A method for risk assessment of water environment microplastics under multiple environmental factors

Abstract

The application relates to the technical field of environmental risk assessment, in particular to a water environment micro-plastic risk assessment method under multiple environmental factors, which comprises the following steps: constructing a micro-plastic pollution database and a multiple environmental factor resource database; pre-processing and feature engineering are performed on original data, core environmental driving factors are screened, and a standardized modeling data set is formed; a risk assessment model is constructed by fusing a harmful outcome path framework, a micro-plastic risk quantitative value is calculated by a weighted comprehensive index method or a machine learning algorithm; a final risk grade is determined in combination with a risk grade division standard and a toxic polymer adjustment mechanism, and an environmental management decision suggestion is output. The application solves the problems of single evaluation index, insufficient consideration of environmental factors, weak machine interpretation and insufficient prediction ability in the prior art, and realizes accurate and dynamic risk assessment of water environment micro-plastic pollution.

Description

Technical Field

[0001] This application relates to the field of environmental risk assessment technology, and in particular to a method for risk assessment of microplastics in aquatic environments under multiple environmental factors. Background Technology

[0002] Microplastics, as an emerging persistent pollutant, have been widely detected in various aquatic environments worldwide, and their potential ecological and health risks are receiving increasing attention. Accurate assessment and scientific management of microplastic risks in aquatic environments have become a critical issue urgently needing to be addressed in the environmental field.

[0003] Currently widely used risk assessment models, such as the pollution load index method and the polymer hazard index method, focus more on the physicochemical properties of microplastics themselves, such as abundance, particle size, and polymer type. This approach essentially isolates microplastics from the complex aquatic environment in which they exist, failing to simultaneously consider the synergistic effects of multiple environmental factors, including the physical, chemical, and biological characteristics of the water body, as well as surrounding human activities. The assessment conclusions are insufficient to accurately reflect the actual ecological risks of microplastics in different environmental scenarios, and cannot effectively explain why the risks exhibit specific spatial distributions.

[0004] Whether it's a simple exponential aggregation or a complex statistical model, most methods fail to clearly elucidate the biological pathways through which key environmental drivers ultimately lead to harmful outcomes in organisms. This lack of mechanistic correlation makes it difficult for risk assessment results to support accurate risk tracing. Even when a high-risk conclusion is obtained, management departments still lack sufficient scientific evidence to formulate targeted intervention measures.

[0005] Current assessment methods suffer from low data utilization efficiency. The assessment model, heavily reliant on point sampling and laboratory analysis, is essentially a post-event assessment lagging behind pollution incidents. This not only limits the applicability of the methods in vast or data-scarce water areas but also hinders rapid risk prediction and early warning. Furthermore, the lack of effective integration and in-depth analysis of the vast amount of existing historical environmental monitoring data and microplastic survey data means that their enormous potential for predicting future risk trends has not been fully realized.

[0006] Therefore, there is an urgent need in this field to develop a new method for risk assessment of microplastics in the water environment that can systematically integrate multi-source environmental information, embed a biological mechanism explanation framework, and have good predictive and early warning capabilities. Summary of the Invention

[0007] The purpose of this application is to provide a method for risk assessment of microplastics in aquatic environments under multiple environmental factors, in order to solve the problems of single assessment indicators, insufficient consideration of environmental factors, weak mechanism explanation and insufficient predictive ability in the existing technology.

[0008] According to an embodiment of this application, a method for risk assessment of microplastics in aquatic environments under multiple environmental factors is proposed, including the following steps: S1: Construct a database of microplastic pollution and a database of resources for multiple environmental factors in the target water area; S2: Preprocess and feature-engineer the raw data from the microplastic pollution database and the multi-environmental factor resource database, screen the core environmental driving factors, and form a standardized modeling dataset; S3: Based on the standardized modeling dataset, construct a risk assessment model that integrates the harmful outcome path framework, and calculate the microplastic risk quantification value of the target water area; S4: Based on the microplastic risk quantification value, combined with the risk level classification standard and the toxic polymer adjustment mechanism, determine the final risk level of the target water area; S5: Based on the final risk level, output environmental management decision recommendations.

