Virus propagation prediction method for large-scale community

Abstract

The invention relates to the technical field of virus propagation prediction, in particular to a virus propagation prediction method for a large-scale community, which comprises the following steps: acquiring a plurality of virus propagation prediction characteristic data; comparing each piece of virus propagation prediction characteristic data with a corresponding reasonable interval in a virus propagation database to identify a plurality of pieces of mutation data; counting the type of the mutation data in the same monitoring period in the plurality of mutation data; based on the number of the types of the mutation data, judging whether the corresponding mutation data is rejected or not; for the non-rejected mutation data, reducing the classification number of the first number of corresponding mutation data meeting the diffusion association condition; and performing mapping verification on a second quantity of corresponding mutation data which does not meet the diffusion association condition, and then replacing the mutation data with corresponding data in the virus propagation database. According to the invention, the prediction accuracy of virus transmission for large communities is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of virus propagation prediction, and in particular to a virus propagation prediction method for a large community. Background Art

[0002] Traditional models rely on manually collected static contact data (such as questionnaires), which struggle to capture sudden human movements and complex environmental interactions (such as central air conditioning systems and public utility usage). This results in input data noise exceeding 35% (WHO 2024 report). Existing anomalous data processing often relies on simple elimination strategies, resulting in loss of valuable information. Macro models (such as cellular automata) ignore microscopic variations within communities, resulting in a 12-24 hour delay in responding to outbreaks at the building unit level. Micro models (such as traditional SEIR) are unable to adapt to dynamic environments (such as wind speed fluctuations and temporary quarantine measures) due to their fixed parameters, resulting in prediction errors fluctuating by as much as ±40%. Cross-regional transmission modeling: Bluetooth / Wi-Fi positioning data only provides coarse-grained location information, lacking quantitative analysis of contact duration and distance attenuation effects, resulting in a cross-regional transmission link misjudgment rate exceeding 25%. System adaptability: Existing solutions lack dynamic calibration mechanisms. When emergencies occur, manual model retraining is required, which takes over 72 hours.

[0003] Chinese Patent Publication No.: CN114999667A discloses a method and system for predicting the cross-regional spread of infectious diseases taking into account traffic intervention; obtains geographic units and main parameters; constructs an infectious disease simulation background based on the geographic unit level and main parameters, and determines the corresponding remote transfer probability matrix; constructs an infectious disease dynamics model based on the remote transfer probability matrix; determines the number of people in different warehouses in each region under a specified time and space based on the infectious disease dynamics model, simulates and visualizes the dynamics of infectious diseases in each region; determines and visualizes the real-time transmission power of the virus and its confidence interval based on the number of people in different warehouses in each region; predicts and visualizes the future infectious disease trends in each region based on the population change data of the warehouses in each region and the real-time transmission power of the virus. It can be seen that the method and system for predicting the cross-regional spread of infectious diseases taking into account traffic intervention has a correlation relationship between certain mutation data. When one mutation data changes, other mutation data will change with the change of this mutation data. When there is a problem in identifying this relationship, the overall effect between the interacting data may cause the problem of invalid virus transmission prediction. Summary of the Invention

[0004] To this end, the present invention provides a virus transmission prediction method for large communities, which is used to overcome the problem in the prior art that there is a correlation between certain mutation data. When one mutation data changes, other mutation data will change with the change of this mutation data. When there is a problem in identifying this relationship, the overall effect between the interacting data may cause the virus transmission prediction to fail.

