0008] Use multi-channel adaptive Fourier decomposition to decompose the pre-processed multi-channel epidemic data and extract epidemic characteristics;
[0009] Input the epidemic characteristics into the pre-trained epidemic prediction model based on the Transformer network to obtain the epidemic prediction results output by the epidemic prediction model.
[0010] In some embodiments, multi-channel adaptive Fourier decomposition is used to decompose the pre-processed multi-channel epidemic data to extract epidemic characteristics, including:
[0011] Multi-channel adaptive Fourier decomposition is used to perform frequency decomposition on each channel in the preprocessed multi-channel epidemic data, extract the main frequency components and fluctuation characteristics in each channel, and obtain the epidemic characteristics.
[0012] In some embodiments, the training process of the epidemic prediction model includes:
[0013] Obtain a historical multi-channel epidemic data set for a specified area, where the historical multi-channel epidemic data set includes multiple pre-processed historical multi-channel epidemic data;
[0014] Use multi-channel adaptive Fourier decomposition to extract historical epidemic characteristics from each historical multi-channel epidemic data, and build a training data set based on the extracted historical epidemic characteristics;
[0015] The epidemic prediction model built based on the Transformer network is iteratively trained using the training data set until the preset end conditions are met to obtain a trained epidemic prediction model.
[0016] In some embodiments, after obtaining the trained epidemic prediction model, the method further includes:
[0017] Collect new multi-channel epidemic data in designated areas;
[0018] Dynamically update the parameters of the epidemic prediction model based on new multi-channel epidemic data.
[0019] In some embodiments, the designated area is a designated region of a designated country; the multi-channel epidemic data includes epidemic data at multiple stages of epidemic development in the designated area; the epidemic data at each stage of epidemic development include one or more of the daily number of newly confirmed cases, number of hospitalizations, utilization rate of intensive care units, vaccination rate, number of tests, implementation of public health intervention measures, meteorological data and population mobility data; epidemic characteristics include the main frequency component characteristics representing the periodic changes of the epidemic and the fluctuation characteristics reflecting the short-term fluctuations and trend changes of the epidemic; the epidemic forecast results include the number of new cases and hospitalization rate in the designated area in the next week.
[0020] In some embodiments, the designated area is multiple countries and regions in the world; the multi-channel epidemic data includes epidemic data of multiple epidemic waves in various countries and regions; the epidemic data of each epidemic wave includes one or more of the daily number of newly confirmed cases, mortality rate, vaccination rate, number of tests, implementation of public health intervention measures, meteorological data and population mobility data; epidemic characteristics include the main frequency component characteristics representing the cyclical changes of the epidemic and the fluctuation characteristics reflecting the short-term drastic changes in the epidemic; the epidemic prediction results include the number of new cases and epidemic trends in each country and region in the next month.
[0021] In some embodiments, the designated area is a designated country; the multi-channel epidemic data includes epidemic data at multiple stages of epidemic development of the designated country; the epidemic data at each stage of epidemic development include one or more of the daily number of newly confirmed cases, vaccination rates, vaccine types and their effectiveness, number of severe cases and deaths, implementation of public health intervention measures and population mobility data; epidemic characteristics include main frequency component characteristics representing periodic changes of the epidemic, fluctuation characteristics reflecting short-term drastic changes in the epidemic, and vaccination impact characteristics reflecting the impact of vaccination on the spread of the epidemic; epidemic forecast results include the development of the epidemic in the designated country in the next two months.
[0022] In a second aspect, the present application provides an epidemic prediction device. In some embodiments, the device includes:
[0023] Data acquisition module, used to obtain multi-channel epidemic data of a specified area;
[0024] A preprocessing module is used to preprocess the multi-channel epidemic data; the preprocessing includes standardization processing;
[0025] A feature extraction module is used to decompose the pre-processed multi-channel epidemic data using multi-channel adaptive Fourier decomposition to extract epidemic characteristics;
[0026] The epidemic prediction module is used to input epidemic characteristics into a pre-trained epidemic prediction model based on the Transformer network to obtain the epidemic prediction results output by the epidemic prediction model.
[0027] In a third aspect, the present application provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the epidemic prediction method provided in the above embodiment is implemented.
[0028] In a fourth aspect, the present application provides a computer device, comprising: one or more processors; a memory; and one or more computer programs, the processor and the memory are connected via a bus, wherein the one or more computer programs are stored in the memory and are configured to be executed by the one or more processors, characterized in that when the processor executes the computer program, the epidemic prediction method provided in the above embodiment is implemented.
