Construction method of blood pressure meteorological risk dynamic evaluation index

Abstract

The invention discloses a construction method of a blood pressure meteorological risk dynamic evaluation index. The method comprises the following steps: using a mixing effect and introducing a cross basis function; selecting a random forest model to calculate a variation mean value, and dividing the variation mean value by the total variation to obtain a weight value; calculating an excess blood pressure value by adopting a mixed effect model, dividing the excess blood pressure value by the maximum excess blood pressure value, and multiplying by 10 to obtain a blood pressure meteorological index; a blood pressure meteorological index integrating various meteorological factors is constructed, risk levels are divided, risk early warning is carried out, decision buffer time is created for different subjects, people with hypertension and related managers are helped to actively cope with the influence of meteorological changes on blood pressure, and the adaptability of the people with hypertension to climate changes is improved; meanwhile, by screening meteorological factors sensitive to the blood pressure and evaluating the influence of composite exposure of the meteorological factors on the blood pressure, the method has more practical significance compared with existing single meteorological factor analysis, the interpretability of the blood pressure-meteorological association is improved, and therefore the method can also be popularized to other health outcomes caused by bad meteorological factors.

Description

Technical Field

[0001] This invention relates to the field of dynamic assessment methods for blood pressure meteorological risk in hypertensive populations, and more particularly to a blood pressure meteorological index that integrates blood pressure meteorological risk factor screening and risk assessment, as well as the exposure-response relationship between composite meteorological exposure and blood pressure. Background Technology

[0002] Hypertension is a major risk factor for cardiovascular death and has already caused a serious disease burden in my country. In the context of climate change, research evidence shows that meteorological factors are an even more important factor affecting blood pressure. Although there are currently health risk indices based on temperature and humidity, considering the complexity of real-world populations being exposed to multiple meteorological factors simultaneously, there is an urgent need to develop a dynamic assessment index for blood pressure health risk based on the coupling effect of multiple meteorological elements. This would facilitate risk warning and is also one of the important means to address the adverse health risks of climate change and improve the population's meteorological adaptability. Summary of the Invention

[0003] To address the aforementioned technical problems, this invention provides a method for constructing a dynamic assessment index of blood pressure meteorological risk. This method can create a decision-making buffer time for different stakeholders, help hypertensive individuals and relevant management parties proactively respond to the impact of meteorological changes on blood pressure, improve the adaptability of hypertensive individuals to climate change, and can also be extended to other health outcomes caused by adverse meteorological factors.

[0004] The technical solution of the present invention is as follows: A method for constructing a dynamic assessment index of blood pressure meteorological risk, comprising the following steps: A. Collect data, including at least meteorological information and air pollutant data; B. Organize the data and match the corresponding meteorological data according to the geographical location of the research subjects and the time of blood pressure measurement to generate an analysis dataset; C. Screen for blood pressure-sensitive meteorological factors, use a mixed-effects model to fit the association between each meteorological factor and blood pressure, apply data from multiple repeated measurements, and introduce cross-basis functions within this framework; D. Constructing the weighting of meteorological factors and the blood pressure meteorological index; The random forest model is selected to randomly sample the samples and their important features, calculate the mean variance caused by each variable, and divide it by the total variance of all variables to obtain the weight value of each variable. ;(Formula 2) ;(Formula 3) ;(Formula 4) In formulas 2~4, This indicates the importance of variable X in node M;N m This indicates the number of samples in the current node; N r and N l Indicates the number of samples in the left and right child nodes; GI r and GI l This represents the new node generated after splitting node m. Gini index; VI(X j ） Indicates the weight of the meteorological factors included; VI ij Representing variables X j In the i The importance of Gini in the trees; QI This refers to the comprehensive meteorological index; VI i This represents the weight of the i-th meteorological variable; X i This represents the actual value of the i-th meteorological variable; The excess blood pressure caused by combined exposure to multiple meteorological factors was calculated using a mixed-effects model. The excess blood pressure value was then divided by the maximum excess blood pressure value during the study period and multiplied by 10 to obtain the blood pressure meteorological index. The specific calculation formula is as follows: ;(Formula 5) ;(Formula 6) ;(Formula 7) In formulas 5~7, E(Y E(Y) or E(Y) both represent the expected value of systolic or diastolic blood pressure; α Indicates a fixed intercept term; β· QI This represents the linear influence term of the meteorological index QI, where β is its linear coefficient. cb Represents the cross basis function; lag Indicates the number of days of lag; df Indicates degrees of freedom; Cor Represent covariates; μ Represents the random intercept term; ε ExcessBP represents the residual term. t BP represents the excess blood pressure value at time t, which is the difference between the population blood pressure value and the fitted intercept blood pressure value under the combined exposure to multiple meteorological factors; min The fitted intercept blood pressure value represents the blood pressure meteorological index under combined exposure to meteorological factors; BPMI represents the blood pressure meteorological index; ExcessBP t_max This indicates the maximum excess blood pressure value during the study period; E. Defining and validating the risk threshold of the blood pressure meteorological index.