[0005] To achieve the above objectives, the present invention provides a method for predicting virus spread in a large community, comprising: Obtaining certain virus transmission prediction feature data; Compare each virus transmission prediction feature data with the corresponding reasonable interval in the virus transmission database to identify several mutation data; Counting the types of mutation data in the same monitoring period among the plurality of mutation data; Determining whether to remove corresponding mutation data based on the number of types of the mutation data; For the mutation data that has not been eliminated, reducing the number of categories of the first number of mutation data that meet the diffusion association condition; Performing mapping verification on the second number of corresponding mutation data that do not meet the diffusion association condition and then replacing them with corresponding data in the virus propagation database; Eliminating the corresponding mutation data to output a basic data set for model training; Based on the basic data set for model training, the large-scale basic model and the small-scale basic model are trained in sequence to form a large-scale virus propagation prediction model and a small-scale virus propagation prediction model respectively; Inputting actual virus propagation feature data into the large-scale virus propagation prediction model and the small-scale virus propagation prediction model respectively to output a first diffusion prediction parameter and a second diffusion prediction parameter respectively; determining a final diffusion parameter for a large community based on the first diffusion prediction parameter, the second diffusion prediction parameter, and a regional connectivity characteristic parameter; Among them, the diffusion association condition is that the linear fit between the mutation data corresponding to the first quantity and the estimated virus diffusion parameters in the virus propagation database is greater than a preset fit.

[0006] Furthermore, judging whether to remove corresponding mutation data based on the number of types of the mutation data includes: If the number of types of mutation data within the same monitoring period is zero, the corresponding mutation data is eliminated; If the number of types of mutation data within the same monitoring period is not zero, the corresponding mutation data is not eliminated.

[0007] Furthermore, the first number of classification numbers of the corresponding mutation data is negatively correlated with the linear fit.

[0008] Furthermore, the mapping verification includes: Performing consistency mapping training on the second number of corresponding mutation data that do not meet the diffusion association condition and the estimated virus diffusion parameters in the virus transmission database; The replacement is performed based on inconsistent results of the consistent mapping training.

[0009] Furthermore, training the large scene basic model and the small scene basic model separately includes: Dividing the basic data set for model training into a training set, a test set, and a validation set; Using the training set to train the large-scale basic model, and sequentially using the test set and the validation set to test and update the large-scale basic model to output a large-scale virus propagation prediction model; The small-scene basic model is trained using the training set, and the small-scene basic model is tested and updated using the test set and the validation set in turn to output a small-scene virus propagation prediction model.

[0010] Furthermore, the regional connectivity characteristic parameters of the single region and each adjacent region are determined based on the first corresponding virus diffusion parameter of the mutation data in the single region within the same monitoring period that meets the diffusion association condition in the virus propagation database and the second corresponding virus diffusion parameter of the mutation data in each adjacent region in the virus propagation database.

[0011] Furthermore, the regional connectivity characteristic parameter between the single region and each adjacent region is the difference between the second corresponding virus diffusion parameter and the first corresponding virus diffusion parameter.

[0012] Further, based on the absolute value of the difference between the first diffusion prediction parameter and the second diffusion prediction parameter being greater than a preset absolute value, the second diffusion prediction parameter is determined as the final diffusion parameter; based on the absolute value of the difference between the first diffusion prediction parameter and the second diffusion prediction parameter being less than or equal to the preset absolute value, the first diffusion prediction parameter is determined as the final diffusion parameter.

[0013] Furthermore, the calculation formula of the second diffusion prediction parameter is:

[0014] Among them, R is the second diffusion prediction parameter, r i is the basic diffusion parameter of the virus in the i-th small area, q ijis the regional connectivity characteristic parameter of the jth adjacent region of the ith small region, n is the number of adjacent regions of the ith small region, and m is the total number of small regions.

[0015] Furthermore, the sum of the first number and the second number is equal to the total number of mutation data that have not been eliminated.

