Environmental exposure data processing and health effect evaluation method, product and application

Abstract

The invention discloses a method and a product for environmental exposure data processing and health effect assessment and application, and belongs to the technical field of air quality health risk assessment. The technical problem to be solved is to carry out parametric fitting on non-linear health effect characteristics of common environmental pollution exposure so as to realize accurate, flexible and operable non-linear change analysis and application. The method comprises the following steps: S1, data processing; and S2, health effect evaluation. According to the method, a shape control function containing a position parameter, a size parameter and a shape parameter is introduced into a shape constraint health effect function, fitting curves which have different characteristics and represent different shapes are successfully constructed, analysis of different nonlinear effect changes is achieved, health effect evaluation is more visual, and the method is more suitable for popularization and application. And the accuracy, flexibility and operability of environmental exposure health effect evaluation are improved.

Description

Technical Field

[0001] This invention relates to the field of air quality health risk assessment technology, specifically to methods, products, and applications for environmental exposure data processing and health effect assessment. Background Technology

[0002] For understanding the technical content of this invention: Air pollution has become one of the major public health problems facing humanity. PM2.5 is a key atmospheric pollutant. 2.5 Fine particulate matter (particulate matter with a particle size of less than 2.5 μm, PM) is a representative example. 2.5 Chronic disease burden and all-cause mortality are major causes of the worsening of the population's chronic disease burden and the frequent occurrence of all-cause mortality events. Even though air pollution has been somewhat controlled, its long-term exposure and cumulative health effects persist, attributable to different air pollutants (PM2.5). 2.5 PM 10 The disease burden from exposure to pollutants such as ozone, nitrogen oxides, and VOCs remains substantial. Therefore, dynamically measuring and monitoring the spatiotemporal heterogeneity of the disease burden caused by air pollution is crucial for optimizing health strategies. Life expectancy and healthy life expectancy, due to their ability to comprehensively reflect mortality levels and healthy lifespan across all age groups of the population, can serve as important indicators for assessing changes and trends in the attributable disease burden of different air pollutants in a region. It is necessary to start with comparative risk assessment techniques and improve them to accurately assess the life expectancy loss caused by long-term exposure to different air pollutants.

[0003] Relevant patent documents retrieved: The document, published in China (publication number CN119763829A) on April 4, 2025, discloses a method for assessing health indices under combined exposure to multiple pollutants. The method includes the following steps: S1: Collecting data on air pollution, meteorology, hospital admissions, and causes of death; obtaining daily values ​​of pollutants, meteorological factors, hospital admissions, and deaths; and obtaining pollutant 0... S2: Select statistically significant pollutants for multi-pollutant composite exposure analysis; S3: Combine daily hospitalization and mortality data, daily average values ​​of meteorological factors, and selected pollutants. S4: Calculate the excess risk corresponding to the 2-day moving average concentration by substituting it into the multi-pollutant model, the Bayesian weighted model, and the mixture analysis model; S5: Calculate the undetermined CEHI under different air pollutant combined exposures corresponding to the excess risk under different models; S6: Select a reasonable undetermined CEHI as the official CEHI.

[0004] Relevant non-patent literature retrieved: Journal name: 《npj climate and atmosphere science》, literature name: 《Health impact assessment on life expectancy gains ascribed to particulate matter reduction》, publication date: February 2025. The literature discloses the following content.

[0005] 为解决这些重要的知识空白,开展了比较风险评估研究,旨在:(i)研究中国人群中广泛的PM水平与死亡率之间的浓度-反应关系的形状,同时纳入世界卫生组织的最新阈值;(ii)使用预期寿命(LE)量化PM减少对人群长寿的贡献程度。我们提出了一个新颖的综合框架,该框架结合了时变Cox回归、形状函数和生命表方法,以激励相关利益攸关方将PM-死亡率关联的形状纳入政策分析和延长寿命的考量中。 (Translation: To fill these important knowledge gaps, this comparative risk assessment study aims to: (i) examine the concentration-response relationship between different PM concentration levels and mortality rates in the Chinese population, using the latest WHO thresholds; and (ii) quantify the contribution of PM concentration reduction to life expectancy using life tables. We propose an innovative and comprehensive framework that integrates time-varying Cox regression, shape functions, and life table methods, aiming to incentivize stakeholders to incorporate the relationship between PM and mortality rates into policy analysis and life extension planning.) The prior art represented by the aforementioned documents has at least the following unresolved technical problems or defects: Current assessment of pollutants (in terms of PM2.5) 2.5 and PM 10 For example, conventional techniques for assessing life expectancy loss often utilize direct regression analysis, such as multilevel linear regression models and panel data models. Direct regression analysis can only reveal the explicit quantitative dependence between exposure and outcome, but cannot distinguish between correlation and causation. Only through causal-oriented research designs such as cohort studies can the temporal relationship between exposure and outcome be clearly defined, thereby reducing reverse causation and confounding bias, and achieving a more accurate estimate of causal effects. Secondly, existing techniques have technical defects in terms of the flexibility and operability of dose-response curve fitting. Relevant evidence includes: existing conventional techniques such as the article "Air pollution trajectories and lifeexpectancy in Nigeria" (International Journal of Social Economics (2022) 49 (7): 1049–1070) and "The impacts of reduction in ambient fine particulate (PM2.5)" (The impacts of reduction in ambient fine particulate (PM2.5)). 2.5 In the article "Air pollution on life expectancy in Taiwan" (Journal of Toxicology and Environmental Health, Part A, 85(22), 913–920), linear regression and panel models were used to analyze the relationship. However, the existing techniques cannot fit the fitting effect characteristics of more different nonlinear changes. Summary of the Invention