[0005] The method for constructing the dynamic assessment index of blood pressure meteorological risk, wherein: E. When classifying and verifying the risk threshold of the blood pressure meteorological index, the regression decision tree method is used to classify the risk threshold of the blood pressure meteorological index. The blood pressure meteorological index is used as the independent variable and the blood pressure value is used as the dependent variable to construct a regression decision number model to determine the risk threshold of the blood pressure meteorological index for elevated blood pressure.

[0006] The method for constructing the dynamic assessment index of blood pressure meteorological risk, wherein E, and the calculation formula for dividing and verifying the risk threshold of the blood pressure meteorological index, are as follows: ;(Formula 8) ;(Formula 9) ;(Formula 10) In formulas 8-10, M or m both represent elements, and R m C represents the region after the m-th partition; m Representing each region R m The corresponding output value or predicted value; I is the indicator function, when x belongs to region R. m The value is 1 if it is true, and 0 otherwise; y i Represents the true output value of the i-th sample; N represents the region R. m The number of samples within the region; c1 and c2 represent the true output values ​​y of all samples within the two sub-regions R1 and R2 after partitioning. i The mean of the values; A represents the attribute used for the division (here, the blood pressure meteorological index); s is the threshold for the division; x i Represents the identifier of the i-th sample; and region R m The predicted value C m It is the region R m The true output value y of all samples i The mean.

[0007] The method for constructing the dynamic assessment index of blood pressure meteorological risk, wherein: When constructing a regression decision number model, the original data is randomly divided into a 70% training set and a 30% validation set. The training set data is used for modeling, and the validation set data is used for validation.

[0008] The method for constructing the dynamic assessment index of blood pressure meteorological risk, wherein: The regression decision tree model takes minimizing the daily blood pressure mean square error (MSE) as its principle. It divides the blood pressure meteorological index into several intervals by using the internal node splitting threshold (i.e., sub-nodes). Each interval corresponds to a leaf node, and the predicted value output by the leaf node is the blood pressure mean value corresponding to each blood pressure meteorological index range.

[0009] The method for constructing the dynamic assessment index of blood pressure meteorological risk, wherein: C. When screening for blood pressure-sensitive meteorological factors, a mixed effect model is used, and a cross-basis function of a distributed lag nonlinear model is introduced to evaluate the nonlinear relationship and lag effect between meteorological factors and blood pressure.

[0010] The method for constructing the dynamic assessment index of blood pressure meteorological risk, wherein: C. When screening for meteorological factors sensitive to blood pressure, consider age, gender, smoking, alcohol consumption, body mass index, exercise, O3, and PM2.5. 2.5 Air pollutants were included as confounding factors in a composite exposure model for control, and meteorological factors that have a significant impact on blood pressure were screened out.

[0011] The method for constructing the dynamic assessment index of blood pressure meteorological risk includes the following calculation formula for the single meteorological factor exposure model: ;(Formula 1) In Formula 1, 𝑔 is the join function. E(Y) For systolic and diastolic blood pressure, Y ij For the first i The individual in the first j Response variables at the time of measurement; 𝑎 The intercept is... cb(X ij , lag, df) As a cross-base of meteorological factors, lag Based on the principle of minimum AIC, this study selected a lag of 2 days for the set lag number. df The set degrees of freedom; X ij Refers to meteorological factors; COVs ij For covariates; u i Individual random effects; ε ij This represents measurement error.

[0012] The method for constructing the dynamic assessment index of blood pressure meteorological risk, wherein: B. When organizing data, for missing values ​​in the data, depending on the situation of the missing data, use the mean or median to fill in the missing values ​​for continuous variables, and use the mode to fill in the missing values ​​for categorical variables, or use multiple imputation methods to imput the missing values.

[0013] The method for constructing the dynamic assessment index of blood pressure meteorological risk, wherein: B. When organizing data, outliers that appear during data collection and processing should be handled by replacement or transformation; duplicate data that appears during data collection and processing should be removed.

[0014] The present invention provides a method for constructing a dynamic assessment index of blood pressure meteorological risk. By constructing a blood pressure meteorological index that integrates multiple meteorological factors, classifying risk levels and conducting risk warnings, it creates a decision-making buffer time for different stakeholders, helps hypertensive individuals and relevant management parties to proactively respond to the impact of meteorological changes on blood pressure, and improves the adaptability of hypertensive individuals to climate change. At the same time, by screening meteorological factors that are sensitive to blood pressure and assessing the impact of their combined exposure on blood pressure, it is more practical than existing single meteorological factor analysis, improves the interpretability of blood pressure-meteorological association, and can thus be extended to other health outcomes caused by adverse meteorological factors. Attached Figure Description

[0015] The accompanying drawings described herein are for illustrative purposes only and are not intended to limit the scope of the invention in any way. Figure 1 This is a technical roadmap for the blood pressure meteorological index in the method for constructing the dynamic assessment index of blood pressure meteorological risk of the present invention. Figure 2 The systolic pressure used in Embodiment 1 of the present invention (Table 1); Figure 3 The diastolic blood pressure used in Embodiment 1 of the present invention is shown in Table 2. Figure 4 This refers to Table 3 (risk threshold) used in Embodiment 1 of the present invention. Figure 5 This refers to Table 4 (recommended measures) used in Embodiment 1 of the present invention. Detailed Implementation

[0016] The specific embodiments and examples of the present invention will be described in detail below with reference to the accompanying drawings. The specific embodiments described are only used to explain the present invention and are not intended to limit the specific embodiments of the present invention.