[0016] Compared with the prior art, the beneficial effect of the present invention is that the method of the present invention determines whether to eliminate the corresponding mutation data based on the number of types of mutation data, thereby overcoming the problem that due to the association between certain mutation data, when one mutation data changes, other mutation data will change with the change of this mutation data. When there is a problem in identifying this relationship, the overall effect between the interacting data may cause the virus transmission prediction to fail; by setting the number of categories of the corresponding mutation data that meets the diffusion association conditions to be reduced when the corresponding mutation data is not eliminated, the problem of increasing and becoming more complex the number of interaction layers and action relationships between the mutation data when there are too many types of associated mutation data is overcome, thereby causing virus transmission. The prediction accuracy is reduced; by mapping and verifying the corresponding mutation data that does not meet the diffusion association conditions and replacing them with the corresponding data in the virus transmission database, the impact of the reduction in the accuracy of virus transmission prediction caused by the incorrect elimination of mutation data due to the inability to identify mutation data that is related to the virus transmission prediction model but has a weak correlation with the diffusion prediction results is reduced; by determining the final diffusion parameter for a large community based on the first diffusion prediction parameter, the second diffusion prediction parameter and the regional connectivity characteristic parameter, the impact of inaccurate monitoring of the connectivity relationship between small areas during the virus transmission process in a large community and the connectivity rules between small areas on the accuracy of virus prediction is reduced; the accuracy of virus transmission prediction in large communities is improved.

[0017] Furthermore, the present invention eliminates mutation data when the number of mutation data types within the same monitoring period is zero, and does not eliminate mutation data when the number is not zero. In the process of modeling the virus transmission prediction model for the virus transmission prediction process in a large community and the process of data collection and processing, when there is a strong time consistency between the monitored mutation data, it indicates that there is a certain invisible correlation between the mutation data within the same monitoring period. For example, in a large community, the spread of the virus may be weakened due to the presence of wind, but a small area of ​​enhanced wind may be formed in a small area between residential buildings, thereby exacerbating the virus transmission and diffusion ability in the small area. Therefore, through the detection and analysis of implicit correlations, abnormal factors that promote the enhancement of virus transmission ability can be identified, and the data cleaning and elimination process can be made more accurate, thereby improving the accuracy of virus transmission prediction.

[0018] Furthermore, the present invention reduces the number of classifications of the corresponding mutation data that meets the diffusion association condition, and the number of classifications of the mutation data is negatively correlated with the linear fit, that is, the greater the linear fit, the smaller the number of classifications of the mutation data. When the linear fit is large, it means that the correlation between the mutation data and the final diffusion parameter is large. The commonality of the mutation data is found, the classification of the mutation data is re-divided, and the number of classifications of the mutation data is reduced. This can reduce the complexity of the basic training set and improve the efficiency of training the model based on the basic data set used for model training.

[0019] Furthermore, through mapping verification, the data with poor correlation with the spread of the virus is replaced, which improves the accuracy of data selection in the basic data set for model training, avoids the interference of corresponding mutation data that does not meet the diffusion association conditions during model training, and thus improves the accuracy of virus spread prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is an overall flow chart of a method for predicting virus spread in a large community according to an embodiment of the present invention; Figure 2 This is a flowchart of a method for predicting virus spread in a large community according to an embodiment of the present invention, which determines whether to exclude corresponding mutation data based on the number of types of the mutation data; Figure 3 This is a flowchart of a mapping verification method for predicting virus spread in a large community according to an embodiment of the present invention; Figure 4 This is a flowchart of separately training a large-scene basic model and a small-scene basic model in a virus propagation prediction method for a large community according to an embodiment of the present invention. DETAILED DESCRIPTION

[0021] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are merely used to explain the present invention and are not intended to limit the present invention.

[0022] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0023] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside", and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.

[0024] Furthermore, it should be noted that, in the description of the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0025] See also Figure 1 、 Figure 2 、 Figure 3 as well as Figure 4 As shown, they are respectively an overall flow chart of the virus propagation prediction method for a large community according to an embodiment of the present invention, a flow chart for determining whether to remove corresponding mutation data based on the number of types of the mutation data, a flow chart for mapping verification, and a flow chart for training the large-scale basic model and the small-scale basic model respectively; The method for predicting virus spread in a large community according to an embodiment of the present invention includes: Step S1, obtaining some virus transmission prediction feature data; Step S2: Compare each virus transmission prediction feature data with the corresponding reasonable interval in the virus transmission database to identify a number of mutation data; Step S3, counting the types of mutation data in the same monitoring period among the plurality of mutation data; Step S4, judging whether to remove corresponding mutation data based on the number of types of the mutation data; Step S5: for the mutation data that has not been eliminated, reducing the number of categories of the first number of mutation data that meet the diffusion association condition; Step S6, performing mapping verification on the second number of corresponding mutation data that do not meet the diffusion association condition and then replacing them with corresponding data in the virus propagation database; Specifically, the corresponding data is the average of historical data of the same type, same region, and same monitoring period in the virus propagation database.