[0006] The purpose of this invention is to provide: A method for processing environmental exposure data and assessing health effects, and related technologies, to address the technical problems of existing technologies, such as the inability to characterize the different variations of more nonlinear effects with flexible parameters, the difficulty in intuitively and accurately assessing health effects, and the poor operability of health effect assessment, or a combination thereof.

[0007] Terminology Explanation: Unless otherwise defined, all technical terms in this document have the same meanings as commonly understood by one of ordinary skill in the art to which the subject matter of the claims pertains. Unless otherwise stated, all patents, patent inventions, and publications cited in this document are incorporated herein by reference in their entirety. If multiple definitions exist for terms in this document, the definitions in this chapter shall prevail.

[0008] It should be understood that the above brief description and the following detailed description are exemplary and for illustrative purposes only, and do not limit the subject matter of the invention in any way. In this invention, the singular is used in conjunction with the plural unless otherwise specifically stated. It should also be noted that, unless otherwise stated, the use of “or” or “or” means “and / or”. Furthermore, the use of the term “comprising” and other forms such as “including,” “containing,” and “contains” are not limiting.

[0009] The definition of standard environmental terminology can be found in the reference "Textbook for Higher Education: Environmental Risk Assessment" (Higher Education Press, January 1, 2009, Edition: 1).

[0010] Unless specifically defined herein, the use of all commercially available products herein employs standard techniques. For example, it may be carried out using the manufacturer's instructions for use with the kit, or in accordance with methods known in the art or the description of this invention. The techniques and methods described herein can generally be implemented according to conventional methods well known in the art, based on the descriptions in the various summary and more specific documents cited and discussed in this specification.

[0011] The terms “optional / arbitrary” or “optionally / arbitrarily” mean that the event or situation described below may or may not occur, including both the occurrence and non-occurrence of the event or situation.

[0012] The term "raster data" as used in this article refers to a data form that divides space into regular grids, each grid being called a cell, and assigns corresponding attribute values ​​to each cell to represent entities.

[0013] The term "restricted cubic spline" used in this paper refers to a special piecewise polynomial spline function that, by imposing linear constraints on the boundary intervals of the nodes, makes the spline function exhibit a linear trend outside the boundary, thereby avoiding excessive fluctuations at the boundary.

[0014] The term "RCS spline function" used in this article refers to a piecewise polynomial function used in regression analysis. It ensures the smoothness of the function by imposing continuity constraints at the nodes and is often used to fit nonlinear relationships.

[0015] The term "B-spline function" used in this paper refers to a set of piecewise polynomial basis functions that have three main characteristics: local support, higher-order continuity, and numerical stability. They are used to expand continuous variables into smooth, differentiable curves in nonlinear modeling.

[0016] The term "nonlinear interpolation" used in this article refers to a method that uses nonlinear methods to interpolate missing values.

[0017] The term "cohort data" as used in this article refers to longitudinal data collected through cohort studies. This involves prospective or retrospective tracking of a population with common characteristics (such as birth time, exposure status, geographic location, and occupation), and the continuous recording of each member's events / state changes in chronological order. The term "time-dependent survival analysis" used in this paper refers to the analysis of survival data in which the effects of covariates on the risk of health outcomes change over time, in order to accurately characterize the risk of health events. It can be used for the analysis of survival data that do not meet the proportional hazards assumption.

[0018] The term “DAG” used in this paper refers to a Directed Acyclic Graph, which is a graph structure consisting of nodes and directed edges without directed cycles. It is used to express causal hypotheses and can be used to identify the minimum sufficient set of adjustments that blocks hybrid paths.

[0019] The term “β of time-dependent survival analysis” used in this paper refers to the covariate regression coefficient of the time-dependent survival analysis model, which characterizes the linear effect of each unit increase in the covariate on the logarithmic risk.

[0020] The term "semi-parametric modeling" used in this article refers to a model construction method that, in survival analysis, does not pre-determine the distribution form of the baseline hazard function, but instead parameterizes the exponential function of the linear combination of covariates.

