Coronavirus prediction method and device based on SEAIQR-PSO-LSTM hybrid model

Abstract

The present invention discloses a coronavirus prediction method and device based on a SEAIQR-PSO-LSTM hybrid model, and relates to the field of public health technology. A specific implementation of the method includes: taking a number sequence of multiple target populations associated with coronavirus in the most recent multiple historical statistical periods as input, training a SEAIQR model and a PSO-LSTM model; constructing a SEAIQR-PSO-LSTM hybrid model by performing linear regression modeling on the output results of the SEAIQR model and the PSO-LSTM model, and using the least squares method to optimize the linear regression parameters to train the SEAIQR-PSO-LSTM hybrid model; the SEAIQR-PSO-LSTM hybrid model outputs the prediction results to obtain the number of target populations in at least one future statistical period; determining the impact of multiple key parameters related to coronavirus on the spread of coronavirus based on the trained SEAIQR-PSO-LSTM hybrid model and the prediction results, and determining the medical resource allocation strategy for the future statistical period based on the prediction results. This implementation can improve the prediction accuracy of the model.

Description

Technical Field

[0001] The present invention relates to the field of public health technology, and in particular to a coronavirus prediction method and device based on a SEAIQR-PSO-LSTM hybrid model. Background Art

[0002] Currently, some teams use the SEIR model to predict the spread of the coronavirus. The SEAIQR model is a modification and optimization of the classic SEIR model, which is more in line with the characteristics of the coronavirus and can show the role of non-drug intervention measures in curbing the spread of the virus. However, the SEAIQR model cannot automatically make adaptive adjustments based on historical fluctuation data. Summary of the invention

[0003] In view of this, the embodiment of the present invention provides a coronavirus prediction method and device based on the SEAIQR-PSO-LSTM hybrid model, which integrates the SEAIQR model and the PSO-LSTM model through a linear regression model, thereby combining the advantages of the SEAIQR and PSO-LSTM models to achieve adaptive adjustment of the prediction model with statistical data and improve prediction accuracy.

[0004] To achieve the above objectives, according to one aspect of the present invention, a coronavirus prediction method based on the SEAIQR-PSO-LSTM hybrid model is provided.

[0005] The coronavirus prediction method based on the SEAIQR-PSO-LSTM hybrid model of the embodiment of the present invention includes: constructing a SEAIQR-PSO-LSTM hybrid model including a SEAIQR model and a PSO-LSTM model; wherein, taking a number sequence of multiple target populations associated with coronavirus in multiple recent historical statistical periods as input, training the SEAIQR model and the PSO-LSTM model; constructing the SEAIQR-PSO-LSTM hybrid model by performing linear regression modeling on the output results of the SEAIQR model and the PSO-LSTM model, and training the SEAIQR-PSO-LSTM hybrid model by optimizing the linear regression parameters using the least squares method; outputting the prediction result by the SEAIQR-PSO-LSTM hybrid model to obtain the number of target populations in at least one future statistical period; determining the influence of multiple key parameters related to coronavirus on the spread of coronavirus based on the trained SEAIQR-PSO-LSTM hybrid model and the prediction result, and determining the medical resource allocation strategy for the future statistical period based on the prediction result.

[0006] Optionally, a data set is created using historical statistical data of the coronavirus to train the SEAIQR model and the PSO-LSTM model, and the SEAIQR-PSO-LSTM hybrid model is constructed by performing linear regression modeling on the output results of the two, including: using least squares method to optimize linear regression parameters to train the SEAIQR-PSO-LSTM hybrid model.

[0007] Optionally, the PSO-LSTM model includes an LSTM model and a PSO algorithm; and the method further includes: before completing the training of the SEAIQR-PSO-LSTM hybrid model, optimizing multiple hyperparameters in the LSTM model as PSO particles in the PSO algorithm.

[0008] Optionally, the optimization of multiple hyperparameters in the LSTM model as PSO particles in the PSO algorithm includes the following steps performed in sequence: dividing the pre-collected experimental data into a training set, a validation set and a test set; wherein the training set and the validation set constitute the data set; taking the hyperparameters of the LSTM model as the optimization object and initializing the adaptive PSO algorithm; dividing subgroups, and constructing the LSTM model with the hyperparameters corresponding to each particle in the subgroup; training the LSTM model with the training set, validating the LSTM model with the validation set, and using the validation result as the fitness value of each particle; determining the global optimal particle position and the local optimal particle position according to the fitness value and the population division result; updating the common particle position and the local optimal particle position respectively according to the PSO algorithm; judging whether the termination condition is met: if so, returning the optimal hyperparameter value; otherwise, returning to the step of executing the subgroup division step and subsequent steps; constructing the final LSTM model with the optimal hyperparameter, training the LSTM model with the training set and the validation set, and predicting with the test set to obtain the prediction result.