[0017] This invention develops a dynamic assessment index for blood pressure meteorological risk based on the exposure-response relationship between combined exposure to multiple meteorological factors and blood pressure, such as... Figure 1 As shown, it includes the following steps: Step A, Data Collection: Conduct follow-up studies, collect basic information, behavioral characteristics, health status and physical activity levels of the population through questionnaires, and measure residents' blood pressure in all four quarters of the year using standard electronic / mercury sphygmomanometers; at the same time, collect meteorological information and air pollutant data, and match them according to the latitude and longitude of the residents' location. Step B, Data Processing: The collected resident questionnaire data and meteorological data were preprocessed to remove invalid information, correct biased data, and integrate duplicate records, ultimately forming a dataset for modeling. Step C: Screening of Blood Pressure-Sensitive Meteorological Factors: A mixed-effects model was used, incorporating a cross-basis function from a distributed lag nonlinear model, to evaluate the nonlinear relationship and lag effect between meteorological factors and blood pressure. Simultaneously, age, gender, smoking, alcohol consumption, body mass index, exercise, O3, and PM2.5 were also considered. 2.5 Air pollutants and other factors were included as confounding factors for control, and meteorological factors that have a significant impact on blood pressure were screened out. Step D: Construction of meteorological factor effect weights and blood pressure meteorological index: Random Forest (RF) model is selected to randomly sample the samples and their important features, calculate the mean variation caused by each variable, and divide it by the total variation of all variables to obtain the weight value of each variable; then these weight values ​​are incorporated into the mixed-effects model and refitted with the blood pressure values ​​to calculate the excess blood pressure value; then each excess blood pressure value is divided by the maximum excess blood pressure value in the study to obtain the blood pressure meteorological index. Step E, Threshold Classification and Validation of Blood Pressure Meteorological Index: Construct a regression decision tree model, divide the nodes of the blood pressure meteorological index into thresholds corresponding to the risk levels, and use the validation set to verify the threshold classification effect.

[0018] In a specific embodiment of the method for constructing the dynamic assessment index of blood pressure meteorological risk of the present invention, in step A, the data collection stage, the following two aspects are emphasized: 1) Basic characteristics of the research subjects Data collection may be derived from cohort studies. The survey (or study) subjects are defined as Chinese residents aged 18 years and older, and data on the subjects are collected multiple times in different seasons. The information collected includes: basic information of the subjects (including gender, age, etc.), behavioral characteristics (including whether they smoke, whether they have consumed alcohol in the past 12 months, etc.), health status (medical history, physical examination, etc.), and physical activity level. Based on the subjects' physical activity level in the past week, all populations are divided into three levels: high, medium, and low. Blood type, blood pressure, height, weight, and other relevant health information are collected during the physical examination. 2) Meteorological information and air pollutant data Meteorological information data can be selected from the fifth-generation global atmospheric reanalysis data of the European Centre for Medium-Range Weather Forecasts (ECMWF), including data such as daily average temperature, relative humidity, wind speed, rainfall, air pressure and precipitation; Air pollutant data can be obtained from the China National Air Quality Real-Time Release Platform (http: / / 106.37.208.233:20035 / ). For example, daily average concentration data of air pollutants, including O3 and PM2.5, from all atmospheric environmental monitoring stations between 2013 and 2015 can be collected. 2.5 Data on air pollutants, etc.

[0019] Specifically, in step B, the data preparation stage, the following three points should be considered: 1> Handling Missing Values During the collection of environmental meteorological data, socioeconomic data, and personal health cohort data, missing values ​​may exist. Depending on the type of missing data, continuous variables are imputed using the mean or median; categorical variables are imputed using the mode, or multiple imputation methods are used to impute missing values. 2> Handling outliers and duplicate values Outliers (such as extreme values) that occur during data collection and acquisition can be handled by replacement or transformation; duplicate data that occurs during data collection and acquisition can be removed. 3> Generate dataset The blood pressure data and other information obtained from the questionnaire survey were integrated, and corresponding meteorological data were matched according to the geographical location of the research subjects and the time of blood pressure measurement to generate an analysis dataset.