[0021] The term “Bayesian generalized Stein-Fojo bi-exponential growth-decay model” used in this paper refers to a parameterized function based on the Bayesian framework that reflects the growth and / or decay characteristics of the response variable as the explanatory variable changes, and can flexibly fit the nonlinear characteristics of the growth and / or decay trends.

[0022] The term “shape-constrained health effect function (SCHIF)” used in this paper refers to a family of parameterized functions used to fit the relationship between exposure and health effects and to impose shape constraints on the relationship curve based on scientific prior knowledge.

[0023] The term “counterfactual distribution” as used in this article refers to the exposure distribution of an alternative population used to compare with actual exposure, usually set as the exposure distribution at the theoretical minimum risk exposure level.

[0024] The term used in this article is "annual population-weighted PM2.5". 2.5 "Exposure level" refers to a value that accurately represents PM by combining temporal and spatial dimensions and using population data as weights. 2.5 The annual average exposure value of the impact on the population throughout the study area.

[0025] The term "life expectancy" as used in this article refers to the average number of years a person of a certain age can live, calculated using life tables based on the age group mortality rates of a population in a country or region.

[0026] The term "healthy life expectancy" as used in this article refers to life expectancy under healthy conditions, that is, the average number of years an individual can live in good health.

[0027] The term “attributable disability life years” used in this article refers to the number of years of life loss due to disability that could be avoided when the exposure level is reduced to the counterfactual distribution, measured in person-years, calculated using a family of exposure-health effect functions and population exposure distributions within the framework of comparative risk assessment.

[0028] The term “Markov Chain Monte Carlo (MCMC) Sampling Estimation of Parameter Posterior Distribution” used in this paper refers to the process of constructing a Markov chain with a stationary parameter posterior distribution in Bayesian inference, and using a series of related samples generated by the chain to approximate the posterior distribution, thereby performing parameter estimation and uncertainty quantification analysis.

[0029] The term "generalized Sigmoid function" used in this paper refers to an extension based on the standard Sigmoid function. In this patent, it specifically refers to adding a shape parameter γ to the position parameter μ and scale parameter ν, and satisfying γ>0, so that the Sigmoid function maintains the bounded monotonic S-shaped characteristics, while adjusting the steepness of the curve by combining the shape parameter γ.

[0030] The term “two-compartment model-Runge-Kutta algorithm” used in this paper refers to the following: In a compartment model constructed with two compartments, health and disease, ordinary differential equations are used to characterize the dynamic transition process between the two compartments, and the Runge-Kutta algorithm is used to numerically integrate the equation system to obtain the time evolution trajectory. Then, the least squares or maximum likelihood methods are combined with data fitting to estimate the model parameters.

[0031] The term "nonparametric penalized spline function" used in this paper refers to a nonparametric estimation method that controls the smoothness of curve fitting and handles overfitting without pre-setting a specific smooth function form.

[0032] Data source The data sources used in this invention come from at least the following two sources: 1. Pearl River Cohort Resources. See the article "Cohort Profile: The Pearl River Cohort Study" (International Journal of Epidemiology, Volume 53, Issue 5, October 2024, dyae112).

[0033] 2. Charls provides public queue resources. See the website https: / / charls.pku.edu.cn / en / .

[0034] It is worth noting that the disclosure of the data sources used in the present invention is only for the purpose of providing a clear and complete description of the present invention, and is sufficient to enable those skilled in the art to understand and implement the present invention, and should not be construed as a limitation on the technical solutions to be protected by the present invention.

[0035] In a first aspect, the present invention provides a method for processing environmental exposure data and assessing health effects, comprising the following steps: S1. Data processing, including: S101. Organize the cohort dataset including age, address and health outcome, clean the variables, and obtain unweighted cohort data. S102. Matching environmental pollutant exposure concentrations based on residential address; S103. Using the unweighted cohort data obtained in step S101 and the environmental pollutant exposure concentrations matched in step S102, raster data is formed, and the population-weighted environmental pollutant exposure concentrations are estimated based on the raster data. S2. Health effect assessment, including: S200. Based on the population-weighted environmental pollutant exposure concentrations obtained in step S103, fit the age-group health effects Θ of pollutant exposure attribution. S201. Based on the population-weighted environmental pollutant exposure concentrations obtained in step S103 and the age and health outcomes in the cohort data obtained in step S101, perform time-dependent survival analysis by age stratification to obtain the regression coefficient β of the time-dependent survival analysis. S202. Define the curve characteristics of the age-specific effect value: When fitting, use the functional relationship between the regression coefficient β of the survival analysis and age to construct the subsequent shape-constrained health effect function SCHIF(z) for each age group. S203. Based on the age-group shape-constrained health effect function SCHIF(z) obtained in step S202 and the population-weighted environmental pollutant exposure concentration obtained in step S103, the quantitative relationship between the population-weighted annual average pollutant exposure level and life expectancy and healthy life expectancy is characterized, and the change in life expectancy and healthy life expectancy caused by changes in pollutant concentration is calculated accordingly.