[0009] Optionally, the verification result is: the average absolute percentage error of the fitness value of the PSO particle relative to the output value in the verification set; the hyperparameters include: the time window size of the LSTM model, the batch size, the number of hidden layer units of the neural network and the Dropout loss rate.

[0010] The key parameters include: effective contact rate.

[0011] To achieve the above objectives, according to another aspect of the present invention, a coronavirus prediction device based on the SEAIQR-PSO-LSTM hybrid model is provided.

[0012] The coronavirus prediction device based on the SEAIQR-PSO-LSTM hybrid model of the embodiment of the present invention includes: a model building unit, which is used to construct a SEAIQR-PSO-LSTM hybrid model including a SEAIQR model and a PSO-LSTM model; a model training unit, which is used to take a number sequence of multiple target populations associated with coronavirus in multiple recent historical statistical periods as input to train the SEAIQR model and the PSO-LSTM model; by performing linear regression modeling on the output results of the SEAIQR model and the PSO-LSTM model, the SEAIQR-PSO-LSTM hybrid model is constructed, and the SEAIQR-PSO-LSTM hybrid model is trained by optimizing the linear regression parameters using the least squares method; a model prediction unit, which is used to output the prediction result by the SEAIQR-PSO-LSTM hybrid model to obtain the number of target populations in at least one future statistical period; a model use unit, which is used to determine the influence of multiple key parameters related to coronavirus on the spread of coronavirus based on the trained SEAIQR-PSO-LSTM hybrid model and the prediction result, and determine the medical resource allocation strategy for the future statistical period based on the prediction result.

[0013] Optionally, the PSO-LSTM model includes an LSTM model and a PSO algorithm; and the model training unit is further used to: divide the pre-collected experimental data into a training set, a validation set and a test set; wherein the training set and the validation set constitute the data set; use the hyperparameters of the LSTM model as the optimization object to initialize the adaptive PSO algorithm; divide the subgroups, and construct the LSTM model with the hyperparameters corresponding to each particle in the subgroup; train the LSTM model using the training set, verify the LSTM model using the validation set, and use the verification result as the fitness value of each particle; determine the global optimal particle position and the local optimal particle position according to the fitness value and the population division result; update the common particle position and the local optimal particle position respectively according to the PSO algorithm; judge whether the termination condition is met: if so, return the optimal hyperparameter value; otherwise, return to the step of executing the subgroup division step and subsequent steps; construct the final LSTM model using the optimal hyperparameter, train the LSTM model using the training set and the validation set, and predict using the test set to obtain the prediction result.

[0014] To achieve the above objective, according to another aspect of the present invention, an electronic device is provided.

[0015] An electronic device of the present invention includes: one or more processors; a storage device for storing one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the coronavirus prediction method based on the SEAIQR-PSO-LSTM hybrid model provided by the present invention.

[0016] To achieve the above objective, according to yet another aspect of the present invention, a computer-readable storage medium is provided.

[0017] A computer-readable storage medium of the present invention stores a computer program thereon, and when the program is executed by a processor, the coronavirus prediction method based on the SEAIQR-PSO-LSTM hybrid model provided by the present invention is implemented.

[0018] According to the technical solution of the present invention, the embodiments of the above invention have the following advantages or beneficial effects:

[0019] First, the present invention constructs a SEAIQR-PSO-LSTM hybrid model including a SEAIQR model and a PSO-LSTM model, and uses a sequence of the number of target populations associated with the coronavirus in the most recent multiple historical statistical cycles as input to train the SEAIQR model and the PSO-LSTM model; then, by performing linear regression modeling on the output results of the SEAIQR model and the PSO-LSTM model, a SEAIQR-PSO-LSTM hybrid model is constructed, and the SEAIQR-PSO-LSTM hybrid model is trained using the least squares method to optimize the linear regression parameters; afterwards, the SEAIQR-PSO-LSTM hybrid model outputs the prediction result, and the number of target populations in at least one future statistical cycle is obtained; finally, the influence of multiple key parameters related to the coronavirus on the spread of the coronavirus is determined according to the trained SEAIQR-PSO-LSTM hybrid model and the prediction result, and the medical resource allocation strategy for the future statistical cycle is determined based on the prediction result. In this way, the advantages of the two models of SEAIQR and PSO-LSTM can be combined to realize the adaptive adjustment of the prediction model with statistical data and the improvement of prediction accuracy and reliability. In addition, the above prediction results can be used as a basis for pre-storing, configuring and deploying medical resources in the future. The present invention can also analyze the impact of key parameters such as effective contact rate on virus transmission based on the above SEAIQR-PSO-LSTM hybrid model.