[0020] Specifically, in step C, the screening stage for blood pressure-sensitive meteorological factors, the following two aspects are mainly considered: (1) Effects of hysteresis and nonlinearity Since the effects of meteorological factors on health often have a certain degree of persistence and lag, and the effects of meteorological factors on blood pressure usually reach a peak in a short period of time and then rapidly weaken, and this effect is not a linear relationship (for example, the effect of rainfall on blood pressure shows a U-shaped curve), a mixed-effects model can be used to fit the relationship between various meteorological factors and blood pressure. By applying data from multiple repeated measurements and introducing cross-basis functions within this framework, the nonlinear effects of meteorological factors on blood pressure at different lag times can be effectively captured, thereby improving the model's fitting effect and interpretability. (2) Control of confounding factors In association analysis, factors such as age, gender, smoking, alcohol consumption, body mass index, exercise, and levels of O3 and PM2.5 in the study participants were considered. 2.5 These factors have been shown in previous literature reviews to be potentially associated with blood pressure; therefore, they can be included as confounding variables for control. The calculation formula for the single meteorological factor exposure model is the same: ;(Formula 1) In Formula 1, 𝑔 is the join function. E(Y) For systolic and diastolic blood pressure, Y ij For the first i The individual in the first j Response variables at the time of measurement; 𝑎 The intercept is... cb(X ij , lag, df) A cross-base of meteorological factors; lag Based on the principle of minimum AIC, this study selected a lag of 2 days for the set lag number. df The set degrees of freedom; X ij Meteorological factors (such as temperature, humidity, air pressure, etc.); COVs ij For covariates (PM) 2.5 (O3, baseline blood pressure, age, etc.); u i Individual random effects; ε ij This represents measurement error.

[0021] Specifically, in step D, the construction stage of meteorological factor effect weights and blood pressure meteorological index, the key lies in the following two models: 1. Random Forest Model A random forest model is used to randomly sample samples and their important features. The mean of the variance generated by each sampling is calculated to characterize the degree of variation caused by that variable. The weight of each variable is then obtained by dividing the mean variance of each variable by the cumulative total variance of all variables. The specific calculation formula is as follows: ;(Formula 2) ;(Formula 3) ;(Formula 4) In formulas 2~4, This indicates the importance of variable X in node M; N m This indicates the number of samples in the current node; N r and Nl Indicates the number of samples in the left and right child nodes; GI r and GI l This represents the new node generated after splitting node m. Gini index; VI(X j ） Indicates the weight of the meteorological factors included; VI ij Representing variables X j In the i The importance of Gini in the trees; QI This refers to the comprehensive meteorological index; VI i This represents the weight of the i-th meteorological variable; X i This represents the actual value of the i-th meteorological variable.

[0022] 2. Mixed-effects model The excess blood pressure caused by combined exposure to multiple meteorological factors was calculated using a mixed-effects model. This excess blood pressure value was then divided by the maximum excess blood pressure value during the study period and multiplied by 10 to obtain a blood pressure meteorological index varying within the range of 0-10. The specific calculation formula is as follows: ;(Formula 5) ;(Formula 6) ;(Formula 7) In formulas 5~7, E(Y E(Y) or E(Y) both represent the expected value of systolic or diastolic blood pressure; α Indicates a fixed intercept term; β· QI This represents the linear influence term of the meteorological index QI, where β is its linear coefficient. cb Represents the cross basis function; lag Indicates the number of days of lag; df Indicates degrees of freedom; Cor Represent covariates; μ Represents the random intercept term; ε ExcessBP represents the residual term. t BP represents the excess blood pressure value at time t, which is the difference between the population blood pressure value and the fitted intercept blood pressure value under the combined exposure to multiple meteorological factors; min The fitted intercept blood pressure value represents the blood pressure meteorological index under combined exposure to meteorological factors; BPMI represents the blood pressure meteorological index; ExcessBP t_max This indicates the maximum excess blood pressure value during the study period.

[0023] Specifically, in step E, the risk threshold classification and validation stage of the blood pressure meteorological index, the original data is randomly divided into a 70% training set and a 30% validation set. The training set is used to build a model, and the regression decision tree method is used to classify the risk threshold of the blood pressure meteorological index. A regression decision tree model is constructed with the blood pressure meteorological index as the independent variable and the blood pressure value as the dependent variable to determine the risk threshold of the blood pressure meteorological index for elevated blood pressure. The calculation formula is as follows: ;(Formula 8) ;(Formula 9) ;(Formula 10) In formulas 8-10, M or m both represent elements, and R m C represents the region after the m-th partition; m Representing each region R m The corresponding output value or predicted value; I is the indicator function, when x belongs to region R. m The value is 1 if it is true, and 0 otherwise; y i Represents the true output value of the i-th sample; N represents the region R. m The number of samples within the region; c1 and c2 represent the true output values ​​y of all samples within the two sub-regions R1 and R2 after partitioning. i The mean of the values; A represents the attribute used for the division (here, the blood pressure meteorological index); s is the threshold for the division; x i Represents the identifier of the i-th sample; and region R m The predicted value C m It is the region R m The true output value y of all samples i The mean.

[0024] Specifically, the regression decision tree model uses the principle of minimizing the daily blood pressure mean square error (MSE) to divide the blood pressure meteorological index into several intervals through the division threshold (sub-node) of the internal nodes. Each interval corresponds to a leaf node. The predicted value output by the leaf node is the blood pressure mean value corresponding to each blood pressure meteorological index range. A larger output predicted value means that the blood pressure meteorological index range has a greater effect on raising blood pressure, i.e., a higher risk.