[0036] Furthermore, the estimation described in step S103 is performed by nonlinear interpolation using a restricted cubic spline function.

[0037] Further, in step S200, the fitting includes the following steps: An estimation method based on semi-parametric modeling is used to obtain the health effects θ of pollutant exposure attribution across all age groups.

[0038] Further, in step S201, the time-dependent survival analysis includes the following steps: A directed acyclic graph (DAG) is introduced to extract variables from the cohort dataset, forming a minimum adequately adjusted set. Using health events observed during follow-up of the cohort data as outcomes, a regression coefficient β for time-dependent survival analysis is obtained based on a semi-parametric modeling estimation method.

[0039] Furthermore, in steps S201 and S200, the estimation method for the semi-parametric modeling is calculated using a time-dependent Cox regression model.

[0040] As some specific embodiments of the present invention, the time-dependent Cox regression model is as follows: The conditional risk function for individual i at time t:

[0041] in, Used as the benchmark risk function; The concentration of pollutants exposed over time (the unit of concentration can be selected from μg / m³). 3 ppm or particles / cm 3 , etc.); β is the regression coefficient (log hazard ratio per unit of X); For covariates; T Represents the matrix transpose operation; for The coefficient vector.

[0042] Further, in step S202, the functional relationship between the regression coefficient β of the time-dependent survival analysis and age is: ; Where age is the age in years; A is the scale parameter, a dimensionless quantity; B is the growth rate parameter, a dimensionless quantity; C is the equilibrium parameter, a dimensionless quantity; D is the decay rate parameter, a dimensionless quantity; and β is the regression coefficient, obtained through step S201. A, B, C, and D were all obtained through fitting. The fitting method was as follows: a Bayesian generalized Stein-Fojo double exponential growth-decay model was constructed, and the U-shaped, J-shaped, or configuration curve relationship between the regression coefficient β of the survival analysis and age was used for fitting. Furthermore, in step S202, the regression coefficient β is used to calculate the age-specific health effect θ based on pollutant exposure attribution, specifically as follows: θ = β; Further, in step S202, the relationship between the shape-constrained health effect function SCHIF(z) and the pollutant concentration z is expressed as follows: ; Where θ represents the age-specific health effect; ; ; z represents the pollutant concentration, WHO cf ν is the counterfactual distribution parameter; α is the upper asymptotic characteristic parameter of the exposure health effect; μ is the location parameter; ν is the scale parameter; and γ is the shape parameter.

[0043] Furthermore, the calculation described in step S203 includes the following steps: S2031. Calculate the probability of death after eliminating risk factors under the counterfactual distribution framework; S2032, Calculate the loss of healthy life expectancy.

[0044] Secondly, the present invention provides the application of the above-described method in the assessment of the health effects of pollutants.

[0045] Thirdly, the present invention provides an apparatus for processing environmental exposure data and assessing health effects, comprising: The data processing module is used to process the data to obtain population-weighted environmental pollutant exposure concentrations; The regression coefficient β calculation module is used to calculate the regression coefficient β based on the population-weighted environmental pollutant exposure concentration transmitted from the data processing module. The shape-constrained health effect function (SCHIF(z)) generation module is used to obtain the relationship expression between the shape-constrained health effect function (SCHIF(z)) and the pollutant concentration z for different age groups based on the regression coefficient β obtained by the regression coefficient β calculation module. The assessment module is used to calculate the relationship between the annual population-weighted pollutant exposure level and life expectancy and healthy life expectancy caused by changes in pollutant concentration, based on the population-weighted environmental pollutant exposure concentration transmitted from the data processing module and the relationship expression between the age-group shape-constrained health effect function SCHIF(z) and pollutant concentration z transmitted from the shape-constrained health effect function SCHIF(z) generation module.

[0046] Fourthly, the present invention provides an electronic device, the electronic device comprising: Memory, which stores executable instructions; A processor that executes the executable commands in the memory to implement the method described above.

[0047] Fifthly, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0048] Compared with the prior art, the beneficial effects of the present invention are at least as follows: By introducing a shape control function containing position, size, and shape parameters into the shape-constrained health effect function, fitting curves with different characteristics and shapes were successfully constructed. This enabled the realization of fitting effect features with more different nonlinear changes, making health effect assessment more intuitive and improving accuracy, flexibility, and operability. Attached Figure Description

[0049] Figure 1 This is a flowchart of Example 1.

[0050] Figure 2 The graph shows the curve shape of different α parameter values ​​in Example 1.

