A dengue fever prediction method and device considering extreme weather driving and a medium

Abstract

The application discloses a dengue fever prediction method and device considering extreme weather driving and a medium. The method comprises the following steps: acquiring case data and multi-source data of a target area, wherein the multi-source data comprises meteorological data, environmental data and socio-economic data; constructing extreme weather characteristics according to the meteorological data and a preset extreme weather threshold; obtaining epidemic spatiotemporal lag data according to the spatial autocorrelation relationship and the temporal autocorrelation relationship of the case data; adding the epidemic spatiotemporal lag term in a zero-inflated negative binomial regression model to obtain an initial epidemic prediction model; the epidemic prediction model comprises a quadratic term of the extreme weather characteristics; and performing parameter fitting on the initial epidemic prediction model based on the case data, the extreme weather characteristics, the epidemic spatiotemporal lag data and the multi-source data to obtain a fitted epidemic prediction model, so as to predict future epidemics. The application can effectively improve the credibility of regional dengue fever epidemic prediction under the background of climate change.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of infectious disease prediction, and in particular to a dengue fever prediction method and device considering extreme weather driving and a medium. BACKGROUND

[0002] Dengue fever is a mosquito-borne infectious disease caused by dengue virus. With global climate change, such as heat waves, extreme precipitation and meteorological drought, these changes have a significant impact on the survival environment of mosquitoes. Climate warming not only expands the distribution range and survival period of mosquitoes, but also expands the incidence area of dengue fever from low altitude to high altitude, from tropical regions to temperate regions, thereby increasing the risk of dengue fever transmission. Therefore, accurate prediction of dengue fever is of great significance for the prevention and control of infectious diseases.

[0003] Current dengue fever prediction techniques focus on routine observation of meteorological, population and socio-economic factors, ignoring in-depth analysis of extreme weather and environmental factors. Existing prediction methods use techniques such as time series statistical analysis, machine learning, deep learning, etc., but generally have weak model variable interpretation, do not fully consider the refined features of extreme weather and urban internal spatial impact factors, etc., resulting in low credibility of dengue fever prediction results. SUMMARY

[0004] To solve the above technical problems, the present application provides a dengue fever prediction method and device considering extreme weather driving and a medium, which can effectively improve the credibility of regional dengue fever epidemic prediction under the background of climate change.

[0005] The present application provides a dengue fever prediction method considering extreme weather driving, comprising:

[0006] Obtaining case data and multi-source data of a target area, the multi-source data including meteorological data, environmental data and socio-economic data;

[0007] According to the meteorological data and a preset extreme weather threshold, an extreme weather feature is constructed;

[0008] According to the spatial autocorrelation relationship and the temporal autocorrelation relationship of the case data, epidemic spatiotemporal lag data is obtained;

[0009] An epidemic prediction model is obtained by adding the epidemic spatiotemporal lag term in the zero-inflated negative binomial regression model; the epidemic prediction model includes a quadratic term of the extreme weather feature;

[0010] Based on the case data, the extreme weather feature, the epidemic spatiotemporal lag data and the multi-source data, parameter fitting is performed on the initial epidemic prediction model to obtain a fitted epidemic prediction model;

[0011] adopting the fitted epidemic prediction model to predict the future epidemic of the target area.

[0012] As an improvement of the above scheme, the case data of the target area is obtained, including:

[0013] Unstructured address text is obtained from the dengue fever case reports of the target area;

[0014] The unstructured address text is mapped to a standard address library to obtain the geographic coordinates corresponding to the cases;

[0015] The target area is divided into grid cells, and based on the geographic coordinates, the number of cases in each grid cell is calculated to obtain the gridded case data.

[0016] As an improvement of the above scheme, the meteorological data includes but is not limited to daily maximum temperature, daily minimum temperature, daily precipitation and drought data;

[0017] The environmental data includes but is not limited to air quality data, vegetation coverage and impervious surface ratio;

[0018] The socio-economic data includes but is not limited to mosquito density, population density, road network density, medical facility distribution and urban construction data.

[0019] As an improvement of the above scheme, the extreme weather features are constructed according to the meteorological data and a preset extreme weather threshold, including:

[0020] Based on the historical meteorological data of the historical base period, the quantile (such as the 90th or 95th quantile) of the historical meteorological data is calculated using the sliding window method to determine the extreme weather threshold;

[0021] According to the extreme weather threshold, the extreme weather features are extracted from the meteorological data, including the frequency, intensity and duration of each extreme weather event.

[0022] As an improvement of the above scheme, the epidemic spatio-temporal lag data is obtained according to the spatial autocorrelation relationship and the temporal autocorrelation relationship of the case data, including:

[0023] Based on the case data, a spatial weight matrix is constructed by calculating the spatial correlation index value of each grid cell and the adjacent area, and a reverse distance weight matrix is obtained by combining a preset reverse distance weighting radius;

[0024] According to the reverse distance weight matrix and the case data, the spatial lag data of each grid cell is calculated;

[0025] performing least square regression on the case data to obtain residual, and calculating a DW statistic according to the residual;

[0026] judging whether there is a first-order positive correlation in the case data according to the DW statistic, and if yes, taking the spatial lag data of lag 1 as epidemic spatiotemporal lag data.