[0009] In summary, the beneficial technical effects of this application are as follows: By systematically integrating four categories of environmental factors—physical, chemical, biological, and human activities—with microplastic characteristic data, a risk assessment framework for the synergistic effects of multiple environmental factors can be established. This overcomes the shortcomings of traditional methods that rely on a single assessment dimension, and improves the authenticity and reliability of the assessment results.

[0010] By integrating the harmful outcome pathway framework, the abstract environmental factor weights in the model can be transformed into specific causal pathways from environmental drivers to biologically harmful outcomes, thus providing a solid mechanistic explanation for risk assessment results and providing key scientific evidence for accurate risk tracing.

[0011] By providing two independent technical paths—the weighted composite index method and machine learning algorithms—we can achieve complementary advantages. The former ensures the computational efficiency and interpretability of the model, while the latter excels at uncovering complex nonlinear relationships. Together, they achieve a balance between accuracy, robustness, and predictability in the evaluation process.

[0012] By using machine learning, key microplastic pollution indices can be predicted using only readily available environmental factor data, thus effectively breaking the dependence on direct, high-frequency monitoring data and providing an effective technical means for large-scale, rapid risk assessment and early warning. Attached Figure Description

[0013] Figure 1 This is a flowchart illustrating the steps of a method for assessing the risk of microplastics in aquatic environments under multiple environmental factors, as described in this application. Detailed Implementation

[0014] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.

[0015] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0016] In the embodiments of this application, the same reference numerals denote the same components, and for the sake of brevity, detailed descriptions of the same components are omitted in different embodiments. It should be understood that the thickness, length, width, and other dimensions of various components in the embodiments of this application shown in the accompanying drawings, as well as the overall thickness, length, width, and other dimensions of the integrated device, are merely illustrative and should not constitute any limitation on this application.

[0017] In this application, "multiple" means two or more (including two).

[0018] The following is combined with Figure 1 A method for risk assessment of microplastics in aquatic environments under multiple environmental factors, according to an embodiment of this application, is described in detail below.

[0019] See appendix Figure 1 This is a flowchart illustrating the steps of a method for assessing the risk of microplastics in aquatic environments under multiple environmental factors, as described in this embodiment.

[0020] This embodiment uses a major river and its connected scenic lakes within a certain urban area as the target water body to illustrate the specific implementation process of this application.

[0021] S1: Construct a database of microplastic pollution and a database of resources for multiple environmental factors in the target water area; Microplastic sample collection and characterization: In the target water area, 15 sampling points were systematically set up based on hydrological characteristics and the distribution of potential pollution sources.

[0022] A 100L surface water sample (0-0.5m depth) was collected at each sampling point using a stainless steel water sampler and then filtered and enriched through a series of stainless steel screens with 5mm and 0.45μm apertures.

[0023] The residue enriched on the 0.45 μm filter membrane was completely transferred to a clean glass petri dish with ultrapure water and dried in an oven at 60 °C to constant weight.

[0024] The dried samples were observed using a stereomicroscope equipped with image analysis capabilities. All suspected microplastic particles were counted, and their morphological characteristics, including fibers, fragments, films, and granules, were recorded. The particle size of at least 200 particles was randomly measured using the microscope's accompanying software, and their distribution was statistically analyzed.

[0025] At least 20% of the particles at each sampling point were randomly selected for polymer identification using Fourier transform infrared spectroscopy. The infrared absorption spectra of unknown particles were compared with the NIST standard polymer spectrum library; a match greater than 75% was considered a valid polymer.

[0026] Finally, the abundance, particle size distribution, and polymer composition of microplastics at each sampling point were statistically obtained.