[0026] Step S7, removing the corresponding mutation data to output a basic data set for model training; Step S8, training the large-scene basic model and the small-scene basic model in sequence based on the basic data set for model training to form a large-scene virus propagation prediction model and a small-scene virus propagation prediction model respectively; Step S9: inputting the actual virus propagation feature data into the large-scale virus propagation prediction model and the small-scale virus propagation prediction model respectively to output a first diffusion prediction parameter and a second diffusion prediction parameter respectively; Step S10, determining a final diffusion parameter for a large community based on the first diffusion prediction parameter, the second diffusion prediction parameter, and a regional connectivity characteristic parameter; Among them, the diffusion association condition is that the linear fit between the mutation data corresponding to the first quantity and the estimated virus diffusion parameters in the virus propagation database is greater than a preset fit.

[0027] In this embodiment, the present invention determines whether to exclude corresponding mutation data based on the number of mutation data types, thereby overcoming the problem that due to the association relationship between certain mutation data, when one mutation data changes, other mutation data will change with the change of this mutation data. When there is a problem in identifying this relationship, the overall effect of the interacting data may cause the virus transmission prediction to fail. By reducing the number of categories of corresponding mutation data that meet the diffusion association conditions when the corresponding mutation data is not excluded, the present invention overcomes the problem that when there are too many types of associated mutation data, the number of interaction layers and action relationships between the mutation data will become more and more complex, thereby causing the accuracy of virus transmission prediction to decrease. By mapping and verifying the corresponding mutation data that does not meet the diffusion association conditions and then replacing them with the corresponding data in the virus transmission database, the impact of the decline in the accuracy of virus transmission prediction caused by the incorrect elimination of mutation data due to the inability to identify mutation data that is related to the virus transmission prediction model but has a weak correlation with the diffusion prediction results is reduced; by determining the final diffusion parameter for a large community based on the first diffusion prediction parameter, the second diffusion prediction parameter and the regional connectivity feature parameter, the impact of inaccurate monitoring of the connectivity relationship between small areas during the virus transmission process in a large community and the connectivity rules between small areas on the accuracy of virus prediction is reduced; and the accuracy of virus transmission prediction in a large community is improved.

[0028] Specifically, the characteristic data for virus transmission prediction include the distribution density of permanent population in large communities, the distribution density of floating population in large communities, the ambient humidity of several sampling points, the ventilation frequency of each public space, the coverage rate of personal protective equipment, the amount of virus residue on the surface of objects, etc.

[0029] Specifically, the permanent population distribution density of a large community is calculated by applying to the statistics bureau where the large community is located to obtain the corresponding number of permanent residents in the large community, and then dividing the permanent population by the land area of ​​the large community; the floating population distribution density of a large community is calculated by applying to the statistics bureau where the large community is located to obtain the corresponding number of floating population in the large community, and then dividing the floating population by the land area of ​​the large community; the ambient humidity at several sampling points is obtained by setting humidity sensors at several sampling points; the ventilation frequency of each public space is calculated by counting the number of times doors and windows of each public space are opened in a day, and the coverage rate of personal protective equipment is calculated by dividing the total amount of personal protective equipment by the population in the large community; the residual amount of virus on the surface of an object is obtained by taking swab samples on the surface of the object and then performing nucleic acid testing on the sampled swabs to obtain the residual amount of virus.

[0030] Specifically, the virus propagation database is a type of database that stores a number of virus propagation prediction feature data. Those skilled in the art will appreciate that the database may be a solid-state drive (SSD), a mechanical hard drive, or a tape storage.