[0027] As an improvement of the above scheme, the parameter fitting of the initial epidemic prediction model based on the case data, the extreme weather feature, the epidemic spatiotemporal lag data, and the multi-source data, to obtain a fitted epidemic prediction model, comprises:

[0028] taking the extreme weather feature, the epidemic spatiotemporal lag data, and the multi-source data as a first variable set, and constructing a basic regression model of each type of extreme weather according to the first variable set and the case data;

[0029] calculating a variance inflation factor of each of the extreme weather features based on the basic regression model, and screening variables in the first variable set according to the variance inflation factor to obtain a second variable set;

[0030] based on the second variable set, adding a penalty term in the basic regression model of each type of extreme weather according to a Lasso regression algorithm, and adopting a stepwise regression method to remove variables with a contribution rate less than a preset threshold to obtain a third variable set;

[0031] performing parameter fitting of the initial epidemic prediction model according to the third variable set and the case data to obtain a fitted epidemic prediction model.

[0032] As an improvement of the above scheme, the epidemic prediction model comprises a counting part and a zero inflation part;

[0033] the expression of the counting part is:

[0034]

[0035] wherein, denotes a grid number, denotes time, denotes a grid number, is the natural logarithm expectation of the number of dengue fever cases in a grid unit, denotes an observation value of an extreme weather feature, denotes an influence coefficient of a quadratic term of each extreme weather feature on dengue fever, denotes an influence coefficient of an extreme weather feature on dengue fever, is epidemic spatiotemporal lag data, is a dengue fever spatial effect coefficient, The multi-source data represents unobservable parts and individual heterogeneity as control variables, observed values representing the control variables, coefficients representing the influence of each control variable on dengue fever;

[0036] observed values of the extreme weather characteristics, including observed values of individual extreme weather characteristics and / or cross terms of different extreme weather characteristics.

[0037] As an improvement of the above scheme, the expression of the zero inflation part is:

[0038]

[0039] wherein, is the probability of the existence of zero inflation, is the influence coefficient of the zero inflation factor, is the observed value of the zero inflation factor, is a constant term; the zero inflation factor includes the nighttime light intensity in the multi-source data and the imported case in the case data.

[0040] The embodiment of the present application also provides a dengue fever prediction device considering extreme weather driving, comprising:

[0041] a data acquisition module configured to acquire case data and multi-source data of a target area, wherein the multi-source data includes meteorological data, environmental data and socio-economic data;

[0042] an extreme weather characteristic module configured to construct extreme weather characteristics according to the meteorological data and a preset extreme weather threshold;

[0043] a spatio-temporal lag data module configured to obtain epidemic spatio-temporal lag data according to the spatial autocorrelation relationship and the temporal autocorrelation relationship of the case data;

[0044] a model construction module configured to add the epidemic spatio-temporal lag term in a zero inflation negative binomial regression model to obtain an initial epidemic prediction model; the epidemic prediction model includes a quadratic term of the extreme weather characteristics;

[0045] a parameter fitting module configured to perform parameter fitting on the initial epidemic prediction model based on the case data, the extreme weather characteristics, the epidemic spatio-temporal lag data and the multi-source data to obtain a fitted epidemic prediction model;

[0046] an epidemic prediction module configured to predict future epidemics of the target area by using the fitted epidemic prediction model.

[0047] The embodiment of the present application also provides a computer readable storage medium, the computer readable storage medium stores a computer program, wherein the computer program controls the device where the computer readable storage medium is located to execute the dengue fever prediction method considering extreme weather driving when the computer program runs.

[0048] Compared with the prior art, the embodiment of the present application provides a dengue fever prediction method, device and medium considering extreme weather driving, which has the beneficial effects that: by constructing a zero-inflated negative binomial spatio-temporal lag model, and combining case data, extreme weather characteristics, spatio-temporal lag data of the epidemic and multi-source data (including meteorological data, environmental data and socio-economic data), reliable prediction of the dengue fever epidemic is realized; by multiple screening of the input variables of the model, the input data of the model has a more reasonable driving mechanism, and the prediction performance of the model is improved; by considering the spatio-temporal autocorrelation of dengue fever cases in adjacent areas, and the time lag effect and nonlinear relationship of climate factors, the dynamic relationship between extreme weather and dengue fever transmission can be more accurately captured, and the accuracy of the model prediction is improved. Compared with the prediction methods such as deep learning model and large model, the model of the present application has low complexity, low cost and strong interpretability, has high credibility, and is helpful for the public health department to accurately grasp the epidemic trend of the dengue fever epidemic, and provides a scientific basis for formulating effective prevention and control strategies. BRIEF DESCRIPTION OF DRAWINGS

[0049] Figure 1 is a flowchart of a dengue fever prediction method considering extreme weather driving provided by the embodiment of the present application.

[0050] Figure 2 is a structural schematic diagram of a dengue fever prediction device considering extreme weather driving provided by the embodiment of the present application. DETAILED DESCRIPTION

[0051] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0052] Please refer to Figure 1 , Figure 1 is a flowchart of a dengue fever prediction method considering extreme weather driving provided by the embodiment of the present application. The dengue fever prediction method considering extreme weather driving comprises:

[0053] S1: acquiring case data of a target area and multi-source data, the multi-source data comprising meteorological data, environmental data and socio-economic data;

[0054] Specifically, first, according to historical dengue fever epidemic data, the target area of the study is determined, and an area with a high incidence of epidemic or a representative area can be selected as the target area, and it is ensured that the target area has available high-precision meteorological data to facilitate subsequent construction of a prediction model. When determining the size of the target area, the spatial scale of the data needs to be considered to avoid the situation that the high-precision data sample is too large due to the excessively large area, which results in excessively high calculation cost, and also to avoid the situation that the driving effect of each variable cannot be accurately evaluated due to the excessively small area.