[0029] In this application, multi-channel adaptive Fourier decomposition is applied to epidemic prediction. By adaptively decomposing multi-channel epidemic data, the main frequency components and dynamic features of each channel used for epidemic prediction can be extracted. This method can better capture the nonlinear and non-stationary characteristics of the data, and it does not rely on fixed basis functions. It is more flexible and can adaptively adjust the feature extraction process according to the dynamic changes of epidemic data. On this basis, this application adopts the Transformer model to predict the epidemic. The Transformer model can fully utilize the features extracted by multi-channel adaptive Fourier decomposition through the self-attention mechanism with powerful long sequence modeling capabilities to achieve accurate prediction of epidemic trends. This application integrates multi-channel adaptive Fourier decomposition and Transformer model into the same framework. When processing long-span, multi-channel epidemic data, it performs well, can more accurately capture the complex dynamic changes in the spread of the epidemic, and improves the accuracy of epidemic prediction results. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] Figure 1 It is a flowchart of an epidemic prediction method provided in one embodiment of the present application.
[0031] Figure 2 This is a functional module block diagram of an epidemic prediction device provided in one embodiment of the present application.
[0032] Figure 3 It is a specific structural block diagram of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION
[0033] In order to make the purpose, technical solution and beneficial effects of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0034] In order to illustrate the technical solution described in this application, a specific embodiment is provided below for illustration.
[0035] See also Figure 1, is a flow chart of an epidemic prediction method provided in an embodiment of the present application. This embodiment mainly uses the epidemic prediction method applied to a computer device as an example for illustration. The epidemic prediction method provided in an embodiment of the present application includes the following steps:
[0036] S101. Obtain multi-channel epidemic data of a specified area.
[0037] The specific content of the designated area and multi-channel epidemic data is related to the prediction task. The designated area can be part or all of the geographical area of one or more countries. Multi-channel epidemic data includes various types of epidemic data collected over a period of time. In some examples, multi-channel epidemic data may include the number of newly confirmed cases, hospitalizations, intensive care unit utilization, vaccination rates, number of tests, implementation of public health intervention measures, meteorological data, and population mobility data per day over a period of time.
[0038] S102. Preprocess multi-channel epidemic data.
[0039] Before feature extraction, multi-channel epidemic data needs to be preprocessed. The preprocessing operation is mainly used to improve the data quality of multi-channel epidemic data, enhance the generalization ability, and help improve the performance of the epidemic prediction model. Accordingly, preprocessing can include missing value processing, outlier detection and processing, and standardization processing (or normalization processing). Among them, the operation of missing value processing can be to delete records with missing values and fill missing values, such as using the mean, median or mode to estimate missing values. The operation of outlier detection can be to identify outliers and process outliers through statistical methods, such as deleting, replacing or retaining outliers based on actual requirements. The operation of standardization processing is to scale the data to a specific range, such as [0,1], using methods such as minimum-maximum normalization and Z-score normalization, which helps to make the epidemic prediction model better generalize to new data.
[0040] S103. Decompose the preprocessed multi-channel epidemic data using multi-channel adaptive Fourier decomposition to extract epidemic characteristics.
[0041] In one embodiment, multi-channel adaptive Fourier decomposition is used to decompose the preprocessed multi-channel epidemic data to extract epidemic characteristics, including: using multi-channel adaptive Fourier decomposition to perform frequency decomposition on each channel in the preprocessed multi-channel epidemic data, extracting the main frequency components and fluctuation characteristics in each channel, and obtaining epidemic characteristics.
[0042] Among them, the operation of decomposing the pre-processed multi-channel epidemic data using multi-channel adaptive Fourier decomposition can be to perform frequency decomposition on the epidemic data of each channel and extract the main frequency components and fluctuation characteristics in the data as epidemic characteristics.
[0043] The main frequency component in the epidemic characteristics indicates the periodic changes in the epidemic. This characteristic represents the long-term spread trend of the epidemic, while the fluctuation characteristics can reflect the short-term fluctuations and trend changes of the epidemic.
[0044] S104. Input the epidemic characteristics into a pre-trained epidemic prediction model based on the Transformer network to obtain the epidemic prediction results output by the epidemic prediction model.
[0045] After extracting the epidemic characteristics through multi-channel adaptive Fourier decomposition, they are input into the Transformer model, i.e., the epidemic prediction model, to predict the epidemic spread trend and obtain the epidemic prediction results. It should be noted that in different prediction tasks, the designated area and multi-channel epidemic data may be different, and the specific content of the corresponding epidemic prediction results will also be different.