[0025] Specifically, to test the predictive and discriminative ability of the blood pressure meteorological index and its risk classification for blood pressure risk, a validation set obtained from a randomly selected 30% of the population was used for validation. The results were compared to see if there were differences in blood pressure under different risk levels. If there were no differences between the validation set and the training set, it indicated that the regression decision number model had good generalization performance, did not overfit to the specific noise of the training set, and could stably adapt to unknown data. If there were differences, it indicated that the regression decision number model was overfitting, which could be corrected by reducing model complexity (pruning), adjusting parameters, and optimizing data. Example

[0026] Embodiment 1 of the present invention is an embodiment of constructing a blood pressure meteorological index provided by the present invention, which includes: (a) Collect the required population data and meteorological data Data collection for the population was conducted using a combination of questionnaires and on-site physical examinations, totaling 10,000 responses. The questionnaires included the following: basic demographic characteristics, health status, and dietary habits of the participants. Basic demographic characteristics included age, sex (male, female), ethnicity (Han, other), marital status (married, unmarried), and education level (junior high school and below, high school, university and above). Health status included smoking (yes / no), alcohol consumption (yes / no), history of hypertension and other diseases (yes / no), physical activity level (divided into exercise and non-exercise groups based on whether moderate-to-vigorous physical activity, such as running or swimming, had been performed in the past week), and resting time (quantified by the cumulative hours spent in sitting, leaning, or lying down, or other non-sleep static behaviors while awake each day). Dietary habits included the intake of grains, fruits, and meat. Physical examinations included measurements of height, weight, and blood pressure.

[0027] Meteorological data were obtained from the fifth-generation global atmospheric reanalysis data (ERA5) provided by the European Centre for Medium-Range Weather Forecasts (ECMWF), extracting daily average temperature, relative humidity, wind speed, air pressure, and precipitation. Atmospheric pollutant data were mainly controlled as confounding factors, and daily average concentrations of ozone (O3) and fine particulate matter (PM2.5) were obtained through the national real-time air quality release platform (http: / / 106.37.208.233:20035 / ). All environmental data were matched according to the latitude and longitude of the district / county where the study subjects were located and the survey date.

[0028] (II) Screening and determining characteristic variables of blood pressure meteorological index based on statistical methods In practice, one can first search for meteorological factors related to blood pressure in the authoritative medical journals defined in the Chinese Science and Technology Journal Citation Report, and identify some meteorological factors (i.e. variables) that may affect blood pressure, including: temperature, humidity, wind speed, precipitation, air pressure, etc.

[0029] The preferred model should be a mixed-effects model, but given that the example in this invention uses cross-sectional data, the model is simplified to a distributed lag nonlinear model. Based on this model, the process of screening and determining the meteorological variables for the blood pressure meteorological index includes: determining the meteorological variables that affect blood pressure fluctuations, constructing lag terms, incorporating them into the model to capture the nonlinear effects of meteorological factors on blood pressure at different lag periods, obtaining the P-values ​​of each variable, and selecting variables with P-values ​​less than a set threshold as the variables required for the index.

[0030] The purpose of the screening was to eliminate variables with poor efficacy from various meteorological variables and select strongly correlated variables as the basis for subsequent index construction. Statistical methods were used to screen meteorological variables, constructing lag terms and incorporating them into a distributed lag nonlinear model to capture the nonlinear effects of meteorological factors on blood pressure at different lag stages, obtaining the p-values ​​for individual variables. Based on the statistical significance of a p-value less than 0.05, factors with minor effects on blood pressure were excluded. The final screened meteorological variables included: temperature, relative humidity, precipitation, air pressure, and wind speed. All five selected meteorological variables met the screening criteria (p < 0.05) in the univariate analysis. Relative humidity was calculated using air temperature and dew point temperature, and the model formula is as follows: (Partial Formula 1) In the above formula, 𝑔 is the join function; E(Y) These are systolic and diastolic blood pressure. Y For individuals, this is the response variable during measurement; The intercept; The cross-base for meteorological factors; lag is the set lag number of days. According to the principle of minimum AIC, the lag number of days is selected as 2. Since the lag effect of meteorological factors on blood pressure is not lasting, the number of days can be adjusted between 0 and 2. df The set degrees of freedom; X i Meteorological factors (such as temperature, humidity, air pressure, etc.); COVs Covariates include PM2.5, O3, baseline blood pressure, and age.

[0031] In this association analysis, age, gender, smoking, alcohol consumption, body mass index, exercise, O3, and PM2.5 were included as confounding variables for control because they have been shown to be potentially associated with blood pressure. Taking temperature as an example, in this case, g[E(systolic blood pressure)] = a + cb(temperature, 2, 3) + β 1⋅PM2.5+ β 2⋅O3+ β 3⋅Baseline blood pressure + β 4⋅Age+ β 5⋅ Gender + β 6⋅BMI+ β 7⋅Smoking+β 8⋅ Drinking + β 9. Exercise.