[0055] As one of the optional embodiments, the case data of the target area is obtained, including:

[0056] Unstructured address text is obtained from the dengue fever case report of the target area;

[0057] The unstructured address text is mapped to a standard address library to obtain the geographic coordinates corresponding to the case;

[0058] The target area is divided into grids, and the number of cases in each grid unit is calculated based on the geographic coordinates to obtain gridded case data.

[0059] Specifically, the dengue fever case report of the target area is obtained, and the unstructured address text of each case is obtained from the attribute data of the report; then, the unstructured address text is mapped to a standard address library by using a place name address matching engine to realize geographic coding of the address. Specifically, a high-quality standard address library is constructed by combining the Internet and open-source geographic information collection data, automatic address data cleaning and quality inspection, and the like. By using a combination of a quantitative matching strategy and a bidirectional index, the unstructured address text to be matched is quickly matched with the standard address after hierarchical segmentation to obtain accurate geographic coordinates of each case. The geographic coordinates of the case are subjected to spatial distributed visualization processing. Specifically, a fishing net with the same size as the research unit is constructed, the number of dengue fever cases in each grid is calculated, and gridded dengue fever case spatial distribution data at a time scale of, for example, a day or a month, that is, case data, is formed. The number of dengue fever cases can also be subdivided into the number of dengue fever cases of different population characteristics (such as gender, age, etc.).

[0060] The embodiment of the application adopts an efficient and accurate place name address matching technology based on a standard address library to ensure that the dengue fever cases can be accurately positioned in space and the precision loss caused by input errors of the model is reduced.

[0061] As one of the optional embodiments, the meteorological data includes but is not limited to daily maximum temperature, daily minimum temperature, daily precipitation, and drought condition data.

[0062] The environmental data includes but is not limited to air quality data, vegetation coverage and impervious surface ratio;

[0063] The socio-economic data includes but is not limited to mosquito density, population density, road network density, medical facility distribution and urban construction data.

[0064] Specifically, high-precision meteorological data (such as 1km or 0.05 latitude and longitude precision) of the target area is obtained, including daily maximum temperature, daily minimum temperature, daily precipitation and drought data.

[0065] Air quality data (such as PM2.5, PM10, O3, SO2 concentration, etc.) that has an impact on the spread of dengue fever, as well as environmental data such as vegetation coverage and impervious surface ratio, are obtained.

[0066] Socio-economic data closely related to the spread of dengue fever, such as mosquito density, population density, road network density, medical facility distribution and urban construction (such as sewage concentration, public toilet density, etc. Facilities that may have water accumulation and provide breeding environment for mosquitoes) are obtained.

[0067] Further, the obtained case data and multi-source data are subjected to data cleaning and quality checking, including 99% tail processing of outliers, deletion of missing values, replacement of missing values with approximate values, etc. At the same time, standardization processing is carried out to ensure the consistency of data coordinates, numerical format and the rationality of dimension units, etc.

[0068] The embodiment of the application comprehensively considers historical case data, meteorological environment, socio-economic conditions and other various driving factors closely related to the incidence of dengue fever, so that the input data of the prediction model has a more reasonable driving mechanism, thereby enhancing the interpretability of the model and facilitating the revelation of the influence mechanism and rules of dengue fever.

[0069] S2: constructing extreme weather features according to the meteorological data and a preset extreme weather threshold;

[0070] As one of the optional embodiments, the constructing extreme weather features according to the meteorological data and a preset extreme weather threshold includes:

[0071] Based on the historical meteorological data of the historical base period, the quantile of the historical meteorological data is calculated by using the sliding window method to determine the extreme weather threshold;

[0072] According to the extreme weather threshold, the extreme weather features are extracted from the meteorological data, and the extreme weather features include the frequency, intensity and duration of each extreme weather event.

[0073] Specifically, the threshold of each extreme weather is determined first: based on the historical meteorological data of a historical base period (e.g., 1961-1990), the quantile of the historical same period is calculated using a sliding window method (e.g., taking 5 days as a window period) as the threshold of the extreme weather event. For example, according to the 90th quantile of the daily maximum temperature, the daytime heat wave threshold is obtained; according to the 90th quantile of the daily minimum temperature, the nighttime heat wave threshold is obtained; according to the 95th quantile of the daily precipitation, the nighttime heat wave threshold is obtained. For extreme climate events such as drought, the grade limit value representing no drought and light drought in the meteorological drought index such as SPEI can be directly used as the corresponding extreme weather threshold.

[0074] Then the extreme weather feature variable is constructed: according to the obtained meteorological data of the target region, the monthly frequency, monthly event average intensity and monthly event average duration of the extreme weather events such as the daytime heat wave event (DHW), the nighttime heat wave event (NHW) and the extreme precipitation event (EP) in the past three days exceeding the extreme weather threshold are calculated; for the drought event, the severity of the drought event is determined according to the drought index and the corresponding threshold.