[0031] During implementation, the permanent population distribution density of large communities is 5,000 people / square kilometer. The reasonable range corresponding to the permanent population distribution density of large communities stored in the virus transmission database is [2,500 people / square kilometer, 8,300 people / square kilometer]. 5,000 people / square kilometer is within the reasonable range, and the permanent population distribution density of large communities is judged to be normal data. If it is not within the reasonable range, the permanent population distribution density of large communities is judged to be mutation data.

[0032] During implementation, the coverage rate of personal protective equipment is 85%. The reasonable range corresponding to the coverage rate of personal protective equipment stored in the virus transmission database is [80%, 99%]. 85% is within the reasonable range, and the coverage rate of personal protective equipment is judged to be normal data. If it is not within the reasonable range, the coverage rate of personal protective equipment is judged to be mutation data.

[0033] Specifically, the types of mutation data include object characteristic data, population distribution characteristic data, and environmental characteristic data; object characteristic data include the coverage rate of personal protective equipment and the amount of virus residue on the surface of objects; population distribution characteristic data include the permanent population distribution density of large communities and the floating population distribution density of large communities; environmental characteristic data include the ambient humidity of several sampling points and the ventilation efficiency index of each public space.

[0034] Specifically, the linear fit between the first number of corresponding mutation data and the estimated virus diffusion parameters in the virus propagation database is the average value of the sum of vertical distances between a number of coordinate points formed by the first number of corresponding mutation data and the estimated virus diffusion parameters and the image of the linear fitting function of the first number of corresponding mutation data and the estimated virus diffusion parameters in the coordinate system with the first number of corresponding mutation data and the estimated virus diffusion parameters as coordinate parameters. Those skilled in the art will understand that the calculation process of the linear fit between the first number of corresponding mutation data and the estimated virus diffusion parameters in the virus propagation database is a conventional technical means well known to those skilled in the art, and therefore the calculation process of the linear fit will not be repeated here. ‌

[0035] In a specific embodiment, the input parameters of the large-scale virus transmission prediction model are the resident population distribution density of a large community and the floating population distribution density of a large community. An embodiment of the large-scale virus transmission prediction model is a first diffusion prediction parameter f(D, V) = k1×D + k2×V; Among them, D is the permanent population distribution density of a large community (people / km²), V is the floating population distribution density of a large community (people / km²), k1+k2=1, k1 is the weight coefficient of the permanent population distribution density, and k2 is the weight coefficient of the floating population distribution density.

[0036] Specifically, in the large-scale virus propagation prediction model, k1 is greater than k2. In this embodiment, k1=0.6 and k2=0.4.

[0037] It will be understood by those skilled in the art that the values ​​of k1 and k2 can be adaptively adjusted according to the density of residential buildings, floor heights and other characteristics of a large community to better suit the characteristics of the large community.

[0038] In a specific embodiment, the input parameters of the small-scene virus transmission prediction model are the resident population distribution density of a single unit area of ​​a large community and the floating population distribution density of a single unit area of ​​a large community. An embodiment of the small-scene virus transmission prediction model is a second diffusion prediction parameter g(P,I) = k3×P+k4×I; Wherein, P is the permanent population distribution density within a single unit area (persons / km²), I is the floating population distribution density within a single unit area (persons / km²), k3 is the weight coefficient of the permanent population distribution density within the unit area, k4 is the weight coefficient of the floating population distribution density within the unit area, k3+k4=1, k3 is less than k4, and in this embodiment, k3=0.45, k2=0.55.

[0039] Specifically, the total area of ​​several unit areas is equal to the total area of ​​a large community.

[0040] Those skilled in the art will appreciate that the values ​​of k3 and k4 may be adaptively adjusted based on the residential building density, floor height, and other characteristics of a single unit area in a large community.

[0041] Optionally, the preset range of the goodness of fit is [0.6 units, 0.9 units].

[0042] Preferably, the preferred embodiment of the preset fitness is 0.8 units.

[0043] It will be understood by those skilled in the art that, in actual application or implementation, those skilled in the art may adaptively adjust the preset degree of fit according to the actual application environment and application scenario.