[0046] The epidemic prediction model is pre-trained. In one embodiment, the training process of the epidemic prediction model includes:
[0047] a. Obtain the historical multi-channel epidemic data set for the specified area.
[0048] In this step, the historical multi-channel epidemic data set includes a plurality of pre-processed historical multi-channel epidemic data. The historical multi-channel epidemic data refers to the multi-channel epidemic data generated in the past in the specified area used to train the epidemic prediction model. For the specific operation of the pre-processing, please refer to the introduction in the above embodiment, which will not be repeated here.
[0049] b. Use multi-channel adaptive Fourier decomposition to extract historical epidemic characteristics from each historical multi-channel epidemic data, and build a training data set based on the extracted historical epidemic characteristics.
[0050] In this step, historical epidemic characteristics refer to epidemic characteristics extracted from historical multi-channel epidemic data. For the content of epidemic characteristics, please refer to the relevant introduction in the above embodiment. The specific operation of extracting historical epidemic characteristics using multi-channel adaptive Fourier decomposition can be referred to the description of the extraction process of epidemic characteristics in the above embodiment, which will not be repeated here.
[0051] Each extracted historical epidemic feature can be used as a training sample, and multiple historical epidemic features can be used as training data sets.
[0052] c. Use the training data set to iteratively train the epidemic prediction model built based on the Transformer network until the preset end conditions are met to obtain a trained epidemic prediction model.
[0053] In this step, a batch of training samples can be called from the training data set each time to train the epidemic prediction model to improve training efficiency. The epidemic prediction model will model the epidemic spread trend based on the historical epidemic characteristics in the input training samples, capture long-term dependencies through the self-attention mechanism, and learn the relationship between epidemic characteristics and epidemic spread trends.
[0054] Furthermore, after obtaining the trained epidemic prediction model, the method also includes: collecting new multi-channel epidemic data in a designated area; and dynamically updating the parameters of the epidemic prediction model based on the new multi-channel epidemic data.
[0055] This embodiment can dynamically adapt to changes in epidemic data and achieve real-time updating and self-adjustment of the model by continuously inputting new epidemic data. It should be pointed out that during the updating process, the adaptive decomposition capability of the multi-channel adaptive Fourier decomposition can enable the epidemic prediction model to maintain high prediction accuracy and stability when the epidemic data changes significantly.
[0056] The epidemic prediction method provided in the above embodiment is described in more detail below using multiple different prediction tasks as examples.
[0057] In one embodiment, the prediction task is a regional COVID-19 epidemic prediction, such as predicting the number of new cases and hospitalization rate in a certain area in the next week. Accordingly, the designated area is specifically a designated area in a designated country, and the multi-channel epidemic data of the designated area specifically includes the epidemic data of multiple epidemic development stages in the designated area, wherein there are multiple epidemic development stages, which may include initial outbreak, accelerated spread, peak period, relief period, trough period, and end or sporadic stage; the epidemic data of each epidemic development stage include one or more of the daily number of newly confirmed cases, hospitalizations, intensive care unit utilization rate, vaccination rate, number of tests, implementation of public health intervention measures, meteorological data and population mobility data. The epidemic characteristics specifically include the main frequency component characteristics representing the periodic changes of the epidemic and the fluctuation characteristics reflecting the short-term fluctuations and trend changes of the epidemic; the epidemic prediction results include the number of new cases and hospitalization rate in the designated area in the next week.
[0058] In this embodiment, the epidemic data of multiple channels such as the number of newly confirmed cases, the number of hospitalized patients, the utilization rate of intensive care units, etc. of COVID-19 in a specific region at different stages of the epidemic development in the past can be used as multi-channel epidemic data; the data needs to be preprocessed first, such as standardization, and then the preprocessed multi-channel epidemic data is decomposed using multi-channel adaptive Fourier decomposition to extract the main frequency components and fluctuation characteristics of each channel as epidemic characteristics; finally, the epidemic characteristics are input into the trained Transformer model for epidemic prediction, and the number of new cases and hospitalization rate in the next week are predicted.
[0059] This embodiment is significantly better than the epidemic prediction method based on the LSTM model and the SIR model in terms of prediction accuracy, especially in capturing the stage of rapid changes in the epidemic. In this embodiment, the frequency components extracted by multi-channel adaptive Fourier decomposition can better reflect the volatility of the epidemic, and the Transformer model can effectively model data with a long time span. The combination of the two can effectively improve the accuracy of the epidemic prediction results.