[0020] The further effects of the above-mentioned non-conventional optional manner will be described below in conjunction with the specific implementation manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings are used to better understand the present invention and do not constitute an improper limitation of the present invention.

[0022] Figure 1 Schematic diagram of the main steps of the coronavirus prediction method based on the SEAIQR-PSO-LSTM hybrid model in an embodiment of the present invention;

[0023] Figure 2 is a schematic diagram of a virus transmission dynamics model according to an embodiment of the present invention;

[0024] Figure 3 It is an overall framework diagram of the LSTM model of an embodiment of the present invention;

[0025] Figure 4 Schematic diagram of PSO-LSTM hyperparameter optimization in an embodiment of the present invention;

[0026] Figure 5 This is an overall framework diagram of the SEAIQR-PSO-LSTM hybrid model of an embodiment of the present invention;

[0027] Figure 6 It is a schematic diagram of components of a coronavirus prediction device based on a SEAIQR-PSO-LSTM hybrid model in an embodiment of the present invention;

[0028] Figure 7 is an exemplary system architecture diagram in which embodiments of the present invention may be applied;

[0029] Figure 8 It is a schematic diagram of the structure of an electronic device used to implement the coronavirus prediction method based on the SEAIQR-PSO-LSTM hybrid model in an embodiment of the present invention. DETAILED DESCRIPTION

[0030] The following is a description of exemplary embodiments of the present invention in conjunction with the accompanying drawings, including various details of the embodiments of the present invention to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0031] It should be pointed out that the embodiments of the present invention and the technical features therein may be combined with each other without conflict.

[0032] Figure 1 Schematic diagram of main steps of the method according to an embodiment of the present invention.

[0033] like Figure 1 As shown, the coronavirus prediction method based on the SEAIQR-PSO-LSTM hybrid model of the embodiment of the present invention can be specifically performed according to the following steps:

[0034] Step S101: Taking the number sequences of multiple target populations associated with the coronavirus in multiple recent historical statistical periods as input, training the SEAIQR model and the PSO-LSTM model.

[0035] The purpose of this step is to train the SEAIQR model and the PSO-LSTM model. It can be understood that the above data set contains multiple training samples, each of which contains training input values ​​and label values. The above statistical period generally refers to the smallest statistical unit, such as one day. The most recent multiple historical statistical periods refer to the most recent continuous multiple statistical periods from the current moment, for example, the 100 days closest to the current moment. In this step, the headcount data of various target populations in the most recent multiple historical statistical periods can be used as input in a fixed order to train the SEAIQR model and the PSO-LSTM model.

[0036] During the training process, the time window size, batch size, number of hidden layer units of the neural network, Dropout loss rate and other hyperparameters of the LSTM model can be used as optimization variables for PSO particles in the PSO algorithm. By updating the speed and position of the particles, the fitness value of the load forecast is minimized, thereby obtaining the optimal model hyperparameters.

[0037] Preferably, in an embodiment of the present invention, the above optimization can perform the following steps: first, the pre-collected experimental data is divided into a training set, a validation set and a test set; wherein the training set and the validation set constitute the above data set. Then, the hyperparameters of the LSTM model are used as the optimization object to initialize the adaptive PSO algorithm. Thereafter, the subgroups are divided, and the LSTM model is constructed with the hyperparameters corresponding to each particle in the subgroup, the LSTM model is trained with the training set, and the LSTM model is verified with the validation set, and the verification result is used as the fitness value of each particle. Exemplarily, the above verification result can be the average absolute percentage error of the output value of the LSTM model relative to the label value in the validation set. Afterwards, the global optimal particle position and the local optimal particle position are determined according to the above fitness value and the population division result, and the ordinary particle position and the local optimal particle position are updated respectively according to the PSO algorithm, and it is judged whether the termination condition is reached: if so, the optimal hyperparameter value is returned; otherwise, the step of executing the subgroup division and the subsequent steps are returned. Finally, the final LSTM model is constructed using the optimal hyperparameters, the LSTM model is trained using the training set and the validation set, and the prediction is performed using the test set to obtain the prediction result. The above hyperparameter optimization steps can help further improve the prediction accuracy and reliability of the model.