[0075] S3: obtaining the epidemic spatiotemporal lag data according to the spatial autocorrelation relationship and the temporal autocorrelation relationship of the case data;

[0076] As one of the optional embodiments, the obtaining the epidemic spatiotemporal lag data according to the spatial autocorrelation relationship and the temporal autocorrelation relationship of the case data comprises:

[0077] Based on the case data, a spatial weight matrix is constructed by calculating the spatial correlation index value of each grid cell and the adjacent region, and a reverse distance weight matrix is obtained in combination with a preset reverse distance weighting radius;

[0078] According to the reverse distance weight matrix and the case data, the spatial lag data of each grid cell is calculated;

[0079] The least square method is used to regress the case data to obtain a residual, and a DW statistic is calculated according to the residual.

[0080] According to the DW statistic, it is determined whether there is a first-order positive correlation in the case data, and if so, the spatial lag data of the lag 1 period is taken as the epidemic spatiotemporal lag data.

[0081] Specifically, the spatial correlation indicators include Moran's I and LISA (Local Local Indicators of Spatial Association) index. The spatial distribution of dengue fever case data in each grid cell of the target region is analyzed by using the global Moran's I and the local LISA index to determine whether there is a spatial autocorrelation relationship in the distribution of dengue fever. When Moran's I > 0, it indicates that there is a positive spatial correlation, and the larger the value of the index, the greater the spatial autocorrelation. The calculation formula of Moran's I index is:

[0082]

[0083] wherein, represents the number of grid cells in the target region, or the number of cases in the cell or , is the average number of cases in cells, is the weight matrix defining the relationship between geographical units - .

[0084] The LISA index represents the clustering relationship of each grid cell with adjacent areas, and the calculation formula is:

[0085]

[0086] wherein, represents the number of grid cells in the target region, or the number of cases in the cell or , is the average number of cases in cells, is the variance, is the weight matrix defining the relationship between geographical units - .

[0087] Further, by calculating the spatial correlation indicators, a spatial weight matrix is constructed. Considering the spatial scale of the study, the spatial distribution range of dengue fever and the range of mosquito activity, a suitable inverse distance weighting radius (such as 2 km) is set to reflect the influence relationship between geographical locations based on the square inverse of the distance. According to the spatial weight matrix and the inverse distance weighting radius, the inverse distance weight matrix of the target region is obtained.

[0088] Further, the spatial lag data of each grid cell is calculated using the inverse distance weight matrix and the dengue case data, which represents the degree of influence of the grid cell by the surrounding cases and reflects the spatial correlation of the dengue epidemic affected by the neighboring areas.

[0089] Further, the least squares OLS regression is performed on the time series data of the case data, and the DW statistic is calculated according to the model residual:

[0090]

[0091] When the DW statistic is located in the interval (0, dL), it indicates that the case data presents a first-order positive autocorrelation in time, where dL is the lower critical value of the DW test. Considering the possible time lag effect between adjacent areas, the dengue epidemic data of the past time unit (such as the last week or the last month) is regarded as a key reference for predicting the current epidemic, and this past time point is referred to as lag 1. That is, the dengue epidemic data of lag 1 considering the spatial and temporal lag effects (i.e., epidemic spatio-temporal lag data) is included in the epidemic prediction model as an input variable of the epidemic prediction model.

[0092] S4: Add the epidemic spatio-temporal lag term in the zero-inflated negative binomial regression model to obtain an initial epidemic prediction model; the epidemic prediction model includes the quadratic term of the extreme weather feature;

[0093] Specifically, the general case data does not occur in every grid cell, so the possibility of zero inflation and over-dispersion distribution of case data distribution is larger in the results of zero inflation and over-dispersion analysis, and therefore the zero-inflated negative binomial regression model is selected as the basic model of the epidemic prediction model.

[0094] Wherein, the judgment of zero inflation is: according to the case data, the frequency distribution graph is drawn, the proportion of zero value of the grid cell is calculated, and if the proportion of zero value is too high, it indicates that there is zero inflation. The judgment of over-dispersion is: according to the distribution characteristics of the case data, the O statistic is calculated, and the calculation formula is:

[0095]

[0096] In the formula, N represents the number of samples, μ represents the mean of the number of dengue cases of all samples, σ represents the standard deviation of the number of dengue cases. If the O statistic ≥ 1.96, it indicates that the data has over-dispersion, and the negative binomial distribution model should be selected.

[0097] Furthermore, if the data exhibits zero inflation, then the zero part needs to be modeled separately. Usually, the occurrence of dengue fever outbreaks is related to imported cases and human activities, while nighttime light intensity is a proxy variable that comprehensively reflects human activities. Therefore, the number of imported cases and nighttime light intensity are selected as the zero inflation factors in the zero-inflation negative binomial regression model.

[0098] Furthermore, considering that the impact of extreme weather on dengue fever generally exhibits a nonlinear relationship of first promoting and then inhibiting, a quadratic term is added to the model to express this nonlinearity. The spatiotemporal lag term of dengue fever cases (corresponding to the spatiotemporal lag data of the epidemic) is integrated into the zero-inflation negative binomial regression model to form the zero-inflation negative binomial spatiotemporal lag regression model (ZINB-STLRM), i.e., the epidemic prediction model, in order to improve the model's fitting effect.