[0060] In one embodiment, the prediction task is global epidemic monitoring and prediction, such as predicting the number of new cases and epidemic trends in the world in the next month. Accordingly, the designated area specifically refers to multiple countries and regions in the world; the multi-channel epidemic data of the designated area specifically includes the epidemic data of multiple epidemic waves in various countries and regions; the epidemic data of each epidemic wave includes one or more of the daily number of newly confirmed cases, mortality rate, vaccination rate, number of tests, implementation of public health intervention measures, meteorological data and population mobility data; the epidemic characteristics include the main frequency component characteristics representing the cyclical changes of the epidemic and the fluctuation characteristics reflecting the short-term drastic changes of the epidemic. The epidemic prediction results include the number of new cases and epidemic trends in various countries and regions in the next month.
[0061] In this embodiment, COVID-19 datasets from multiple countries and regions around the world can be used as multi-channel epidemic data (including epidemic data from multiple epidemic waves in various countries and regions over the past period of time); the data needs to be preprocessed first, such as standardization, and then the preprocessed multi-channel epidemic data is decomposed using multi-channel adaptive Fourier decomposition to extract the main frequency components and fluctuation characteristics of each channel as epidemic characteristics; finally, the epidemic characteristics are input into the trained Transformer model for epidemic prediction, and the number of new cases and epidemic trends in each country and region in the next month are predicted.
[0062] This embodiment shows high prediction accuracy on a global scale and can accurately predict the epidemic trends in multiple countries and regions, which is better than the method based on ARIMA model and convolutional neural network for epidemic prediction. In addition, multi-channel adaptive Fourier decomposition shows strong feature extraction capabilities in the decomposition of multi-channel and multi-country data, and the Transformer model can integrate global data to predict more accurate epidemic prediction results.
[0063] In one embodiment, the prediction task is to predict the epidemic situation under the influence of vaccination, such as predicting the development of the epidemic situation in a certain country in the next two months. Accordingly, the designated area is specifically a designated country; the multi-channel epidemic data of the designated area specifically includes the epidemic data of multiple epidemic development stages of the epidemic situation in the designated country; the epidemic data of each epidemic development stage includes one or more of the daily number of newly confirmed cases, vaccination rate, vaccine type and its effectiveness, number of severe cases and deaths, implementation of public health intervention measures and population mobility data; the epidemic characteristics include the main frequency component characteristics representing the periodic changes of the epidemic, the fluctuation characteristics reflecting the short-term drastic changes of the epidemic, and the vaccination impact characteristics reflecting the impact of vaccination on the spread of the epidemic; the epidemic prediction results include the development of the epidemic situation in the next two months in the designated country.
[0064] In this embodiment, the epidemic data of multiple channels such as the number of newly confirmed cases, number of hospitalized patients, utilization rate of intensive care units, vaccination rate, types of vaccines and their effectiveness of COVID-19 in a specified country at different stages of the epidemic development in the past can be used as multi-channel epidemic data; the data needs to be preprocessed first, such as standardization, and then the preprocessed multi-channel epidemic data is decomposed using multi-channel adaptive Fourier decomposition to extract the main frequency components and fluctuation characteristics of each channel as epidemic characteristics; finally, the epidemic characteristics are input into the trained Transformer model for epidemic prediction, and the epidemic development of the country in the next two months is predicted.
[0065] This embodiment can better capture the dynamic impact of vaccination on the epidemic. The prediction results show that in areas with high vaccination rates, the spread of the epidemic is effectively controlled, and the prediction accuracy is better than the method of using support vector machines for epidemic prediction. In addition, the multi-dimensional features of vaccination and epidemic data are decomposed through multi-channel adaptive Fourier decomposition, which enables the Transformer model to make efficient long-term predictions based on these features.
[0066] The present application also provides an epidemic prediction device. In some embodiments, see Figure 2 , the device comprises:
[0067] A data acquisition module 10 is used to acquire multi-channel epidemic data of a specified area;
[0068] A preprocessing module 20 is used to preprocess the multi-channel epidemic data; the preprocessing includes standardization processing;
[0069] A feature extraction module 30 is used to decompose the pre-processed multi-channel epidemic data using multi-channel adaptive Fourier decomposition to extract epidemic characteristics;
[0070] The epidemic prediction module 40 is used to input the epidemic characteristics into a pre-trained epidemic prediction model constructed based on the Transformer network, and obtain the epidemic prediction results output by the epidemic prediction model.