[0038] Step S102: By performing linear regression modeling on the output results of the SEAIQR model and the PSO-LSTM model, a SEAIQR-PSO-LSTM hybrid model is constructed, and the SEAIQR-PSO-LSTM hybrid model is trained by optimizing the linear regression parameters using the least squares method.

[0039] In this step, the SEAIQR model and the PSO-LSTM model are used as data input terminals and are connected to the linear regression model together. That is, the SEAIQR model and the PSO-LSTM model input the historical statistical data of the coronavirus, and the output results of the SEAIQR model and the PSO-LSTM model enter the linear regression model to build the SEAIQR-PSO-LSTM hybrid model. In practical applications, the SEAIQR-PSO-LSTM hybrid model can be trained by using the least squares method to optimize the linear regression parameters.

[0040] Step S103: The SEAIQR-PSO-LSTM hybrid model outputs the prediction result to obtain the number of people in the target population within at least one future statistical period.

[0041] Step S104: Determine the impact of multiple key parameters related to coronavirus on the spread of coronavirus based on the trained SEAIQR-PSO-LSTM hybrid model and the prediction results, and determine the medical resource allocation strategy for the future statistical period based on the prediction results.

[0042] In specific applications, the above key parameters may include: effective contact rate. In this step, medical resources matching the predicted number of target populations may be stored in advance based on the prediction results of the model to ensure that there is no danger of an emergency. The impact of each key parameter on virus transmission may also be evaluated based on the above model and prediction results.

[0043] The following describes a technical solution of a specific embodiment of the present invention.

[0044] Infectious disease dynamics models can be used to study the epidemic patterns and dynamic development of diseases, and provide evidence-based support for quantifying the risk of virus transmission and seeking optimal treatment strategies. The Susceptible-Exposed-Infected-Removed (SEIR) model is a common infectious disease dynamics model, which is suitable for infectious diseases with a latent period and lifelong immunity after recovery. Long Short Term Memory Network (LSTM) is very good at processing and predicting time series data. It can "recall" patterns in past or future data without artificially adding time features, and can maximize the mining of nonlinear relationships between time series data.

[0045] Public health crises caused by major infectious diseases have always been an important practical problem faced by the development of human society. Especially in modern society with highly close global division of labor and cooperation, the impact on economic and social development and people's well-being is more prominent.

[0046] The research contents of this program are as follows: First, data collection. Second, model construction. By integrating multi-source big data, the SEAIQR-PSO-LSTM hybrid model is constructed with the advantages of both the SEAIQR model of infectious disease dynamics and the deep network PSO-LSTM model. Third, treatment effect evaluation. Sensitivity analysis is performed on each key parameter. Fourth, medical resource allocation simulation.

[0047] The key issues that this solution intends to solve are: First, a new framework for the SEAIQR-PSO-LSTM hybrid model. The SEAIQR model is a correction and optimization of the classic SEIR model. The SEAIQR model constructed in this solution is more in line with the development characteristics of the coronavirus, but the SEAIQR model cannot automatically adjust according to historical fluctuation data. The PSO-LSTM model is one of the models based on time series prediction. It can explore the nonlinear correlation between time series data to a large extent, and through historical data, it can learn some factors of human interference. This solution uses the prediction results of the above two models to learn the linear relationship between the prediction results of the two models and the real data through linear regression, thereby improving the overall accuracy and reliability of the SEAIQR-PSO-LSTM hybrid model. Second, a new perspective on the allocation of medical resources can dynamically adjust medical resources.

[0048] The prediction model used in this solution is described below. Specifically, the SEAIQR model is first constructed.

[0049] In the framework of the classic SEIR model of susceptible (S)-exposed (E)-infected (I)-removed (R), a time-varying SEAIQR model with phased measures is constructed, taking into account the characteristics of the virus and the types of cases {mild (I1), common (I2), severe and critical (I3)}. Figure 2 SEAIQR stands for Susceptible-Exposed-Asymptomatic-Infected-Quarantined-Removed. The differential equations are as follows:

[0050] dS / dt=-c(t)[β+q(1-β)]S(I+θE+kA) / N+λSq-vS+hR

[0051] dSq / dt=c(t)q(1-β)S(I+θE+kA) / N-λSq

[0052] dE / dt=c(t)β(1-q)S(I+θE+kA) / N-pσ A E-(1-p)σ I E

[0053] dEq / dt=c(t)βqS(I+θE+kA) / N-δ q Eq

[0054] dA / dt=pσ A E-(δ I +α+γ A )A

[0055]

[0056]

[0057]

[0058] dH / dt=δ I (I1+I2+I3+A)+δ q Eq-(α+γ H )H

[0059] dR / dt=γ I (I1+I2+I3)+γ A A+γ H H+vS-hR+α(I1+I2+I3+A+H)