[0099] As one optional embodiment, the epidemic prediction model includes a counting component and a zero-inflation component;

[0100] The expression for the counting part is:

[0101]

[0102] The expression for the zero-inflation part is:

[0103]

[0104] in, Indicates the grid number, Indicates time, Indicates the number of grid cells. Let e ​​be the expected value of the natural logarithm of the number of dengue fever cases within a grid cell. Observations representing the characteristics of extreme weather The quadratic term representing the influence coefficient of various extreme weather characteristics on the dengue fever epidemic. This represents the influence coefficient of extreme weather characteristics on dengue fever outbreaks. For the spatiotemporal lag data of the epidemic (i.e., the first...) Each grid cell in time Observations at time Based on spatial weight The weighted result represents dengue fever cases in adjacent regions with a lag of one period. This is the dengue fever spatial effect coefficient. Representing unobservable components and individual heterogeneity, the multi-source data serves as control variables. Represents the observed values ​​of the control variables. This represents the influence coefficient of each control variable on the dengue fever epidemic; The probability of zero inflation exists. an influence coefficient of a zero inflation factor, an observed value of a zero inflation factor, a constant term;

[0105] The zero inflation factor includes night light intensity in the multi-source data and imported case number in the case data, to reflect population density and socio-economic activity characteristics.

[0106] The zero inflation negative binomial spatio-temporal lag regression model constructed in the embodiment comprehensively considers zero inflation phenomenon, over-dispersion phenomenon and spatio-temporal autocorrelation of neighboring dengue fever cases, and improves prediction accuracy. In the context of global warming, extreme weather is a driving force for the spread of dengue fever, and the occurrence of dengue fever is the result of the comprehensive action of various meteorological factors. Small-scale extreme weather may promote the spread of dengue fever virus, while extreme weather exceeding a certain intensity may inhibit the occurrence of dengue fever, and the driving effect may have a time lag effect. Therefore, by considering the time lag effect, nonlinear relationship and the like of climate factors, the embodiment can more accurately capture the dynamic relationship between extreme weather and the spread of dengue fever, and improve the credibility of the prediction result.

[0107] S5: fitting parameters of the initial epidemic prediction model based on the case data, the extreme weather feature, the epidemic spatio-temporal lag data and the multi-source data, to obtain a fitted epidemic prediction model;

[0108] As one of the optional embodiments, the fitting of parameters of the initial epidemic prediction model based on the case data, the extreme weather feature, the epidemic spatio-temporal lag data and the multi-source data to obtain a fitted epidemic prediction model comprises:

[0109] taking the extreme weather feature, the epidemic spatio-temporal lag data and the multi-source data as a first variable set, and constructing a basic regression model of each type of extreme weather according to the first variable set and the case data;

[0110] calculating a variance inflation factor of each of the extreme weather features based on the basic regression model, and screening variables in the first variable set according to the variance inflation factor to obtain a second variable set;

[0111] based on the second variable set, adding a penalty term to the basic regression model of each type of extreme weather according to a Lasso regression algorithm, and using a stepwise regression method to remove variables with a contribution rate less than a preset threshold to obtain a third variable set;

[0112] fitting parameters of the initial epidemic prediction model according to the third variable set and the case data to obtain a fitted epidemic prediction model.

[0113] Specifically, to improve the predictive performance of the model, the input variables were first screened. First, a basic regression model for each type of extreme weather was constructed based on extreme weather characteristics, spatiotemporal lag data of the dengue fever outbreak, multi-source data, and case data. The variance inflation factor (VIF) was used to detect multicollinearity. Multicollinearity refers to a high degree of linear correlation between variables in the regression model, which can lead to inaccurate estimation of regression coefficients. Therefore, highly correlated variables were screened out using a VIF > 10 criterion. Next, extreme weather variables were identified, and Lasso regression was used to select control variables. A penalty term was added to the linear regression model for each type of extreme weather, that is, the sum of the absolute values ​​of the feature coefficients multiplied by the parameter λ, to control the complexity of the model, allowing the model to select fewer features and thus reducing the risk of overfitting. Then, stepwise regression was used to further screen control variables, identifying those significantly associated with dengue fever outbreaks. Specifically, the control variables were analyzed based on their relationship to the model. The contribution rate was eliminated in order. Variables whose contribution rate is less than a threshold (e.g., 0.01) are used to obtain the final model input variables (i.e., the third variable set).

[0114] Furthermore, based on case data and screened variables (including multi-source data, extreme weather characteristics, and spatiotemporal lag data of the epidemic), the parameters of the epidemic prediction model are fitted. The fitting effect of the model is judged by the AIC criterion and / or BIC criterion. The parameters of the optimal model are selected as the fitted model parameters to obtain the fitted epidemic prediction model.

[0115] During model fitting, the Reporting Relative Risk Ratio (IRR), or exp(β), is selected to explain the regression coefficients of the model using scientific theory. The IRR coefficient illustrates the promoting or inhibiting effects of various extreme weather characteristics and their related driving factors on the dengue fever transmission risk. Simultaneously, the model's performance in terms of spatiotemporal lag and nonlinear effects is examined to verify the rationality of the model selection. For example, a positive spatiotemporal lag term indicates that the number of cases in neighboring regions in the previous period will contribute to an increase in the number of dengue fever cases in the current period in this region.

[0116] Furthermore, the observations of the extreme weather features in the epidemic prediction model include observations of individual extreme weather features and / or interaction terms of different extreme weather features.

[0117] Specifically, if we take the observed value of a single extreme weather characteristic as... Substituting these into the model allows us to analyze the impact of individual extreme weather features on model predictions. Furthermore, by adding interaction terms of different extreme weather features to the model, we can study the interactions between different extreme weather events. For example, we can use daytime heatwave intensity and nighttime heatwave intensity, along with their interaction terms, as... Substitute the model, if the model fitting cross product term appears negative coefficient, namely the relative risk ratio IRR < 1, it indicates that the two extreme weather characteristics show antagonism, can reduce the risk of dengue fever transmission.