[0051] Figure 3 This is a curve diagram showing the shape of the l(z) parameter adjusted in Example 1.

[0052] Figure 4 The graph shows the curve shape of γ with different values ​​in Example 1.

[0053] Figure 5 For PM 2.5 and PM 10The plot shows the hazard ratio results of the Bayesian generalized Stein-Fojo model and the age-stratified time-dependent survival analysis model. The red line segment represents the results after S202 fitting, while the blue line segment represents the results without S202 fitting. Detailed Implementation

[0054] The following non-limiting embodiments are intended to enable those skilled in the art to gain a more comprehensive understanding of the present invention, but do not limit the invention in any way. The following content is merely an exemplary description of the scope of protection claimed by the present invention, and those skilled in the art can make various changes and modifications to the present invention based on the disclosed content, and such changes should also fall within the scope of protection claimed by the present invention.

[0055] The present invention will be further described below by way of specific embodiments. Unless otherwise specified, all instruments, devices, equipment, reagents, products, etc., used in the embodiments of the present invention are obtained through conventional commercial means.

[0056] Example 1 A method for processing environmental exposure data and assessing health effects, flowchart as follows: Figure 1 As shown. The specific steps are as follows.

[0057] S1, Data Processing.

[0058] The process involves organizing and cleaning the queue dataset, and transforming addresses. It matches environmental exposure with individual addresses (based on Baidu API (Application Programming Interface) coordinate lookup). When processing environmental exposure, population weighting is considered simultaneously, i.e., estimating the weighted population concentration within a 1-kilometer radius of an individual's residential address based on raster data. Population data comes from the SEDAC (Socioeconomic Data and Applications Center) public database, which can also be extracted from the GEE (Google Earth Engine) platform. The innovative method utilizes the flexible nonlinear fitting characteristics, boundary estimation stability, and intuitive derivability of the restricted cubic spline RCS function, based on its population raster data, to replace the less robust endpoint fitting of the B-spline function, achieving nonlinear interpolation of the population raster data.

[0059] S2. Age-stratified time-dependent survival analysis based on cohort data. Wherein: S200. Based on the population-weighted environmental pollutant exposure concentrations obtained in step S103, fit the health effects Θ of pollutant exposure attribution across all age groups.

[0060] The total health effect across all ages attributing pollutant exposure to the fitted pollutant exposure is denoted as Θ, satisfying exp(Θ)≥1. Specifically, a semi-parametric modeling-based estimation method (using a time-dependent Cox regression model) is employed to obtain the total health effect Θ across all ages. Corrected covariates include influencing factors such as smoking, age, mean temperature, and ozone exposure.

[0061] The covariate Cox model is as follows: The conditional hazard function for individual i (regardless of age group) at time t :

[0062] in, This is the baseline risk function for all ages; The concentration of pollutants exposed over time (the unit of concentration can be selected from μg / m³). 3 ppm or particles / cm 3 , etc.); Θ represents the total health effect across all ages (model regression coefficients, log hazard ratio for each unit of X); For covariates; T Represents the matrix transpose operation; for The coefficient vector.

[0063] S201. Based on the population-weighted environmental pollutant exposure concentrations obtained in step S103 and the age and health outcomes in the cohort data obtained in step S101, perform age-stratified time-dependent survival analysis to obtain the regression coefficient β for the time-dependent survival analysis. Details are as follows: The dataset is stratified by age, dividing it into multiple subsets based on age. Generally, the step size for age stratification is 1-10 years. For example, with a step size of 10 years, the age stratifications are 0-10 years, 10-20 years, 20-30 years, 30-40 years, 40-50 years, 50-60 years, 60-70 years, 70-80 years, and 80-90 years.

[0064] After age stratification, a DAG (Directed Acyclic Graph) is introduced to refine the minimum sufficient adjustment set of the cohort data for time-dependent survival analysis of age stratification (covariates considered include smoking, average temperature, and ozone exposure). Observable health events during follow-up of the cohort data are used as outcomes, such as hospitalization for chronic diseases, death from chronic diseases, and all-cause mortality. Based on a semi-parametric modeling estimation method, the β of the time-dependent survival analysis is obtained.

[0065] The estimation method for semi-parametric modeling is to use a time-dependent covariate Cox model for calculation.

[0066] Furthermore, the time-dependent covariate Cox model is as follows: Construct conditional risk functions for time t for individual i by age group. as follows:

[0067] in, This is the baseline risk function for each age group; The concentration of pollutants exposed over time (the unit of concentration can be selected from μg / m³). 3 ppm or particles / cm 3 , etc.); β is the regression coefficient (log hazard ratio per unit of X); For covariates; T Represents the matrix transpose operation; for The coefficient vector.