[0027] It should be noted that stereomicroscopes, due to their three-dimensional imaging capabilities and long working distance, are particularly suitable for the preliminary identification, counting, and physical morphological classification of microplastic particles with diverse shapes. Their optical principles provide a wide field of view and deep depth of field, facilitating the observation of complex samples.

[0028] Fourier transform infrared spectroscopy is an identification technique based on molecular vibrational energy level transitions. When a sample is irradiated with infrared light, specific chemical bonds or functional groups in the molecule absorb infrared light at frequencies matching their vibrations, forming a unique molecular fingerprint absorption spectrum. By comparing this spectrum with standard spectra of known polymers, qualitative analysis of the composition of microplastic polymers can be achieved.

[0029] Synchronous acquisition and spatiotemporal matching of environmental factor data: At the same time and location when water samples are collected, the following four types of environmental factor data should be measured or collected simultaneously: Physical environmental factors: water temperature, pH value, dissolved oxygen, conductivity, turbidity Chemical environmental factors: total nitrogen, total phosphorus, ammonia nitrogen, chemical oxygen demand Biological environmental factors: chlorophyll a concentration Factors influencing human activities: land use type within a 500-meter buffer zone around the sampling point, and distance to the nearest sewage outlet. All the above data were spatiotemporally matched to ensure that each microplastic data point had corresponding environmental factor data at the same spatiotemporal coordinates.

[0030] All microplastic characteristic data will be stored in the microplastic pollution database, and all environmental factor data will be stored in the multi-environmental factor resource database.

[0031] The database can be built using a relational database, and each record contains a unique sampling point ID, latitude and longitude coordinates, and sampling timestamp.

[0032] It should be noted that spatiotemporal matching is a prerequisite for establishing a reliable causal relationship. It requires that environmental factor data and microplastic data come from the exact same geographical location and the same sampling time to avoid systematic errors introduced by spatiotemporal asynchrony and ensure that subsequent models can accurately reveal the real relationship between environmental factors and microplastic pollution.

[0033] Preliminary risk quantification: Based on the abundance of microplastics in the characteristic data, the pollution load index P of the target water area is calculated using the following formula: ; in, Let be the contamination coefficient of the i-th sampling point; denoted as , where is the measured abundance of microplastics at the i-th sampling point; B is the background abundance of microplastics in the surface water, taken as 6.65 items / L; and n is the total number of sampling points.

[0034] Based on the polymer composition in the microplastic characteristic data, the polymer hazard index H of the target water area is calculated using the following formula: ; in, Let j be the percentage of the j-th polymer. Let be the hazard factor corresponding to the j-th polymer.

[0035] The specific values ​​for the polymer hazard factor are as follows: Polyvinyl chloride corresponds to 10551, polyurethane corresponds to 13844, acrylonitrile-butadiene-styrene copolymer corresponds to 6552, polycarbonate corresponds to 1177, polystyrene corresponds to 30, polyamide corresponds to 47, polyethylene corresponds to 11, polyethylene terephthalate corresponds to 4, and polypropylene corresponds to 1.

[0036] It should be noted that the core idea of ​​the pollution load index is to quantify the multiple of the pollution level in the current study area relative to the general global level by comparing it with a recognized background value. It can be regarded as the pollution coefficient of the i-th point, while the pollution load index is the average pollution level of the entire study area.

[0037] The polymer hazard index is established based on the significant differences in the potential hazards of different polymers to organisms. This index, through weighted averaging, transforms polymer composition information into a comprehensive toxicity potential indicator. The hazard coefficients in the table are primarily derived from the toxicity data of the polymer monomers and the polymer's own environmental behavior characteristics; their values ​​can vary by several orders of magnitude, scientifically reflecting the relative risk levels of different polymers.

[0038] S2: Preprocess and feature engineering the raw data, screen the core environmental driving factors, and form a standardized modeling dataset; Data preprocessing: Box plots were generated for each environmental factor and microplastic abundance data. Values ​​below the lower quartile -1.5 × interquartile range or above the upper quartile +1.5 × interquartile range were identified as outliers and removed.