[0118] Further, a test set is obtained, the fitted epidemic prediction model is verified, and indicators such as MAE and RMSE are used to evaluate the prediction performance of the model, to ensure the reliability of the epidemic prediction model.

[0119] S6: using the fitted epidemic prediction model to predict the future epidemic of the target area.

[0120] Specifically, the future climate prediction data set and the related control variable prediction data set of the target area are obtained, and the fitted epidemic prediction model is used to predict the dengue fever epidemic of the target area, to obtain the prediction result. Further, according to the mechanism behind the model coefficients and the development prediction of the regional dengue fever epidemic, early warning and prevention and control suggestions are provided for the public health department, and the dengue fever prevention and control strategy is optimized.

[0121] The embodiment of the present application realizes reliable prediction of dengue fever epidemic by constructing a zero-inflated negative binomial spatio-temporal lag model and combining case data, extreme weather characteristics, epidemic spatio-temporal lag data and multi-source data (including meteorological data, environmental data and socio-economic data); through multiple screening of model input variables, the model input data have a more reasonable driving mechanism, and the prediction performance of the model is improved; by considering the spatio-temporal autocorrelation of dengue fever cases in adjacent areas and the time lag effect and nonlinear relationship of climate factors, the dynamic relationship between extreme weather and dengue fever transmission can be more accurately captured, and the accuracy of model prediction is improved. Compared with the prediction methods such as deep learning model and large model, the model of the present application has low complexity, low cost and strong interpretability, has high credibility, and is helpful for the public health department to accurately master the epidemic trend of dengue fever, and provides a scientific basis for formulating effective prevention and control strategies.

[0122] Correspondingly, the present application also provides a dengue fever prediction device considering extreme weather driving, which can realize all processes of the dengue fever prediction method considering extreme weather driving in the above embodiment.

[0123] Please refer to Figure 2 , Figure 2 is a structure diagram of a dengue fever prediction device considering extreme weather driving provided by the embodiment of the present application. The dengue fever prediction device considering extreme weather driving comprises:

[0124] The data acquisition module 201 is configured to acquire case data and multi-source data of a target area, wherein the multi-source data comprises meteorological data, environmental data and socio-economic data.

[0125] an extreme weather feature module 202, configured to construct an extreme weather feature according to the meteorological data and a preset extreme weather threshold value;

[0126] a spatio-temporal lag data module 203, configured to obtain spatio-temporal lag data of the epidemic according to a spatial autocorrelation relationship and a temporal autocorrelation relationship of the case data;

[0127] a model construction module 204, configured to add a spatio-temporal lag term of the epidemic into a zero-inflated negative binomial regression model to obtain an initial epidemic prediction model; the epidemic prediction model comprises a quadratic term of the extreme weather feature;

[0128] a parameter fitting module 205, configured to perform parameter fitting on the initial epidemic prediction model based on the case data, the extreme weather feature, the spatio-temporal lag data of the epidemic, and the multi-source data, to obtain a fitted epidemic prediction model;

[0129] an epidemic prediction module 206, configured to predict a future epidemic of a target region by using the fitted epidemic prediction model.

[0130] Preferably, the case data of the target region is obtained by:

[0131] obtaining unstructured address text from dengue fever case reports of the target region;

[0132] mapping the unstructured address text to a standard address library to obtain geographical coordinates corresponding to the cases;

[0133] performing grid segmentation on the target region, and calculating the number of cases of each grid unit based on the geographical coordinates to obtain gridded case data.

[0134] Preferably, the meteorological data includes but is not limited to daily maximum temperature, daily minimum temperature, daily precipitation, and drought condition data;

[0135] The environmental data includes but is not limited to air quality data, vegetation coverage, and impervious surface ratio;

[0136] The socio-economic data includes but is not limited to mosquito vector density, population density, road network density, medical facility distribution, and urban construction condition data.

[0137] Preferably, the extreme weather feature module 202 is specifically configured to:

[0138] based on historical meteorological data of a historical base period, calculate quantiles of the historical meteorological data by using a sliding window method to determine an extreme weather threshold value;

[0139] According to the extreme weather threshold, an extreme weather feature is extracted from the weather data, the extreme weather feature including frequency, intensity and duration of each extreme weather event.

[0140] Preferably, the spatio-temporal lag data module 203 is specifically configured to:

[0141] Based on the case data, a spatial weight matrix is constructed by calculating a spatial correlation index value of each grid cell and adjacent areas, and a reverse distance weight matrix is obtained in combination with a preset reverse distance weighting radius;

[0142] According to the reverse distance weight matrix and the case data, the spatial lag data of each grid cell is calculated;

[0143] The case data is subjected to least square regression to obtain a residual, and a DW statistic is calculated according to the residual;

[0144] According to the DW statistic, it is determined whether there is a first-order positive correlation in the case data, and if so, the spatial lag data of the lag 1 period is taken as the epidemic spatio-temporal lag data.