[0068] S202. Define the curve characteristics of age-specific effect values: When fitting, construct the shape-constrained health effect function SCHIF(z) for each age group based on the functional relationship between the regression coefficient β of the survival analysis and age.

[0069] Based on the literature review, the curve characteristics of age-specific effect sizes are defined, and a function of β and age based on survival analysis is constructed, i.e. The innovative approach involves introducing a parametric growth model from oncology to construct a Bayesian generalized Stein-Fojo bi-exponential growth-decay model, fitting U-shaped, J-shaped, and configurational curve relationships between β and age. A, B, C, and D are then determined. The age can be extrapolated based on the corresponding age stratification.

[0070] β is used as the age-specific health effect θ based on pollutant exposure attribution.

[0071] The specific method for calculating the age-specific health effect θ based on pollutant exposure attribution is as follows: β= ; A, B, C, and D are obtained through step S202; age is the age in years, which is determined according to the different age categories.

[0072] At this point, the β values ​​for each age group obtained based on S202 also satisfy β≥1. Specifically, as follows... Figure 5 As shown in the figure, the fitted experimental results converge more closely and the simulation effect is better than that of the unfitted experimental results.

[0073] This improvement is based on the framework of the Global Exposure Mortality Model (GEMM). The core of the GEMM model is... SCHIF stands for Shape Constrained Health Impact Function, and its key innovative improvements are as follows: ① Compared to existing technologies where θ is derived from pooled evidence from cohort studies, the parameter in this invention is directly derived from the actual assessment results of regional population cohorts. This provides more valuable reference for comparative risk assessment studies of environmental factors in regions and can provide evidence-based basis for the formulation and adjustment of refined regional environmental health policies.

[0074] ② Compared to the linear and log-linear structures of existing technologies, a counterfactual distribution (CF) is introduced for comparison, i.e. Where z represents the pollutant concentration, WHO cf α is the counterfactual distribution parameter; α is the upper asymptotic characteristic parameter. Based on the degree of difference in this comparison, the overall monotonicity of the nonlinear logarithmic structure can be measured. The curve shapes for different values ​​of the α parameter are as follows: Figure 2 As shown, it can be seen that the shape of the curve can be easily adjusted by taking different values ​​for α. ③ Extending to a generalized sigmoid function, introducing the γ parameter, satisfying...

[0075] (The constraints are) (κ and ρ are different shape adjustment parameters), which adjust the curve form of the Sigmoid function, such as the inflection point, smoothness, and tail variation. The adjusted parameters are in the form of: Figure 2 .

[0076] μ (position parameter): strictly anchors the inflection point, i.e. the point of maximum slope; ν (scale parameter): controls the smoothness, ν↑→ the curve is gentler and the transition bandwidth is wider; ν↓→ the curve is steeper and the abrupt change is obvious; γ (shape parameter): controls the change of the curve tail, γ↑→ rapid rise, γ↓→ slow rise.

[0077] The shape of the curve varies as γ takes different values. Figure 4 .

[0078] Figures 2-4 In the diagram, the horizontal axis represents the independent variable z, which indicates the pollutant concentration. In this embodiment, the unit of concentration is μg / m³. 3 The units for pollutant concentration can be adjusted according to the actual situation, but different units for pollutants do not affect the shape of the curve.

[0079] S203. Based on the age-group shape-constrained health effect function SCHIF(z) obtained in step S202 and the population-weighted environmental pollutant exposure concentration obtained in step S103, the quantitative relationship between the population-weighted annual average pollutant exposure level and life expectancy and healthy life expectancy is characterized, and the change in life expectancy and healthy life expectancy caused by changes in pollutant concentration is calculated accordingly.

[0080] Based on age-group-specific SCHIF(z) functions, combined with the Jiang life table and Sullivan method, the loss of life expectancy and healthy life expectancy attributable to pollutant exposure is estimated. Pollutants can be selected arbitrarily, specifically particulate pollutants such as PM2.5. 2.5 and PM 10 Any one of them. Using PM 2.5 For example, under different emission reduction scenarios, PM2.5 levels are based on different intervention levels. 2.5 Changes in concentration can be used to estimate the annual population-weighted PM2.5 concentration. 2.5 The impact of changes in exposure levels on life expectancy and healthy life expectancy. The main improvements and innovations here are: ① Corrected mortality probability after eliminating risk factors under the counterfactual distribution framework:

[0081] ② Based on the estimation of attributable disability years of life (YLD) due to risk factors, a health-adjusted weight is constructed to calculate the loss of healthy life expectancy:

[0082] YLD estimation employs a framework of "two-compartment model - automatic step-size Runge-Kutta algorithm" to simulate morbidity. A system of nonlinear coupled differential equations is constructed with relative risk (RR), incidence rate, mortality rate, and remission rate as inputs. The automatic step-size Runge-Kutta numerical integration method is used to solve this system, incorporating an adaptive step-size update mechanism into the Runge-Kutta algorithm: during integration, the step size is dynamically adjusted by comparing the local truncation errors of the full step size and half step size to balance computational accuracy and efficiency. The adaptive step-size Runge-Kutta method can efficiently and accurately capture the temporal evolution characteristics of YLD under the synergistic effects of relative risk, incidence rate, mortality rate, and remission rate, providing a solid numerical foundation for the refined calculation of healthy life expectancy loss in multi-state life table models.