[0071] In some embodiments, the feature extraction module 30 includes:
[0072] A decomposition unit, used for performing frequency decomposition on each channel in the preprocessed multi-channel epidemic data using multi-channel adaptive Fourier decomposition;
[0073] The extraction unit is used to extract the main frequency components and fluctuation characteristics in each channel to obtain the epidemic characteristics.
[0074] In some embodiments, the device further includes a training module for an epidemic prediction model, the training module including:
[0075] A data set acquisition unit, used to acquire a historical multi-channel epidemic data set of a specified area, where the historical multi-channel epidemic data set includes a plurality of pre-processed historical multi-channel epidemic data;
[0076] A data set construction unit, used for extracting historical epidemic characteristics from each historical multi-channel epidemic data by using multi-channel adaptive Fourier decomposition, and constructing a training data set based on the extracted historical epidemic characteristics;
[0077] The iterative training unit is used to iteratively train the epidemic prediction model built based on the Transformer network using the training data set until the preset end condition is met to obtain a trained epidemic prediction model.
[0078] In some embodiments, the device further includes a model updating module. The model updating module includes:
[0079] A collection unit, used to collect new multi-channel epidemic data in a specified area;
[0080] The parameter updating unit is used to dynamically update the parameters of the epidemic prediction model based on new multi-channel epidemic data.
[0081] In some embodiments, the designated area is a designated region of a designated country; the multi-channel epidemic data includes epidemic data at multiple stages of epidemic development in the designated area; the epidemic data at each stage of epidemic development include one or more of the daily number of newly confirmed cases, number of hospitalizations, utilization rate of intensive care units, vaccination rate, number of tests, implementation of public health intervention measures, meteorological data and population mobility data; epidemic characteristics include the main frequency component characteristics representing the periodic changes of the epidemic and the fluctuation characteristics reflecting the short-term fluctuations and trend changes of the epidemic; the epidemic forecast results include the number of new cases and hospitalization rate in the designated area in the next week.
[0082] In some embodiments, the designated area is multiple countries and regions in the world; the multi-channel epidemic data includes epidemic data of multiple epidemic waves in various countries and regions; the epidemic data of each epidemic wave includes one or more of the daily number of newly confirmed cases, mortality rate, vaccination rate, number of tests, implementation of public health intervention measures, meteorological data and population mobility data; epidemic characteristics include the main frequency component characteristics representing the cyclical changes of the epidemic and the fluctuation characteristics reflecting the short-term drastic changes in the epidemic; the epidemic prediction results include the number of new cases and epidemic trends in each country and region in the next month.
[0083] In some embodiments, the designated area is a designated country; the multi-channel epidemic data includes epidemic data at multiple stages of epidemic development of the designated country; the epidemic data at each stage of epidemic development include one or more of the daily number of newly confirmed cases, vaccination rates, vaccine types and their effectiveness, number of severe cases and deaths, implementation of public health intervention measures and population mobility data; epidemic characteristics include main frequency component characteristics representing periodic changes of the epidemic, fluctuation characteristics reflecting short-term drastic changes in the epidemic, and vaccination impact characteristics reflecting the impact of vaccination on the spread of the epidemic; epidemic forecast results include the development of the epidemic in the designated country in the next two months.
[0084] The epidemic prediction device provided in one embodiment of the present application and the epidemic prediction method provided in the present application belong to the same concept. The specific implementation process is detailed in the full text of the specification and will not be repeated here.
[0085] An embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the epidemic prediction method provided in an embodiment of the present application.
[0086] Figure 3A specific structural block diagram of a computer device provided in an embodiment of the present application is shown, and the computer device 100 includes: one or more processors 101, a memory 102, and one or more computer programs, wherein the processor 101 and the memory 102 are connected via a bus, the one or more computer programs are stored in the memory 102, and are configured to be executed by the one or more processors 101, and when the processor 101 executes the computer program, the epidemic prediction method provided in an embodiment of the present application is implemented.
[0087] It should be understood that each step in each embodiment of the present application is not necessarily performed in sequence according to the order indicated by the step number. Unless there is clear explanation in this article, the execution of these steps does not have strict order restriction, and these steps can be performed in other orders. Moreover, at least a part of the steps in each embodiment can include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a part of other steps or sub-steps or stages of other steps.
[0088] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink), DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
[0089] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0090] The above embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for those of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.