[0145] Preferably, the parameter fitting module 205 is specifically configured to:

[0146] The extreme weather feature, the epidemic spatio-temporal lag data and the multi-source data are taken as a first variable set, and a basic regression model of each type of extreme weather is constructed according to the first variable set and the case data;

[0147] Based on the basic regression model, a variance inflation factor of each extreme weather feature is calculated, and variables are screened in the first variable set according to the variance inflation factor, to obtain a second variable set;

[0148] Based on the second variable set, a penalty term is added to the basic regression model of each type of extreme weather according to a Lasso regression algorithm, and a stepwise regression method is adopted to remove variables with a contribution rate less than a preset threshold to the model, to obtain a third variable set;

[0149] According to the third variable set and the case data, the initial epidemic prediction model is subjected to parameter fitting to obtain a fitted epidemic prediction model.

[0150] Preferably, the epidemic prediction model includes a count part and a zero inflation part;

[0151] The expression of the count part is:

[0152]

[0153] wherein, represents a grid number, denotes time, denotes the number of grids, is the expected natural logarithm with base e of the number of dengue cases in a grid unit, denotes the observation value of the extreme weather feature, denotes the impact coefficient of the quadratic term of each extreme weather feature on dengue fever, denotes the impact coefficient of the extreme weather feature on dengue fever, is the spatiotemporal lag data of the epidemic, is the spatial effect coefficient of dengue fever, represents the unobservable part and individual heterogeneity, and the multi-source data as a control variable, denotes the observation value of the control variable, denotes the impact coefficient of each control variable on dengue fever;

[0154] The observation value of the extreme weather feature includes the observation value of a single extreme weather feature and / or the cross term of different extreme weather features.

[0155] Preferably, the expression of the zero inflation part is:

[0156]

[0157] wherein, is the probability of the existence of zero inflation, is the impact coefficient of the zero inflation factor, is the observation value of the zero inflation factor, is a constant term; the zero inflation factor includes the nighttime light intensity in the multi-source data and the imported case in the case data.

[0158] In specific implementations, the working principle, control process and technical effects of the dengue fever prediction device considering extreme weather driving provided by the embodiments of the present application correspond to the dengue fever prediction method considering extreme weather driving in the above embodiments, and will not be repeated here.

[0159] The embodiments of the present application also provide a computer readable storage medium, which includes a stored computer program, wherein when the computer program runs, the device where the computer readable storage medium is located executes the dengue fever prediction method considering extreme weather driving provided by any of the above embodiments.

[0160] The embodiment of the present application provides a dengue fever prediction method and device considering extreme weather driving and a medium, which has the beneficial effect that: by constructing a zero-inflated negative binomial spatio-temporal lag model, and combining case data, extreme weather characteristics, epidemic spatio-temporal lag data and multi-source data (including meteorological data, environmental data and socio-economic data), reliable prediction of dengue fever epidemic is realized; through multiple screening of model input variables, the model input data has a more reasonable driving mechanism, and the prediction performance of the model is improved; by considering the spatio-temporal autocorrelation of dengue fever cases in adjacent areas, and the time lag effect and nonlinear relationship of climate factors, the dynamic relationship between extreme weather and dengue fever transmission can be more accurately captured, and the accuracy of model prediction is improved. Compared with the prediction method of deep learning model and large model, the model of the present application has low complexity, low cost and strong interpretability, has high credibility, is helpful for the public health department to accurately master the epidemic trend of dengue fever, and provides a scientific basis for formulating effective prevention and control strategies.

[0161] The above is the preferred embodiment of the present application, it should be pointed out that, for those skilled in the art, without departing from the principles of the present application, a number of improvements and refinements can be made, these improvements and refinements are also considered to be within the scope of protection of the present application.

Claims

1. A dengue fever prediction method considering extreme weather drivers, characterized in that, include: Acquire case data and multi-source data for the target area, including meteorological data, environmental data, and socioeconomic data; the socioeconomic data includes, but is not limited to, mosquito density, population density, road network density, distribution of medical facilities, and urban construction data. Based on the meteorological data and preset extreme weather thresholds, extreme weather characteristics are constructed; Based on the spatial and temporal autocorrelation relationships of the case data, spatiotemporal lag data of the epidemic are obtained. This process includes: based on the case data, constructing a spatial weight matrix by calculating the spatial correlation index value between each grid cell and its neighboring areas, and obtaining an inverse distance weight matrix by combining it with a preset inverse distance weighting radius; calculating the spatial lag data of each grid cell based on the inverse distance weight matrix and the case data; performing least squares regression on the case data to obtain residuals, and calculating the DW statistic based on the residuals; determining whether the case data exhibits a first-order positive correlation based on the DW statistic, and if so, using the spatial lag data lagged by one period as the spatiotemporal lag data of the epidemic. By adding a spatiotemporal lag term for the epidemic to the zero-inflation negative binomial regression model, an initial epidemic prediction model is obtained; the epidemic prediction model includes a quadratic term of the extreme weather characteristics. Based on the case data, extreme weather characteristics, spatiotemporal lag data of the epidemic, and multi-source data, the initial epidemic prediction model is fitted with parameters to obtain a fitted epidemic prediction model. The process of fitting parameters to the initial epidemic prediction model based on the case data, extreme weather characteristics, spatiotemporal lag data of the epidemic, and multi-source data to obtain a fitted epidemic prediction model includes: using the extreme weather characteristics, the spatiotemporal lag data of the epidemic, and the multi-source data as a first variable set; constructing a basic regression model for each type of extreme weather based on the first variable set and the case data; calculating the variance inflation factor for each extreme weather characteristic based on the basic regression model; filtering variables in the first variable set based on the variance inflation factor to obtain a second variable set; adding a penalty term to the basic regression model for each type of extreme weather based on the second variable set using the Lasso regression algorithm, and using stepwise regression to remove variables whose contribution rate to the model is less than a preset threshold to obtain a third variable set; and fitting parameters to the initial epidemic prediction model based on the third variable set and the case data to obtain a fitted epidemic prediction model. The fitted epidemic prediction model is used to predict the future epidemic situation in the target area.