[0039] For data with random missing values, the K-nearest neighbor algorithm is used for imputation. In this embodiment, K=5 is set, and the Euclidean distance is used to find the 5 most similar samples, and their mean is used to imput the missing values.

[0040] For data segments with consecutive missing values, linear interpolation is used to repair them based on known data points before and after the time series. This repair operation is only performed on data segments with consecutive missing values ​​≤ 72 hours. Data segments with consecutive missing values ​​> 72 hours are marked as invalid data and removed.

[0041] It should be noted that the box plot method is a robust outlier identification method based on the distribution of the data itself. It uses the quartiles of the data to describe the dispersion of the data and regards extreme values ​​that are far from the main distribution of the data as outliers. This method does not depend on the data following a specific distribution.

[0042] The principle of the K-nearest neighbor algorithm for imputation is based on the idea that similar things cluster together. That is, the missing value of a sample is likely to be similar to the values ​​of its K nearest neighbors in the feature space. The algorithm finds the nearest neighbors by calculating the distance between the samples in the multidimensional feature space, and is suitable for situations with complex data structures and random missing patterns.

[0043] Linear interpolation assumes that the trend of data change is approximately linear over a short time interval. It uses two known data points before and after a missing point to construct a linear function to estimate the missing value. It is suitable for repairing consecutive small missing values ​​in time series data.

[0044] Data standardization: For biological and chemical data that conform to or are approximately normally distributed, the Z-score standardization method is used, and the mean of the processed data is 0 and the standard deviation is 1.

[0045] For physical and concentration data, the extreme value normalization method is used to linearly transform their values ​​to the [0,1] interval. The pH value is used as physical data and is also standardized using this method.

[0046] It should be noted that standardization aims to eliminate the influence of differences in units and value ranges among different features, bringing all features to the same order of magnitude. Z-score standardization is suitable for data with unrestricted distributions; it preserves the original distribution shape but shifts its center to 0 and scales it to 1. Extreme value normalization can strictly limit the data to the range [0,1], which is particularly suitable for data with clear physical thresholds or concentration-based data, and helps improve the convergence speed and performance of subsequent distance-based models.

[0047] Screening core environmental drivers: The calculated pollution load index P and polymer hazard index H were used as proxy indicators of microplastic risk; all standardized environmental factors were used as independent variables, and the random forest regression algorithm was used to assess feature importance.

[0048] The calculated feature importance is sorted in descending order, and the top N environmental factors with a cumulative contribution of more than 80% are selected to form the core environmental driving factors.

[0049] It should be noted that proxy indicators are used because it is difficult and expensive to obtain direct and comprehensive data on the ecotoxicological effects of microplastics, while P and H indices are widely accepted indirect measures that are highly correlated with the ultimate ecological risk.

[0050] The Random Forest algorithm evaluates feature importance by calculating the reduction in Gini impurity or the reduction in mean squared error. The principle is that the higher the average improvement in model purity brought about by a feature when it is used to split nodes in all decision trees, the more important the feature is.

[0051] The cumulative contribution of 80% reflects the Pareto principle commonly used in statistics, which aims to capture the most important information with the fewest factors, achieve dimensionality reduction, and avoid overcomplicating the model.

[0052] S3: Construct a risk assessment model that integrates a harmful outcome pathway framework and calculate the quantification value of microplastic risk; Two parallel technical paths are provided, and either one can be chosen for implementation.

[0053] The weighted composite index method is adopted: The weights of each core environmental driving factor, as well as the weights of the pollution load index P and the polymer hazard index H, are determined using the analytic hierarchy process (AHP). The AHP requires the construction of a 1-9 scale judgment matrix, which undergoes a consistency test (CR < 0.1) to ensure the scientific validity of the weight allocation. The standardized P and H are then weighted and summed to calculate the basic microplastic pollution index C. Finally, the microplastic risk quantification value R for the target water area is calculated using the following formula: ; in, ω represents the standardized k-th core environmental driver; ω is the weight of the microplastic pollution baseline index C. Let be the weight of the k-th core environmental driving factor; m be the total number of core environmental driving factors; all weights satisfy: ; The weight ω∈[0.3,0.5] of the basic index C of microplastic pollution.