Claims
1. A method for predicting an epidemic, characterized in that: The method comprises: Obtain multi-channel epidemic data for a specified area; Preprocessing the multi-channel epidemic data; the preprocessing includes standardization processing; Use multi-channel adaptive Fourier decomposition to decompose the pre-processed multi-channel epidemic data and extract epidemic characteristics; The epidemic characteristics are input into a pre-trained epidemic prediction model based on a Transformer network to obtain the epidemic prediction results output by the epidemic prediction model.
2. The method according to claim 1, characterized in that Multi-channel adaptive Fourier decomposition is used to decompose the pre-processed multi-channel epidemic data and extract epidemic characteristics, including: Multi-channel adaptive Fourier decomposition is used to perform frequency decomposition on each channel in the preprocessed multi-channel epidemic data, and the main frequency components and fluctuation characteristics in each channel are extracted to obtain the epidemic characteristics.
3. The method according to claim 1, characterized in that The training process of the epidemic prediction model includes: Acquire a historical multi-channel epidemic data set of the designated area, wherein the historical multi-channel epidemic data set includes a plurality of pre-processed historical multi-channel epidemic data; Use multi-channel adaptive Fourier decomposition to extract historical epidemic characteristics from each historical multi-channel epidemic data, and build a training data set based on the extracted historical epidemic characteristics; The training data set is used to iteratively train the epidemic prediction model built based on the Transformer network until a preset end condition is met, thereby obtaining a trained epidemic prediction model.
4. The method according to claim 3, characterized in that After obtaining the trained epidemic prediction model, the method further includes: Collect new multi-channel epidemic data of the designated area; Dynamically update the parameters of the epidemic prediction model based on the new multi-channel epidemic data.
5. The method according to claim 1, characterized in that The designated area is a designated region of a designated country; The multi-channel epidemic data includes epidemic data of multiple epidemic development stages of the designated area; the epidemic data of each epidemic development stage includes one or more of the daily number of newly confirmed cases, number of hospitalizations, utilization rate of intensive care units, vaccination rate, number of tests, implementation of public health intervention measures, meteorological data and population mobility data; The epidemic characteristics include the main frequency component characteristics indicating the periodic changes of the epidemic and the fluctuation characteristics reflecting the short-term fluctuations and trend changes of the epidemic; The epidemic forecast results include the number of new cases and hospitalization rate in the designated area in the next week.
6. The method according to claim 1, characterized in that The designated areas are multiple countries and regions around the world; The multi-channel epidemic data includes epidemic data of multiple epidemic waves in various countries and regions; the epidemic data of each epidemic wave includes one or more of the daily number of newly confirmed cases, mortality rate, vaccination rate, number of tests, implementation of public health intervention measures, meteorological data and population mobility data; The epidemic characteristics include the main frequency component characteristics representing the periodic changes of the epidemic and the fluctuation characteristics reflecting the short-term drastic changes of the epidemic; The epidemic forecast results include the number of new cases and epidemic trends in each country and region in the next month.
7. The method according to claim 1, characterized in that The designated area is a designated country; The multi-channel epidemic data includes epidemic data at multiple epidemic development stages of the epidemic in the designated country; the epidemic data at each epidemic development stage includes one or more of the daily number of newly confirmed cases, vaccination rate, vaccine types and their effectiveness, number of severe cases and deaths, implementation of public health intervention measures and population mobility data; The epidemic characteristics include main frequency component characteristics indicating periodic changes in the epidemic, fluctuation characteristics reflecting short-term dramatic changes in the epidemic, and vaccination impact characteristics reflecting the impact of vaccination on the spread of the epidemic; The epidemic forecast results include the development of the epidemic in the designated country in the next two months.
8. An epidemic prediction device, characterized in that: The device comprises: Data acquisition module, used to obtain multi-channel epidemic data of a specified area; A preprocessing module, used for preprocessing the multi-channel epidemic data; the preprocessing includes standardization processing; A feature extraction module is used to decompose the pre-processed multi-channel epidemic data using multi-channel adaptive Fourier decomposition to extract epidemic characteristics; The epidemic prediction module is used to input the epidemic characteristics into a pre-trained epidemic prediction model based on the Transformer network to obtain the epidemic prediction results output by the epidemic prediction model.
9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the epidemic prediction method according to any one of claims 1 to 7 are implemented.
10. A computer device, characterized in that: include: one or more processors; Memory; as well as One or more computer programs, the processor and the memory are connected via a bus, wherein the one or more computer programs are stored in the memory and are configured to be executed by the one or more processors, wherein the processor implements the steps of the epidemic prediction method according to any one of claims 1 to 7 when executing the computer program.