[0083] Example 2 The method described in Example 1 is used for PM2.5 PM 10 Application of exposure attribution in health risk assessment.

[0084] Data preparation: A directed acyclic graph (DAG) is used to extract and organize variables from the cohort dataset, forming the dataset. Key individual-level variables collected include observable health outcome events from cohort follow-up (e.g., all-cause mortality, chronic disease visits or hospitalizations), as well as age, time of outcome, and address. First, adverse environmental exposures (e.g., temperature, PM2.5) are matched based on address. 2.5 PM 10 The raster data (etc.) are then considered. Next, a population-weighted average algorithm is considered, using the restricted cubic spline RCS function to perform multi-year interpolation estimation of the population raster data, which is used as a parameter for calculating the population-weighted average of regional adverse environmental exposure.

[0085] Data Analysis: First, survival analysis was performed using age stratification to obtain the β value ( Figure 5 ); Fitted PM 2.5 The age-specific health effect value θ of exposure attribution was determined. Based on a literature review, the curve characteristics of the age-specific effect value were defined. Using a Bayesian generalized Stein-Fojo bi-exponential growth-decay model, U-shaped, J-shaped, and configurational curves were fitted to the relationship between β and age. The dose-response relationship between exposure and outcome was fitted using an improved shape-constrained health effect function (SCHIF). The relevant parameter values ​​are shown in Table 1 below (γ=1).

[0086] Table 1

[0087] Example 3 An apparatus for performing environmental exposure data processing and health effect assessment using the method described in Example 1 comprises the following modules: The data processing module is used to process the data to obtain population-weighted environmental pollutant exposure concentrations; The regression coefficient β calculation module is used to calculate the regression coefficient β based on the population-weighted environmental pollutant exposure concentration transmitted from the data processing module. The shape-constrained health effect function (SCHIF(z)) generation module is used to obtain the relationship expression between the shape-constrained health effect function (SCHIF(z)) and the pollutant concentration z for different age groups based on the regression coefficient β obtained by the regression coefficient β calculation module. The assessment module is used to calculate the relationship between the annual population-weighted pollutant exposure level and life expectancy and healthy life expectancy caused by changes in pollutant concentration, based on the population-weighted environmental pollutant exposure concentration transmitted from the data processing module and the relationship expression between the age-group shape-constrained health effect function SCHIF(z) and pollutant concentration z transmitted from the shape-constrained health effect function SCHIF(z) generation module.

[0088] Example 4 An electronic device includes: a memory storing executable instructions; and a processor that executes the executable instructions in the memory to implement a method for environmental exposure data processing and health effect assessment as described in this invention (including embodiment 1).

[0089] In the electronic device described in this embodiment, the memory is used to store non-transitory computer-readable instructions. Specifically, the memory may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may, for example, include random access memory (RAM) and / or cache memory. The non-volatile memory may, for example, include read-only memory (ROM), hard disk, flash memory, etc.

[0090] In the electronic device described in this embodiment, the processor may be a central processing unit (CPU) or other form of processing unit with data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device to perform desired functions. In one embodiment of the present invention, the processor is used to execute computer-readable instructions stored in the memory.

[0091] Example 5 A computer-readable storage medium storing a computer program that, when executed by a processor, implements the method for environmental exposure data processing and health effect assessment described in this invention.

[0092] The aforementioned computer-readable storage medium stores non-transitory computer-readable instructions. When these non-transitory computer-readable instructions are executed by a processor, all or part of the steps of the method described in Embodiment 1 of the present invention are performed.

[0093] The aforementioned computer-readable media include, but are not limited to: Optical storage media, such as CD-ROM and DVD; magneto-optical storage media, such as MO; magnetic storage media, such as magnetic tape or portable hard drives; media with built-in rewritable non-volatile memory, such as memory cards; media with built-in ROM, such as ROM cartridges.

[0094] Finally, it should be noted that the above content is only used to illustrate the technical solution of the present invention, and is not intended to limit the scope of protection of the present invention. Simple modifications or equivalent substitutions made by those skilled in the art to the technical solution of the present invention do not depart from the essence and scope of the technical solution of the present invention.