[0032] This study included 10,000 participants. The average temperature, wind speed, air pressure, relative humidity, rainfall, O3, and PM2.5 were 19.70℃, 4.07 m / s, 100.22 kPa, 63.68%, 3.16 mm, 42.22 μg / m3, and 30.54 μg / m3, respectively. Univariate results showed that the average temperature, wind speed, air pressure, relative humidity, and rainfall on the day of the survey all had significant effects on blood pressure. Daily average temperature and relative humidity showed negative correlations with both systolic and diastolic blood pressure. Daily average wind speed, air pressure, and blood pressure showed positive correlations, while rainfall and blood pressure exhibited a U-shaped relationship. The correlations between meteorological factors and blood pressure were generally consistent from two days prior to the survey to the day of the survey.

[0033] To further construct a blood pressure meteorological index, a random forest model was used to evaluate the effect weights of five meteorological factors on blood pressure, in order to reveal the heterogeneous contribution characteristics of meteorological factors to blood pressure. The specific calculation formula is as follows: ;(Formula 2) ;(Formula 3) ;(Formula 4) In the above formula, This indicates the importance of variable X in node M; N m This indicates the number of samples in the current node; N r and N l Indicates the number of samples in the left and right child nodes; GI r and GI l This represents the new node generated after splitting node m. Gini index; VI(X j ） Indicates the weight of the meteorological factors included; VI ij Representing variables X j In the i The importance of Gini in the trees; QI This refers to the comprehensive meteorological index; VI i This represents the weight of the i-th meteorological variable; X i This represents the actual value of the i-th meteorological variable.

[0034] The results obtained in this example show that the weights of importance scores calculated by the Random Forest Gini index are as follows: On the day of the survey, the weights of the effects of temperature, humidity, air pressure, wind speed, and precipitation on systolic blood pressure were 21.5%, 20.6%, 18.3%, 18.4%, and 21.2%, respectively; and the weights of the effects on diastolic blood pressure were 21.3%, 19.3%, 21.6%, 18.7%, and 19.2%, respectively; and the results were consistent between the day before the survey and the two days before the survey.

[0035] Then, based on the above effect weights, a blood pressure meteorological index was constructed. A distributed lag nonlinear model was used to calculate the excess blood pressure caused by combined exposure to five meteorological factors. The excess blood pressure value was then divided by the maximum excess blood pressure value during the study period and multiplied by 10 to obtain a blood pressure meteorological index varying within the range of 0-10. The specific calculation formula is as follows: (Partial Formula 5) ;(Formula 6) ;(Formula 7) In the above formula, E(Y E(Y) represents the expected value of systolic or diastolic blood pressure; α represents the fixed intercept term; cb(AQI,lag,df) is the standardized cross-basis function multiplied by the weights of the research variables; AQI refers to the meteorological composite index; cb represents the cross-basis function; lag represents the lag days; df represents the degrees of freedom; Cor represents the covariates; ExcessBP t All indicate the first t Excess blood pressure at any given time, i.e., the difference between the blood pressure value of a population under the combined exposure to multiple meteorological factors and the fitted intercept blood pressure value; BP min The fitted intercept blood pressure value represents the blood pressure meteorological index under combined exposure to meteorological factors; BPMI represents the blood pressure meteorological index; ExcessBP t_max This indicates the maximum excess blood pressure value during the study period.

[0036] In this example, with a cumulative lag of 0-2 days, the average systolic blood pressure meteorological index is 5.9 with a standard deviation of 1.5; the average diastolic blood pressure meteorological index is 7.4 with a standard deviation of 1.0; and as the corresponding indices increase, the changes in systolic and diastolic blood pressure also increase.

[0037] Because the obtained index is between 0 and 10, it is not convenient to intuitively determine the risk level. Therefore, in the stage of risk threshold classification and verification of the blood pressure meteorological index, the original data was randomly divided into a 70% training set and a 30% validation set. The training set was used to build a model, and a regression decision tree method was used to classify the risk threshold of the blood pressure meteorological index. A regression decision tree model was constructed with the blood pressure meteorological index as the independent variable and the blood pressure value as the dependent variable to determine the risk threshold of the blood pressure meteorological index for elevated blood pressure. The calculation formula is as follows: ;(Formula 8) ;(Formula 9) ;(Formula 10) In the above formula, M or m both represent elements, and R m C represents the region after the m-th partition; m Representing each region R m The corresponding output value or predicted value; I is the indicator function, when x belongs to region R. m The value is 1 if it is true, and 0 otherwise; y i Represents the true output value of the i-th sample; N represents the region R. m The number of samples within the region; c1 and c2 represent the true output values ​​y of all samples within the two sub-regions R1 and R2 after partitioning. i The mean of the values; A represents the attribute used for the division (here, the blood pressure meteorological index); s is the threshold for the division; x i Represents the identifier of the i-th sample; and region R m The predicted value C m It is the region R m The true output value y of all samples i The mean.

[0038] Based on the risk warning decision tree, we divide the blood pressure meteorological index into three risk levels: low risk, medium risk, and high risk. The results are as follows: Figure 2 The systolic blood pressure shown in Table 1 and Figure 3 The diastolic blood pressure is shown in Table 2.

[0039] Finally, based on the risk threshold obtained from 70% of the training set, validation was performed on the 30% validation set; such as Figure 4 Table 3 shows the results. In the validation set, with increasing blood pressure meteorological risk level, the levels of systolic and diastolic blood pressure significantly increased, and the risk of developing hypertension also significantly increased. Pairwise difference tests showed no statistically significant differences. The results indicate that this threshold has a good effect. Furthermore, based on the above threshold levels, corresponding recommendations can be proposed, such as... Figure 5 Table 4 is shown.