[0060] In this model, c(t) represents the number of contacts of a population that changes over time. Assume that before t = t1, the number of contacts is a constant, that is, c(t) = c0; after t = t1 and before t = t2, the number of contacts decreases exponentially at a rate w; after t = t2, the number of contacts decreases exponentially at a faster rate r, then the number of contacts of a population can be expressed by the following piecewise function:

[0061]

[0062] Next, we build a PSO-LSTM model. Considering the temporal characteristics of virus data and the nonlinear relationship between them, we build a deep learning LSTM model. Figure 3 , the calculation formula is as follows:

[0063] f t =σ(W f [h t-1 ,X t ]+b f)

[0064] i t =σ(W i [h t-1 ,X t ]+b i )

[0065]

[0066]

[0067] O t =σ(W o [h t-1 ,X t ]+b o )

[0068] h t =O t ⊙tanh(C t )

[0069] Where: f t 、i t , O t Represent the outputs of the forget gate, input gate, and output gate respectively. is the memory of the current input, C t-1 is the cell state at the previous moment, C t is the cell state at the current moment, h t is the output at the current moment, W is the weight matrix connecting the two layers, b is the bias term, and σ is the sigmoid activation function.

[0070] In order to effectively avoid the overfitting problem of LSTM neural network, Dropout technology is used in the non-cyclic part of the hidden layer to randomly inactivate neurons. This solution uses a long short-term memory network prediction model (PSO-LSTM) based on particle swarm optimization (PSO), which is good at dealing with complex nonlinear problems with long-term dependencies. The PSO algorithm with adaptive learning strategy is used to optimize the hyperparameters of the LSTM model, including the initial learning rate, the number of hidden layer units, Dropout and batch_size, so that the data features match the network topology and improve the prediction accuracy. See the PSO-LSTM model algorithm flow for details. Figure 4 .

[0071] Afterwards, the SEAIQR-PSO-LSTM hybrid model is constructed.

[0072] In order to more accurately predict the development trend of the virus, this solution forms a SEAIQR-PSO-LSTM hybrid model by performing linear regression modeling on the prediction results of the SEAIQR model and the PSO-LSTM model. The advantage of the SEAIQR-PSO-LSTM hybrid model is that it not only learns the changing patterns of general infectious diseases, but also comprehensively predicts the next virus situation based on the characteristics of historical data. This solution believes that there is a linear correlation between the SEAIQR model and the PSO-LSTM model, that is, the internal relationship between the SEAIQR model prediction value and the PSO-LSTM model prediction value is found through a linear equation to form the best linear fitting line. The overall framework diagram of the SEAIQR-PSO-LSTM hybrid model is as follows: Figure 5 .

[0073] The SEAIQR-PSO-LSTM hybrid model uses the predicted value g of the SEAIQR model and the predicted value h of the PSO-LSTM model as input values ​​and the real data as the target value to perform linear regression modeling:

[0074]

[0075] In the formula, is the predicted value of the SEAIQR-PSO-LSTM hybrid model. The coefficient matrix k and the offset b can be solved by x = [g, h] and the corresponding true value y, so that the predicted value The SEAIQR-PSO-LSTM hybrid model training process uses the least squares method to optimize the linear regression parameters:

[0076]

[0077] In summary, the present invention can combine the advantages of SEAIQR and PSO-LSTM models to achieve adaptive adjustment of the prediction model with statistical data and improve the prediction accuracy and reliability. In addition, the above prediction results can be used as a basis for pre-reserve, configuration and deployment of medical resources in the future. The present invention can also analyze the impact of key parameters such as effective contact rate on virus transmission based on the above SEAIQR-PSO-LSTM hybrid model.

[0078] It should be noted that, for the convenience of description, the aforementioned method embodiments are described as a series of action combinations, but those skilled in the art should be aware that the present invention is not limited by the described action sequence, and some steps can actually be performed in other sequences or simultaneously. In addition, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily necessary to implement the present invention.

[0079] In order to better implement the above-mentioned solution of the embodiment of the present invention, relevant devices for implementing the above-mentioned solution are also provided below.

[0080] See also Figure 6 As shown, the coronavirus prediction device 600 based on the SEAIQR-PSO-LSTM hybrid model provided by an embodiment of the present invention may include: a model building unit 601, a model training unit 602, a model prediction unit 603 and a model use unit 604.