[0054] It should be noted that the weighted composite index method model is transparent, computationally efficient, and the results are easy to interpret and trace.

[0055] The range of values ​​for the weight ω is designed to ensure that the pollution status of microplastics themselves accounts for a significant and fundamental proportion of the final risk, while environmental driving factors jointly determine the remainder.

[0056] The Analytic Hierarchy Process (AHP) is a multi-criteria decision-making method that combines qualitative and quantitative analysis. By constructing a judgment matrix and performing consistency tests, it scientifically determines the relative importance of each factor.

[0057] Machine learning algorithms are used: Historical data of at least 3 years were retrieved from the database, using core environmental driving factors as features and pollution load index P and polymer hazard index H as labels. A risk assessment model was established using the gradient boosting decision tree algorithm, and grid search and 5-fold cross-validation were used for model selection and hyperparameter optimization. The hyperparameters optimized by grid search included a learning rate of 0.01-0.1, a tree depth of 3-10 layers, and a number of trees of 100-300. The standardized modeling dataset of the target water area was input into the trained model, and the predicted values ​​of P and H were output. The predicted values ​​of P and H were weighted and summed in a 1:1 ratio to calculate the final microplastic risk quantification value R.

[0058] It should be noted that the advantage of the machine learning approach lies in its ability to automatically capture the complex nonlinear relationships and interactions between environmental factors and microplastic risks.

[0059] Gradient boosting decision trees are an ensemble learning algorithm that trains multiple weak decision trees sequentially, with each tree working to correct the residuals of the previous tree, ultimately combining them into a powerful predictive model.

[0060] Grid search and cross-validation are standard model optimization processes designed to systematically search for the optimal combination of hyperparameters, objectively evaluate the model's generalization ability, prevent overfitting, and ensure the model's stability and reliability.

[0061] Integrating harmful outcome path frameworks: This framework is not a specific calculation formula, but a mechanistic explanation framework. When building models or analyzing results, it is necessary to establish a chain of causal pathways from core environmental drivers to biologically harmful outcomes.

[0062] It should be noted that the harmful outcome pathway framework is a conceptual model in the field of toxicology used to describe a series of known, sequential biological events leading from molecular initiation events to harmful outcomes at the individual / population level. Integrating it into the model of this application means logically and interpretively linking the key inputs and outputs of the mathematical model with the key events in the harmful outcome pathway. For example, when the model shows turbidity as a significant driving factor, it can be interpreted in conjunction with the harmful outcome pathway as follows: high turbidity may affect the filter-feeding behavior of aquatic organisms, increasing the probability of microplastic ingestion, thereby triggering cellular events such as physical damage and inflammation, ultimately leading to growth inhibition, providing a scientific explanation consistent with existing toxicological knowledge.

[0063] S4: Determine the final risk level based on the microplastic risk quantification value, combined with the risk level classification standard and the toxic polymer adjustment mechanism; Risk level classification: The risk level is divided into three levels: low risk, medium risk, and high risk. The threshold of the risk quantification value R corresponding to each level is determined by the ternary digits of the historical R values ​​stored in the microplastic pollution database.

[0064] Determine the initial risk level: Based on the microplastic risk quantification value R calculated by S3, determine the initial risk level of each sampling point.

[0065] Toxicity polymer adjustment mechanism: If the polymer hazard index H of a sampling point is greater than 1000, or if at least two of the following polymers are detected at the sampling point: polyvinyl chloride, polyurethane, acrylonitrile-butadiene-styrene copolymer and polycarbonate, the final risk level of the sampling point is determined to be one level higher than its initial risk level; when the initial risk level is high risk, the final risk level remains high risk.