[0044] Specifically, judging whether to remove corresponding mutation data based on the number of types of the mutation data includes: If the number of types of mutation data within the same monitoring period is zero, the corresponding mutation data is eliminated; If the number of types of mutation data within the same monitoring period is not zero, the corresponding mutation data is not eliminated.

[0045] In this embodiment, the present invention eliminates mutation data when the number of mutation data types within the same monitoring period is zero, and does not eliminate mutation data when the number is not zero. In the process of modeling the virus transmission prediction model for the virus transmission prediction process in a large community and the process of data collection and processing, when there is a strong time consistency between the monitored mutation data, it indicates that there is a certain invisible correlation between the mutation data within the same monitoring period. For example, in a large community, the spread of the virus may be weakened due to the presence of wind, but a small area of ​​wind enhancement may be formed in a small area between residential buildings, thereby exacerbating the virus transmission and diffusion ability in the small area. Therefore, through the detection and analysis of invisible correlations, abnormal factors that promote the enhancement of virus transmission ability can be identified, and the data cleaning and elimination process can be made more accurate, thereby improving the accuracy of virus transmission prediction.

[0046] Specifically, multiple types of virus transmission prediction feature data can be obtained within a single monitoring cycle. For example, the monitoring period for personal protective equipment coverage is 12 hours, and the monitoring period for virus residues on the surface of objects is 6 hours.

[0047] Specifically, the first number of classification numbers of the corresponding mutation data is negatively correlated with the linear fit.

[0048] In implementation, when the linear fit exceeds the preset linear fit by a value within 0.1, the number of categories of the mutation data is reduced by 1; when the linear fit exceeds the preset linear fit by a value exceeding 0.1, the number of categories of the mutation data is reduced by 1 for every 0.1 exceeding the preset linear fit; for example, in a possible embodiment, the linear fit exceeds the preset linear fit by a value of 0.2, and at this time the number of categories of the mutation data is the original number of categories of the mutation data minus 2.

[0049] In this embodiment, the present invention reduces the number of classifications of the corresponding mutation data that meets the diffusion association conditions, and the number of classifications of the mutation data is negatively correlated with the linear fit, that is, the greater the linear fit, the smaller the number of classifications of the mutation data. When the linear fit is large, it means that the correlation between the mutation data and the final diffusion parameter is large. The commonality of the mutation data is found, the classification of the mutation data is re-divided, and the number of classifications of the mutation data is reduced. This can reduce the complexity of the basic training set and improve the efficiency of training the model based on the basic data set used for model training.

[0050] Specifically, the mapping verification includes: Performing consistency mapping training on the second number of corresponding mutation data that do not meet the diffusion association condition and the estimated virus diffusion parameters in the virus transmission database; The replacement is performed based on inconsistent results of the consistent mapping training.

[0051] In this implementation, through mapping verification, the data with poor correlation with the spread of the virus is replaced, which improves the accuracy of data selection in the basic data set for model training, avoids the interference of corresponding mutation data that does not meet the diffusion association conditions during model training, and thus improves the accuracy of virus transmission prediction.

[0052] Specifically, the second quantity corresponds to the mutation data and the estimated virus diffusion parameters in the virus propagation database for consistency mapping training so that when the corresponding mutation data changes, the estimated virus diffusion parameters in the virus propagation database also change synchronously. For example, if the value of the mutation data increases and the estimated virus diffusion parameters in the virus propagation database also increases, it means that the corresponding mutation data is consistent with the estimated virus diffusion parameters in the virus propagation database.

[0053] Specifically, training the large-scale basic model and the small-scale basic model separately includes: Dividing the basic data set for model training into a training set, a test set, and a validation set; Using the training set to train the large-scale basic model, and sequentially using the test set and the validation set to test and update the large-scale basic model to output a large-scale virus propagation prediction model; The small-scene basic model is trained using the training set, and the small-scene basic model is tested and updated using the test set and the validation set in turn to output a small-scene virus propagation prediction model.