[0081] Among them, the model building unit 601 is used to construct a SEAIQR-PSO-LSTM hybrid model including a SEAIQR model and a PSO-LSTM model; the model training unit 602 is used to take the number sequences of multiple target populations associated with the coronavirus in the most recent multiple historical statistical periods as input to train the SEAIQR model and the PSO-LSTM model; the SEAIQR-PSO-LSTM hybrid model is constructed by performing linear regression modeling on the output results of the SEAIQR model and the PSO-LSTM model, and the SEAIQR-PSO-LSTM hybrid model is trained by optimizing the linear regression parameters using the least squares method; the model prediction unit 603 is used to output the prediction result by the SEAIQR-PSO-LSTM hybrid model to obtain the number of target populations in at least one future statistical period; the model use unit 604 is used to determine the influence of multiple key parameters related to the coronavirus on the spread of the coronavirus based on the trained SEAIQR-PSO-LSTM hybrid model and the prediction result, and determine the medical resource allocation strategy for the future statistical period based on the prediction result.

[0082] In an embodiment of the present invention, the PSO-LSTM model includes an LSTM model and a PSO algorithm; and the model training unit 602 is further used to: divide the pre-collected experimental data into a training set, a validation set and a test set; wherein the training set and the validation set constitute the data set; use the hyperparameters of the LSTM model as the optimization object to initialize the adaptive PSO algorithm; divide the subgroups, and construct the LSTM model with the hyperparameters corresponding to each particle in the subgroup; train the LSTM model using the training set, verify the LSTM model using the validation set, and use the verification result as the fitness value of each particle; determine the global optimal particle position and the local optimal particle position according to the fitness value and the population division result; update the common particle position and the local optimal particle position respectively according to the PSO algorithm; judge whether the termination condition is met: if so, return the optimal hyperparameter value; otherwise, return to the step of executing the subgroup division and subsequent steps; construct the final LSTM model using the optimal hyperparameter, train the LSTM model using the training set and the validation set, and predict using the test set to obtain the prediction result.

[0083] As a preferred solution, the model training unit 602 is further used to: before completing the training of the SEAIQR-PSO-LSTM hybrid model, optimize multiple hyperparameters in the LSTM model as PSO particles in the PSO algorithm.

[0084] Preferably, the verification result is: the average absolute percentage error of the fitness value of the PSO particle relative to the output value in the verification set; the hyperparameters include: the time window size of the LSTM model, the batch size, the number of hidden layer units of the neural network and the Dropout data.

[0085] In addition, in the embodiment of the present invention, the key parameters include: effective contact rate.

[0086] According to the technical solution of the embodiment of the present invention, the advantages of the SEAIQR and PSO-LSTM models can be combined to realize the adaptive adjustment of the prediction model with statistical data and the improvement of prediction accuracy and reliability. In addition, the above prediction results can be used as a basis for the advance reserve, configuration and deployment of medical resources in the future period. The present invention can also analyze the impact of key parameters such as effective contact rate on virus transmission based on the above SEAIQR-PSO-LSTM hybrid model, so as to implement targeted measures.

[0087] Figure 7An exemplary system architecture 700 is shown to which a coronavirus prediction method based on a SEAIQR-PSO-LSTM hybrid model or a coronavirus prediction device based on a SEAIQR-PSO-LSTM hybrid model according to an embodiment of the present invention can be applied.

[0088] like Figure 7 As shown, the system architecture 700 may include terminal devices 701, 702, 703, a network 704 and a server 705 (this architecture is only an example, and the components included in the specific architecture may be adjusted according to the specific application). The network 704 is used to provide a medium for a communication link between the terminal devices 701, 702, 703 and the server 705. The network 704 may include various connection types, such as wired, wireless communication links or optical fiber cables.

[0089] The user can use the terminal devices 701, 702, 703 to interact with the server 705 through the network 704 to receive or send messages, etc. Various client applications can be installed on the terminal devices 701, 702, 703, such as machine learning model applications (only as an example).

[0090] The terminal devices 701 , 702 , and 703 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, and desktop computers.

[0091] The server 705 may be a server that provides various services, such as a backend server (only an example) that provides support for machine learning model applications operated by users using the terminal devices 701, 702, and 703. The backend server may process the received prediction request and feed back the processing result (such as the prediction result - only an example) to the terminal devices 701, 702, and 703.

[0092] It should be noted that the coronavirus prediction method based on the SEAIQR-PSO-LSTM hybrid model provided in the embodiment of the present invention is generally executed by the server 705, and accordingly, the coronavirus prediction device based on the SEAIQR-PSO-LSTM hybrid model is generally arranged in the server 705.