[0066] Generate spatial distribution map: Based on the spatial coordinates of each sampling point and the corresponding final risk level, generate a risk level spatial distribution vector file in the geographic information system software.

[0067] It should be noted that using ternary grading is an unsupervised, data-driven method. It adaptively performs relative grading based on the risk distribution of the current dataset, avoiding the difficulty of subjectively setting fixed thresholds and making the grading more objective and targeted.

[0068] The toxic polymer adjustment mechanism is a safety conservatism strategy based on the precautionary principle. Its scientific basis is that even if a certain highly hazardous polymer constitutes a small percentage of the total mass, its monomer leaching or intrinsic toxicity can still pose a disproportionately high risk. This mechanism, as an escalation rule, ensures that assessment results have sufficient sensitivity and early warning capability for such potentially high-risk scenarios.

[0069] S5: Based on the final risk level, output environmental management decision recommendations; Based on the final determined risk level and its spatial distribution, specific and actionable environmental management decision recommendations are output. For example: For high-risk areas, it is recommended to designate them as priority control areas for microplastics, strengthen the supervision of sewage discharge along the coast, and carry out source tracing; For medium-risk areas where adjustment mechanisms are triggered by toxic polymers, it is recommended to issue specific polymer alerts, with a focus on emissions from relevant industrial enterprises; For low-risk areas, it is recommended to maintain existing protective measures and implement annual routine monitoring.

[0070] It should be noted that this step directly transforms the technical assessment results into actionable environmental management decision recommendations, forming a complete closed loop from scientific understanding to management action.

[0071] The above description is merely a preferred embodiment of this application and is not intended to limit the scope of protection of this application. Therefore, all equivalent changes made to the structure, shape, and principle of this application should be included within the scope of protection of this application.

Claims

1. A method for risk assessment of microplastics in aquatic environments under multiple environmental factors, characterized in that, Includes the following steps: S1: Construct a database of microplastic pollution and a database of resources for multiple environmental factors in the target water area; S2: Preprocess and feature-engineer the raw data from the microplastic pollution database and the multi-environmental factor resource database, screen the core environmental driving factors, and form a standardized modeling dataset; S3: Based on the standardized modeling dataset, construct a risk assessment model that integrates the harmful outcome path framework, and calculate the microplastic risk quantification value of the target water area; S4: Based on the microplastic risk quantification value, combined with the risk level classification standard and the toxic polymer adjustment mechanism, determine the final risk level of the target water area; S5: Based on the final risk level, output environmental management decision recommendations.

2. The method for risk assessment of microplastics in aquatic environments under multiple environmental factors according to claim 1, characterized in that, Step S1 includes: Microplastic samples were obtained from the target water area and counted and morphologically analyzed using a stereomicroscope. No less than 20% of the individuals are randomly selected from the microplastic samples, and the polymer type is identified by Fourier transform infrared spectroscopy to obtain the microplastic characteristic data of the target water area. The microplastic characteristic data includes abundance, particle size distribution and polymer composition. Simultaneously collect data on four types of environmental factors: physical environmental factors, chemical environmental factors, biological environmental factors, and human activity influencing factors. Perform spatiotemporal matching of the four types of environmental factor data with the microplastic characteristic data. Store the four types of environmental factor data in a multi-environmental factor resource database and the microplastic characteristic data in a microplastic pollution database.

3. The method for risk assessment of microplastics in aquatic environments under multiple environmental factors according to claim 2, characterized in that, Step S1 further includes preliminary risk quantification of the microplastic feature data: Based on the abundance of microplastics in the characteristic data, the pollution load index P of the target water area is calculated using the following formula: ； in, Let be the contamination coefficient of the i-th sampling point; is the measured value of microplastic abundance at the i-th sampling point; B is the background value of microplastic abundance in surface water; n is the total number of sampling points; Based on the polymer composition in the microplastic characteristic data, the polymer hazard index H of the target water area is calculated using the following formula: ； in, Let j be the percentage of the j-th polymer. Let be the hazard factor corresponding to the j-th polymer.