2. The dengue fever prediction method considering extreme weather driving as described in claim 1, characterized in that, The acquisition of case data in the target area includes: Obtain unstructured address text from dengue fever case reports in the target area; The unstructured address text is mapped to a standard address library to obtain the geographical coordinates corresponding to the case. The target area is divided into grids, and the number of cases in each grid cell is calculated based on the geographic coordinates to obtain gridded case data.

3. The dengue fever prediction method considering extreme weather as described in claim 1, characterized in that, The meteorological data includes, but is not limited to, daily maximum temperature, daily minimum temperature, daily precipitation, and drought data; The environmental data includes, but is not limited to, air quality data, vegetation coverage, and the proportion of impermeable surfaces.

4. The dengue fever prediction method considering extreme weather driving as described in claim 1, characterized in that, The step of constructing extreme weather characteristics based on the meteorological data and preset extreme weather thresholds includes: Based on historical meteorological data from the historical base period, the sliding window method is used to calculate the quantiles of the historical meteorological data to determine the extreme weather threshold. Based on the extreme weather threshold, extreme weather features are extracted from the meteorological data, including the frequency, intensity, and duration of each extreme weather event.

5. The dengue fever prediction method considering extreme weather driving as described in claim 1, characterized in that, The epidemic prediction model includes a counting component and a zero-inflation component; The expression for the counting part is: in, Indicates the grid number, Indicates time, Indicates the number of grid cells. Let e ​​be the expected value of the natural logarithm of the number of dengue fever cases within a grid cell. Observations representing the characteristics of extreme weather The quadratic term representing the influence coefficient of various extreme weather characteristics on the dengue fever epidemic. This represents the influence coefficient of extreme weather characteristics on dengue fever outbreaks. For data with a time and space lag due to the pandemic, This is the dengue fever spatial effect coefficient. Representing unobservable components and individual heterogeneity, the multi-source data serves as control variables. Represents the observed values ​​of the control variables. This represents the influence coefficient of each control variable on the dengue fever epidemic; The observed values ​​of the extreme weather features include the observed values ​​of a single extreme weather feature and / or the interaction terms of different extreme weather features.

6. The dengue fever prediction method considering extreme weather driving as described in claim 5, characterized in that, The expression for the zero-inflation part is: in, The probability of zero inflation exists. The influence coefficient of zero expansion factor, The observed value is the one with zero inflation factor. The zero inflation factor is a constant term; it includes the nighttime light intensity in the multi-source data and the number of imported cases in the case data.

7. A dengue fever prediction device considering extreme weather-driven factors, characterized in that, include: The data acquisition module is used to acquire case data and multi-source data of the target area. The multi-source data includes meteorological data, environmental data and socio-economic data. The socio-economic data includes, but is not limited to, data on mosquito density, population density, road network density, distribution of medical facilities and urban construction. The extreme weather feature module is used to construct extreme weather features based on the meteorological data and preset extreme weather thresholds; The spatiotemporal lag data module is used to obtain spatiotemporal lag data of the epidemic based on the spatial and temporal autocorrelation relationships of the case data. Specifically, the spatiotemporal lag data module is used to: construct a spatial weight matrix based on the case data by calculating the spatial correlation index value between each grid cell and its neighboring areas, and obtain an inverse distance weight matrix by combining it with a preset inverse distance weighting radius; calculate the spatial lag data of each grid cell based on the inverse distance weight matrix and the case data; perform least squares regression on the case data to obtain residuals, and calculate the DW statistic based on the residuals; determine whether there is a first-order positive correlation in the case data based on the DW statistic, and if so, use the spatial lag data with a lag of 1 period as the spatiotemporal lag data of the epidemic. The model building module is used to add a spatiotemporal lag term of the epidemic to the zero-inflation negative binomial regression model to obtain an initial epidemic prediction model; the epidemic prediction model includes a quadratic term of the extreme weather characteristics; A parameter fitting module is used to perform parameter fitting on the initial epidemic prediction model based on the case data, the extreme weather characteristics, the spatiotemporal lag data of the epidemic, and the multi-source data to obtain a fitted epidemic prediction model. Specifically, the parameter fitting module is used to: use the extreme weather characteristics, the spatiotemporal lag data of the epidemic, and the multi-source data as a first variable set; construct a basic regression model for each type of extreme weather based on the first variable set and the case data; calculate the variance inflation factor for each extreme weather characteristic based on the basic regression model; filter variables in the first variable set based on the variance inflation factor to obtain a second variable set; add a penalty term to the basic regression model for each type of extreme weather based on the second variable set using the Lasso regression algorithm, and use stepwise regression to remove variables whose contribution rate to the model is less than a preset threshold to obtain a third variable set; and perform parameter fitting on the initial epidemic prediction model based on the third variable set and the case data to obtain a fitted epidemic prediction model. The epidemic prediction module is used to predict the future epidemic situation in the target area using the fitted epidemic prediction model.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein when the device containing the computer-readable storage medium executes the computer program, it implements the dengue fever prediction method considering extreme weather as described in any one of claims 1 to 6.

Images