[0093] It should be understood that Figure 7 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0094] The present invention also provides an electronic device. The electronic device of an embodiment of the present invention includes: one or more processors; a storage device for storing one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the coronavirus prediction method based on the SEAIQR-PSO-LSTM hybrid model provided by the present invention.

[0095] Reference below Figure 8 , which shows a schematic diagram of the structure of a computer system 800 of an electronic device suitable for implementing an embodiment of the present invention. Figure 8 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0096] like Figure 8 As shown, the computer system 800 includes a central processing unit (CPU) 1201, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 802 or a program loaded from a storage part 808 into a random access memory (RAM) 803. In the RAM 803, various programs and data required for the operation of the computer system 800 are also stored. The CPU 801, the ROM 802, and the RAM 803 are connected to each other via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0097] The following components are connected to the I / O interface 805: an input section 806 including a keyboard, a mouse, etc.; an output section 807 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 808 including a hard disk, etc.; and a communication section 809 including a network interface card such as a LAN card, a modem, etc. The communication section 809 performs communication processing via a network such as the Internet. A drive 810 is also connected to the I / O interface 805 as needed. A removable medium 811, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 810 as needed so that a computer program read therefrom is installed into the storage section 808 as needed.

[0098] In particular, according to the embodiments disclosed in the present invention, the process described in the main step diagram above can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program code for executing the method shown in the main step diagram. In the above embodiment, the computer program can be downloaded and installed from the network through the communication part 809, and / or installed from the removable medium 811. When the computer program is executed by the central processing unit 801, the above functions defined in the system of the present invention are executed.

[0099] It should be noted that the computer-readable medium shown in the present invention may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present invention, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer readable signal media may also be any computer readable medium other than computer readable storage media, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0100] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present invention. In this regard, each box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the above-mentioned module, program segment or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0101] The units involved in the embodiments of the present invention may be implemented by software or hardware. The units described may also be arranged in a processor, for example, may be described as: a processor including a model building unit, a model training unit, a model prediction unit, and a model use unit. The names of these units do not constitute a limitation on the units themselves in some cases, for example, the model building unit may also be described as a "unit that provides a SEAIQR-PSO-LSTM hybrid model to a model training unit".

[0102] As another aspect, the present invention further provides a computer-readable medium, which may be included in the device described in the above embodiment; or may exist independently without being assembled into the device. The computer-readable medium carries one or more programs. When the one or more programs are executed by the device, the steps executed by the device include: constructing a SEAIQR-PSO-LSTM hybrid model including a SEAIQR model and a PSO-LSTM model; wherein, the number sequences of multiple target populations associated with the coronavirus in the most recent multiple historical statistical periods are used as input to train the SEAIQR model and the PSO-LSTM model; the SEAIQR-PSO-LSTM hybrid model is constructed by performing linear regression modeling on the output results of the SEAIQR model and the PSO-LSTM model, and the SEAIQR-PSO-LSTM hybrid model is trained by optimizing the linear regression parameters using the least squares method; the SEAIQR-PSO-LSTM hybrid model outputs a prediction result to obtain the number of target populations in at least one future statistical period; the influence of multiple key parameters related to the coronavirus on the spread of the coronavirus is determined based on the trained SEAIQR-PSO-LSTM hybrid model and the prediction result, and the medical resource allocation strategy for the future statistical period is determined based on the prediction result.

[0103] In the technical solution of the embodiment of the present invention, the advantages of the SEAIQR and PSO-LSTM models can be combined to achieve adaptive adjustment of the prediction model with statistical data and improve the accuracy and reliability of prediction. In addition, the above prediction results can be used as a basis for pre-reservation, configuration and deployment of medical resources in the future. The present invention can also analyze the impact of key parameters such as effective contact rate on virus transmission based on the above SEAIQR-PSO-LSTM hybrid model.