Claims

1. A method of environmental exposure data processing and health effect assessment, characterized by, The method comprises the following steps: S1, data processing, comprising: S101, collating a cohort dataset comprising age, address and health outcome, variable cleaning, obtaining un-population-weighted cohort data; S102, matching environmental pollutant exposure concentration based on address; S103, forming a grid data using the un-population-weighted cohort data obtained in step S101 and the environmental pollutant exposure concentration matched in step S102, and estimating population-weighted environmental pollutant exposure concentration based on the grid data; S2, health effect evaluation, comprising: S200, fitting pollutant exposure-attributed health effects Θ of all ages based on the population-weighted environmental pollutant exposure concentration obtained in step S103; S201, performing age-stratified time-dependent survival analysis based on the population-weighted environmental pollutant exposure concentration obtained in step S103 and the age and health outcome in the cohort data obtained in step S101, and obtaining regression coefficients β of the time-dependent survival analysis; S202, setting the curve characteristics of the age-specific effect value: fitting the functional relationship between the regression coefficients β of the time-dependent survival analysis and the age age, and constructing an age-stratified shape-constrained health effect function SCHIF(z); S203, based on the age-stratified shape-constrained health effect function SCHIF(z) obtained in step S202 and the population-weighted environmental pollutant exposure concentration obtained in step S103, depicting the quantitative relationship between the population-weighted annual pollutant exposure level and the life expectancy and health life expectancy, and calculating the change amount of the life expectancy and health life expectancy caused by the change of the pollutant concentration.

2. The method of claim 1, wherein, The estimation in step S103 is a non-linear interpolation and completion estimation using a restricted cubic spline function.

3. The method of claim 1, wherein, In step S200, the fitting comprises the following steps: Based on the estimation method of semi-parametric modeling, the health effects Θ of all ages attributed to pollutant exposure are obtained; In step S201, the time-dependent survival analysis comprises the following steps: A DAG (Directed Acyclic Graph) is introduced to extract variables of the cohort dataset to form a minimum sufficient adjustment set; based on the estimation method of semi-parametric modeling, the regression coefficients β of the time-dependent survival analysis are obtained based on the health events observed in the cohort data follow-up as the outcome.

4. The method of claim 3, wherein, In steps S201 and S200, the estimation method based on semi-parametric modeling is to calculate using a time-dependent Cox regression model.

5. The method of claim 4, wherein, The functional relationship between the regression coefficients β of the survival analysis and the age age in the time-dependent Cox regression model is: ； wherein age is the age, unit: years; A is a scale parameter, dimensionless; B is a growth rate parameter, dimensionless; C is a balance parameter, dimensionless; D is a decay rate parameter, dimensionless; β is the regression coefficient, obtained through step S201; A, B, C and D are obtained by fitting: a Bayesian generalized Stein-Fojo double exponential growth-decay model is constructed to fit the U-shaped, J-shaped or configuration curve relationship between the regression coefficients β of the time-dependent survival analysis and the age age. In step S202, the shape-constrained health impact function SCHIF(z) and the pollutant concentration z are expressed as: ； where θ is the age-specific health effect, obtained by step S202, θ=β. ； ； z is the pollutant concentration, WHO cf is the counterfactual distribution parameter; a is the characteristic parameter of the upper asymptote of the health effect of exposure; μ is the location parameter; v is the scale parameter; γ is the shape parameter.

6. The method of claim 1, wherein, In step S203, the calculating includes the following steps: S2031, calculating the mortality probability after eliminating the risk factors in the counterfactual distribution framework; S2032, measuring the loss of healthy life expectancy.

7. The method of any one of claims 1-6 is applied to the assessment of the health effects of pollutants.

8. An apparatus for environmental exposure data processing and health effect assessment, comprising: Comprising: a data processing module for processing data to obtain population-weighted environmental pollutant exposure concentration; a regression coefficient β calculation module for calculating the regression coefficient β according to the population-weighted environmental pollutant exposure concentration transmitted by the data processing module; a shape-constrained health impact function SCHIF(z) generation module for obtaining the relationship expression of the shape-constrained health impact function SCHIF(z) and the pollutant concentration z according to the regression coefficient β obtained by the regression coefficient β calculation module; an evaluation module for calculating the relationship between the annual population-weighted pollutant exposure level and the life expectancy and healthy life expectancy due to the change of the pollutant concentration according to the population-weighted environmental pollutant exposure concentration transmitted by the data processing module and the relationship expression of the shape-constrained health impact function SCHIF(z) and the pollutant concentration z transmitted by the shape-constrained health impact function SCHIF(z) generation module.

9. An electronic device, comprising: The electronic device comprises: a memory storing executable instructions; a processor running the executable instructions in the memory to implement the method of any one of claims 1-6.

10. A computer readable storage medium, characterized in that, The computer readable storage medium stores a computer program which is executed by the processor to implement the method of any one of claims 1-6. The computer readable storage medium stores a computer program which is executed by the processor to implement the method of any one of claims 1-6.

Images