[0040] To further verify the function of this index, the present invention provides another embodiment 2, which assumes that the meteorological data of the city from which the above data is sourced on a future day is as follows: average daily temperature of 28°C, average daily relative humidity of 70%, average daily air pressure of 100 kPa, average daily wind speed of 3 m / s, and daily precipitation of 5 mm, and predicts the risk of systolic pressure fluctuation on that day based on this index.

[0041] Multiplying the corresponding quantities by the effect weights of systolic pressure, the values ​​are: temperature 28 × 21.5% = 6.02, relative humidity 70 × 20.6% = 14.42, air pressure 100 × 18.3% = 18.3, wind speed 3 × 18.4% = 0.552, and precipitation 5 × 21.2% = 1.06, totaling 40.352. Substituting these values ​​into the model, the predicted systolic pressure is approximately 132.7812. BPmin Approximately 125.4979, ExcessBP t Approximately 7.2833, while ExcessBP t_max The reading was approximately 9.3826, resulting in a blood pressure meteorological index of approximately 7.76. The risk level was assessed as Level 3 (high risk). Residents of the city were advised to minimize outdoor activities and exercise intensity, keep doors and windows closed, use air conditioning, and regulate the indoor microclimate. Sensitive individuals were advised to avoid strenuous outdoor activity and prolonged exposure to the outdoors.

[0042] Based on the above-mentioned method for constructing a dynamic assessment index of blood pressure meteorological risk, this invention proposes a system for constructing a dynamic assessment index of blood pressure meteorological risk, including a data collection unit, a data processing unit, a blood pressure-sensitive meteorological factor screening unit, a meteorological factor effect weight and blood pressure meteorological index construction unit, and a blood pressure meteorological index risk threshold division and verification unit, which are respectively used to correspond to steps A to E above.

[0043] Based on the above-mentioned method for constructing the dynamic assessment index of blood pressure meteorological risk, this invention also proposes an electronic device, including a memory and a processor, including but not limited to smartphones, mobile tablets, laptops, desktop devices, etc. When the processor executes a computer management program stored in the memory, it implements the above steps A to E.

[0044] Based on the above-mentioned method for constructing the dynamic assessment index of blood pressure meteorological risk, this invention also proposes a computer-readable storage medium, including but not limited to storage chips, mobile hard drives, USB flash drives, optical discs, etc., which stores a computer management program. When the computer management program is executed by the processor of an electronic device, it implements the above-mentioned steps A to E.

[0045] It should be noted that this invention includes processes such as data collection, data processing, determining the effect weight of each meteorological factor, establishing a blood pressure meteorological risk index, classifying risk levels and verifying them. It establishes a dynamic assessment index for blood pressure meteorological risk based on the coupling effect of multiple meteorological elements. It does not achieve simple functions with a complex structure, nor does it use conventional or simple features for combination or stacking. Therefore, it conforms to the common sense of technological improvement and has practicality.

[0046] Any content not described in detail in this specification is prior art known to those skilled in the art.

[0047] It should be understood that the above description is only a preferred embodiment of the present invention and is not sufficient to limit the technical solution of the present invention. For those skilled in the art, within the spirit and principles of the present invention, additions, subtractions, substitutions, transformations or improvements can be made based on the above description, and all such additions, subtractions, substitutions or improvements should fall within the protection scope of the appended claims of the present invention.

Claims

1. A method for constructing a dynamic assessment index for blood pressure meteorological risk, characterized in that, Includes the following steps: A. Collect data, including at least meteorological information and air pollutant data; B. Organize the data and match the corresponding meteorological data according to the geographical location of the research subjects and the time of blood pressure measurement to generate an analysis dataset; C. Screen for blood pressure-sensitive meteorological factors, use a mixed-effects model to fit the association between each meteorological factor and blood pressure, apply data from multiple repeated measurements, and introduce cross-basis functions within this framework; D. Constructing the weighting of meteorological factors and the blood pressure meteorological index; The random forest model is selected to randomly sample the samples and their important features, calculate the mean variance caused by each variable, and divide it by the total variance of all variables to obtain the weight value of each variable. ； (Formula 2) ； (Formula 3) ； (Formula 4) In formulas 2~4, This indicates the importance of variable X in node M; N m This indicates the number of samples in the current node; N r and N l Indicates the number of samples in the left and right child nodes; GI r and GI l This represents the new node generated after splitting node m. Gini index; VI (X) j ） Indicates the weight of the meteorological factors included; VI ij Representing variables X j In the i The importance of Gini in the trees; QI This refers to the comprehensive meteorological index; VI i This represents the weight of the i-th meteorological variable; X i This represents the actual value of the i-th meteorological variable; The excess blood pressure caused by combined exposure to multiple meteorological factors was calculated using a mixed-effects model. The excess blood pressure value was then divided by the maximum excess blood pressure value during the study period and multiplied by 10 to obtain the blood pressure meteorological index. The specific calculation formula is as follows: ； (Formula 5) ； (Formula 6) ； (Formula 7) In formulas 5~7, E(Y E(Y) or E(Y) both represent the expected value of systolic or diastolic blood pressure; α Indicates a fixed intercept term; β·QI This represents the linear influence term of the meteorological index QI, where β is its linear coefficient. cb Represents the cross basis function; lag Indicates the number of days of lag; df Indicates degrees of freedom; Cor Represent covariates; μ Represents the random intercept term; ε ExcessBP represents the residual term. t BP represents the excess blood pressure value at time t, which is the difference between the population blood pressure value and the fitted intercept blood pressure value under the combined exposure to multiple meteorological factors; min The fitted intercept blood pressure value represents the blood pressure meteorological index under combined exposure to meteorological factors; BPMI represents the blood pressure meteorological index; ExcessBP t_max This indicates the maximum excess blood pressure value during the study period; E. Defining and validating the risk threshold of the blood pressure meteorological index.