[0104] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may occur depending on design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A coronavirus prediction method based on the SEAIQR-PSO-LSTM hybrid model, characterized in that: include: Constructing a SEAIQR-PSO-LSTM hybrid model including a SEAIQR model and a PSO-LSTM model; wherein the number sequences of multiple target populations associated with the coronavirus in multiple recent historical statistical periods are used as input to train the SEAIQR model and the PSO-LSTM model; The SEAIQR-PSO-LSTM hybrid model is constructed by performing linear regression modeling on the output results of the SEAIQR model and the PSO-LSTM model, and the SEAIQR-PSO-LSTM hybrid model is trained by optimizing the linear regression parameters using the least squares method; The SEAIQR-PSO-LSTM hybrid model outputs a prediction result to obtain the number of target people in at least one future statistical period; Determine the impact of multiple key parameters related to coronavirus on the spread of coronavirus based on the trained SEAIQR-PSO-LSTM hybrid model and the prediction results, and determine the medical resource allocation strategy for the future statistical period based on the prediction results; Constructing the SEAIQR-PSO-LSTM hybrid model, including: using historical statistical data of coronavirus to create a data set to train the SEAIQR model and the PSO-LSTM model, and constructing the SEAIQR-PSO-LSTM hybrid model by performing linear regression modeling on the output results of the two; The PSO-LSTM model includes an LSTM model and a PSO algorithm; and the method further includes: before completing the training of the SEAIQR-PSO-LSTM hybrid model, optimizing multiple hyperparameters in the LSTM model as PSO particles in the PSO algorithm, including the following steps performed in sequence: Dividing the pre-collected experimental data into a training set, a validation set and a test set; wherein the training set and the validation set constitute the data set; Taking the hyperparameters of the LSTM model as the optimization object, initializing the adaptive PSO algorithm; Divide the particles into subgroups, and construct the LSTM model using the hyperparameters corresponding to the particles in the subgroups; Using the training set to train the LSTM model, using the verification set to verify the LSTM model, and using the verification result as the fitness value of each particle; Determine the global optimal particle position and the local optimal particle position according to the fitness value and the population division result; According to the PSO algorithm, the common particle positions and the local optimal particle positions are updated respectively; Determine whether the termination condition is met: if so, return the optimal hyperparameter value; otherwise, return to the step of executing subgroup division and subsequent steps; The final LSTM model is constructed using the optimal hyperparameters, the LSTM model is trained using the training set and the validation set, and prediction is performed using the test set to obtain the prediction result.

2. The method according to claim 1, characterized in that The verification result is: the average absolute percentage error of the fitness value of the PSO particle relative to the output value in the verification set; The hyperparameters include: the time window size, batch size, number of hidden layer units of the neural network and Dropout loss rate of the LSTM model.

3. The method according to claim 1 or 2, characterized in that: The key parameters include: effective contact rate.

4. A coronavirus prediction device based on the SEAIQR-PSO-LSTM hybrid model, characterized in that: include: A model building unit, used for building a SEAIQR-PSO-LSTM hybrid model including a SEAIQR model and a PSO-LSTM model; A model training unit is used to take a number sequence of multiple target populations associated with the coronavirus in multiple recent historical statistical periods as input to train the SEAIQR model and the PSO-LSTM model; construct the SEAIQR-PSO-LSTM hybrid model by performing linear regression modeling on the output results of the SEAIQR model and the PSO-LSTM model, and use the least squares method to optimize the linear regression parameters to train the SEAIQR-PSO-LSTM hybrid model; A model prediction unit, used to output a prediction result from the SEAIQR-PSO-LSTM hybrid model to obtain the number of people in the target population within at least one future statistical period; A model using unit, used to determine the influence of multiple key parameters related to coronavirus on the spread of coronavirus based on the trained SEAIQR-PSO-LSTM hybrid model and the prediction results, and determine the medical resource allocation strategy for the future statistical period based on the prediction results; The model training unit is also used to: create a data set using historical statistical data of coronavirus to train the SEAIQR model and the PSO-LSTM model, and construct the SEAIQR-PSO-LSTM hybrid model by performing linear regression modeling on the output results of the two; the PSO-LSTM model includes an LSTM model and a PSO algorithm; before the SEAIQR-PSO-LSTM hybrid model is trained, multiple hyperparameters in the LSTM model are optimized as PSO particles in the PSO algorithm; the model training unit is further used to: Dividing the pre-collected experimental data into a training set, a validation set and a test set; wherein the training set and the validation set constitute the data set; Taking the hyperparameters of the LSTM model as the optimization object, initializing the adaptive PSO algorithm; Divide the particles into subgroups, and construct the LSTM model using the hyperparameters corresponding to the particles in the subgroups; Using the training set to train the LSTM model, using the verification set to verify the LSTM model, and using the verification result as the fitness value of each particle; Determine the global optimal particle position and the local optimal particle position according to the fitness value and the population division result; According to the PSO algorithm, the common particle positions and the local optimal particle positions are updated respectively; Determine whether the termination condition is met: if so, return the optimal hyperparameter value; otherwise, return to the step of executing subgroup division and subsequent steps; The final LSTM model is constructed using the optimal hyperparameters, the LSTM model is trained using the training set and the validation set, and prediction is performed using the test set to obtain the prediction result.

5. An electronic device, characterized in that: include: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 3.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 3 is implemented.

Images