[0054] Specifically, the training set is the main data set used to learn model parameters, and the weights are adjusted to make the model fit the data patterns.

[0055] Validation set: used for hyperparameter tuning and model selection, and to monitor overfitting during training.

[0056] ‌Test set‌: An independent dataset for the final evaluation of the model's generalization ability, used only once after training is complete.

[0057] Specifically, the regional connectivity characteristic parameters of the single region and each adjacent region are determined based on the first corresponding virus diffusion parameter of the mutation data in the virus propagation database that meets the diffusion association condition and the second corresponding virus diffusion parameter of the mutation data in each adjacent region in the virus propagation database.

[0058] Specifically, the first corresponding virus diffusion parameter is a virus diffusion capacity characterization parameter corresponding to the mutation data in a single area within the same monitoring period in the virus transmission database; the virus diffusion capacity characterization parameter can characterize the virus diffusion capacity in the single area; for example, in a specific embodiment, when the mutation data in a single area within the same monitoring period is the ambient humidity of the corresponding number of sampling points in the single area and the ventilation frequency of all public spaces in the single area, then the virus transmission database has a virus diffusion capacity characterization parameter corresponding to the ambient humidity of the corresponding number of sampling points in the above single area and the ventilation frequency of all public spaces in the single area (except for the ambient humidity of the corresponding number of sampling points and the ventilation frequency of all public spaces in the single area, other virus transmission prediction feature data in a single area associated with the virus diffusion capacity adopt the standard value or average value corresponding to the single area).

[0059] Specifically, the parameters that characterize the virus's ability to spread and spread can be the basic reproduction number and the infection incidence rate.

[0060] Those skilled in the art will understand that the basic reproduction number is the average number of new infected persons that a typical infected person can infect in a completely susceptible population during the entire infectious period; the infection incidence rate is the proportion of new infections in a population in a specific time period and a specific area to the total population.

[0061] Specifically, the second corresponding virus diffusion parameter is a parameter characterizing the virus propagation and diffusion capability corresponding to the mutation data in each adjacent area of ​​a single area within the same monitoring period in the virus propagation database.

[0062] Specifically, the estimated virus diffusion parameters in the virus propagation database are parameters characterizing the virus propagation and diffusion capabilities.

[0063] Specifically, the regional connectivity characteristic parameter between the single region and each adjacent region is the difference between the second corresponding virus diffusion parameter and the first corresponding virus diffusion parameter.

[0064] Specifically, based on the absolute value of the difference between the first diffusion prediction parameter and the second diffusion prediction parameter being greater than a preset absolute value, the second diffusion prediction parameter is determined as the final diffusion parameter; based on the absolute value of the difference between the first diffusion prediction parameter and the second diffusion prediction parameter being less than or equal to the preset absolute value, the first diffusion prediction parameter is determined as the final diffusion parameter.

[0065] Optionally, the optional range of the preset absolute value is [500, 800].

[0066] Preferably, the preset absolute value is 700.

[0067] Those skilled in the art will appreciate that [500, 800], 700 are several optional embodiments and preferred embodiments. In actual applications or implementations, those skilled in the art may adaptively adjust the preset absolute values ​​according to the actual application environment and application scenarios.

[0068] Specifically, the calculation formula of the second diffusion prediction parameter is:

[0069] Among them, R is the second diffusion prediction parameter, ri is the basic virus diffusion parameter of the i-th small area, qij is the regional connectivity characteristic parameter of the j-th adjacent area of ​​the i-th small area, n is the number of adjacent areas of the i-th small area, and m is the total number of small areas.

[0070] Specifically, the sum of the first number and the second number is equal to the total number of mutation data that have not been eliminated.

[0071] Thus far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art may make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will fall within the scope of protection of the present invention.