2. The method for constructing the dynamic assessment index of blood pressure meteorological risk according to claim 1, characterized in that: E. When classifying and verifying the risk threshold of the blood pressure meteorological index, the regression decision tree method is used to classify the risk threshold of the blood pressure meteorological index. The blood pressure meteorological index is used as the independent variable and the blood pressure value is used as the dependent variable to construct a regression decision number model to determine the risk threshold of the blood pressure meteorological index for elevated blood pressure.

3. The method for constructing the dynamic assessment index of blood pressure meteorological risk according to claim 2, characterized in that, E. The calculation formula for classifying and verifying the risk threshold of the blood pressure meteorological index is as follows: ； (Formula 8) ； (Formula 9) ； (Formula 10) In formulas 8-10, M or m both represent elements, and R m C represents the region after the m-th partition; m Representing each region R m The corresponding output value or predicted value; I is the indicator function, when x belongs to region R. m The value is 1 if it is true, and 0 otherwise; y i Represents the true output value of the i-th sample; N represents the region R. m The number of samples within the region; c1 and c2 represent the true output values ​​y of all samples within the two sub-regions R1 and R2 after partitioning. i The mean of the values; A represents the attribute used for the division (here, the blood pressure meteorological index); s is the threshold for the division; x i Represents the identifier of the i-th sample; and region R m The predicted value C m It is the region R m The true output value y of all samples i The mean.

4. The method for constructing the dynamic assessment index of blood pressure meteorological risk according to claim 2, characterized in that: When constructing a regression decision number model, the original data is randomly divided into a 70% training set and a 30% validation set. The training set data is used for modeling, and the validation set data is used for validation.

5. The method for constructing the dynamic assessment index of blood pressure meteorological risk according to claim 2, characterized in that: The regression decision tree model takes minimizing the daily blood pressure mean square error (MSE) as its principle. It divides the blood pressure meteorological index into several intervals by using the internal node splitting threshold (i.e., sub-nodes). Each interval corresponds to a leaf node, and the predicted value output by the leaf node is the blood pressure mean value corresponding to each blood pressure meteorological index range.

6. The method for constructing the dynamic assessment index of blood pressure meteorological risk according to claim 1, characterized in that: C. When screening for blood pressure-sensitive meteorological factors, a mixed effect model is used, and a cross-basis function of a distributed lag nonlinear model is introduced to evaluate the nonlinear relationship and lag effect between meteorological factors and blood pressure.

7. The method for constructing the dynamic assessment index of blood pressure meteorological risk according to claim 1, characterized in that: C. When screening for meteorological factors sensitive to blood pressure, consider age, gender, smoking, alcohol consumption, body mass index, exercise, O3, and PM2.

5. 2.5 Air pollutants were included as confounding factors in a composite exposure model for control, and meteorological factors that have a significant impact on blood pressure were screened out.

8. The method for constructing the dynamic assessment index of blood pressure meteorological risk according to claim 7, characterized in that, The calculation formula for the single meteorological factor exposure model is as follows: ； (Formula 1) In Formula 1, 𝑔 is the join function. E(Y) For systolic and diastolic blood pressure, Y ij For the first i The individual in the first j Response variables at the time of measurement; 𝑎 The intercept is... cb(X ij ,lag,df) As a cross-base of meteorological factors, lag Based on the principle of minimum AIC, this study selected a lag of 2 days for the set lag number. df The set degrees of freedom; X ij Refers to meteorological factors; COVs ij For covariates; u i Individual random effects; ε ij This represents measurement error.

9. The method for constructing the dynamic assessment index of blood pressure meteorological risk according to claim 1, characterized in that: B. When organizing data, for missing values ​​in the data, depending on the situation of the missing data, use the mean or median to fill in the missing values ​​for continuous variables, and use the mode to fill in the missing values ​​for categorical variables, or use multiple imputation methods to imput the missing values.

10. The method for constructing the dynamic assessment index of blood pressure meteorological risk according to claim 1, characterized in that: B. When organizing data, outliers that appear during data collection and processing should be handled by replacement or transformation; duplicate data that appears during data collection and processing should be removed.

Images