Claims

1. A method for predicting virus spread in a large community, characterized in that: include: Obtaining certain virus transmission prediction feature data; Compare each virus transmission prediction feature data with the corresponding reasonable interval in the virus transmission database to identify several mutation data; Counting the types of mutation data in the same monitoring period among the plurality of mutation data; Determining whether to remove corresponding mutation data based on the number of types of the mutation data; For the mutation data that has not been eliminated, reducing the number of categories of the first number of mutation data that meet the diffusion association condition; Performing mapping verification on the second number of corresponding mutation data that do not meet the diffusion association condition and then replacing them with corresponding data in the virus propagation database; Eliminating the corresponding mutation data to output a basic data set for model training; Based on the basic data set for model training, the large-scale basic model and the small-scale basic model are trained in sequence to form a large-scale virus propagation prediction model and a small-scale virus propagation prediction model respectively; Inputting actual virus propagation feature data into the large-scale virus propagation prediction model and the small-scale virus propagation prediction model respectively to output a first diffusion prediction parameter and a second diffusion prediction parameter respectively; determining a final diffusion parameter for a large community based on the first diffusion prediction parameter, the second diffusion prediction parameter, and a regional connectivity characteristic parameter; Among them, the diffusion association condition is that the linear fit between the mutation data corresponding to the first quantity and the estimated virus diffusion parameters in the virus propagation database is greater than a preset fit.

2. The method for predicting virus spread in a large community according to claim 1, wherein: Determining whether to eliminate corresponding mutation data based on the number of types of the mutation data includes: If the number of types of mutation data within the same monitoring period is zero, the corresponding mutation data is eliminated; If the number of types of mutation data within the same monitoring period is not zero, the corresponding mutation data is not eliminated.

3. The method for predicting virus spread in a large community according to claim 2, wherein: The first number of classifications of the corresponding mutation data is negatively correlated with the linear fit.

4. The method for predicting virus spread in a large community according to claim 3, wherein: The mapping verification includes: Performing consistency mapping training on the second number of corresponding mutation data that do not meet the diffusion association condition and the estimated virus diffusion parameters in the virus transmission database; The replacement is performed based on inconsistent results of the consistent mapping training.

5. The method for predicting virus spread in a large community according to claim 4, characterized in that: Training the large scene basic model and the small scene basic model separately includes: Dividing the basic data set for model training into a training set, a test set, and a validation set; Using the training set to train the large-scale basic model, and sequentially using the test set and the validation set to test and update the large-scale basic model to output a large-scale virus propagation prediction model; The small-scene basic model is trained using the training set, and the small-scene basic model is tested and updated using the test set and the validation set in turn to output a small-scene virus propagation prediction model.

6. The method for predicting virus spread in a large community according to claim 5, characterized in that: Based on the first corresponding virus diffusion parameter of the mutation data in a single area within the same monitoring period that meets the diffusion association condition and the second corresponding virus diffusion parameter of the mutation data in each adjacent area in the virus transmission database, the regional connectivity characteristic parameters of the single area and each adjacent area are determined.

7. The method for predicting virus spread in a large community according to claim 6, wherein: The regional connectivity characteristic parameter between the single region and each adjacent region is the difference between the second corresponding virus diffusion parameter and the first corresponding virus diffusion parameter.

8. The method for predicting virus spread in a large community according to claim 7, wherein: determining the second diffusion prediction parameter as the final diffusion parameter based on an absolute value of a difference between the first diffusion prediction parameter and the second diffusion prediction parameter being greater than a preset absolute value; Based on the absolute value of the difference between the first diffusion prediction parameter and the second diffusion prediction parameter being less than or equal to the preset absolute value, the first diffusion prediction parameter is determined as the final diffusion parameter.

9. The method for predicting virus spread in a large community according to claim 8, wherein: The calculation formula of the second diffusion prediction parameter is: Among them, R is the second diffusion prediction parameter, r i is the basic diffusion parameter of the virus in the i-th small area, q ij is the regional connectivity characteristic parameter of the jth adjacent region of the ith small region, n is the number of adjacent regions of the ith small region, and m is the total number of small regions.

10. The method for predicting virus spread in a large community according to claim 1, wherein: The sum of the first number and the second number is equal to the total number of mutation data that have not been eliminated.

Images