4. The method for risk assessment of microplastics in aquatic environments under multiple environmental factors according to claim 3, characterized in that, When calculating the polymer hazard index, the hazard coefficient of the j-th polymer is taken as follows: Polyvinyl chloride corresponds to 10551, polyurethane corresponds to 13844, acrylonitrile-butadiene-styrene copolymer corresponds to 6552, polycarbonate corresponds to 1177, polystyrene corresponds to 30, polyamide corresponds to 47, polyethylene corresponds to 11, polyethylene terephthalate corresponds to 4, and polypropylene corresponds to 1.

5. The method for risk assessment of microplastics in aquatic environments under multiple environmental factors according to claim 1, characterized in that, Step S2 includes: The raw data from the microplastic pollution database and the multi-environmental factor resource database are preprocessed, the preprocessing including: Box plots were used to remove outlier data points that were below the lower quartile minus 1.5 interquartile range or above the upper quartile plus 1.5 interquartile range. For data with random missing information, the K-nearest neighbor algorithm is used for imputation. For data segments with continuous missing segments, linear interpolation is used for repair. The preprocessed raw data is standardized, and the standardization is performed according to the data type, including: For biological and chemical data that conform to a normal distribution, the Z-score standardization method is used to transform them into a distribution with a mean of 0 and a standard deviation of 1. For physical and concentration data, the extreme value normalization method is used to linearly transform their values ​​to the [0,1] interval.

6. The method for risk assessment of microplastics in aquatic environments under multiple environmental factors according to claim 3, characterized in that, The process of screening the core environmental driving factors in step S2 includes: Based on the feature importance assessment method, the pollution load index and polymer hazard index are used as risk proxy indicators to quantify the correlation strength between each environmental factor and microplastic pollution. Based on the correlation strength, each environmental factor is ranked, and the top N environmental factors with a cumulative contribution of more than 80% are selected to constitute the core environmental driving factors.

7. The method for risk assessment of microplastics in aquatic environments under multiple environmental factors according to claim 3, characterized in that, In step S3, the weighted composite index method is used to calculate the risk quantification value of the microplastics, including: The weights of each core environmental driving factor are determined using a weighting method. The basic microplastic pollution index C is calculated by weighted summing of the standardized pollution load index P and the polymer hazard index H. The microplastic risk quantification value R for the target water area is calculated using the following formula: ； in, ω represents the standardized k-th core environmental driver factor; ω is the weight of the microplastic pollution baseline index C. Let be the weight of the k-th core environmental driving factor; m be the total number of core environmental driving factors; all weights satisfy: ； The weight ω∈[0.3,0.5] of the microplastic pollution baseline index C.

8. The method for risk assessment of microplastics in aquatic environments under multiple environmental factors according to claim 3, characterized in that, In step S3, a machine learning algorithm is used to calculate the risk quantification value of the microplastics, including: A supervised learning training set was constructed, and historical data were retrieved from the microplastic pollution database and the multi-environmental factor resource database. The core environmental driving factors were used as features, and the pollution load index and polymer hazard index were used as regression targets. A risk assessment model was established using a supervised learning regression algorithm, and the model was trained using model selection and hyperparameter optimization techniques. The standardized modeling dataset of the target water area is input into the trained risk assessment model, which outputs predicted values ​​of pollution load index and polymer hazard index. The predicted values ​​of the pollution load index and the polymer hazard index are weighted and summed to calculate the quantified value of the microplastic risk.

9. The method for risk assessment of microplastics in aquatic environments under multiple environmental factors according to claim 1, characterized in that, The aforementioned fusion harmful outcome pathway framework refers to: establishing a chain causal pathway from core environmental drivers to biological harmful outcomes in model construction or outcome interpretation; and transforming abstract feature weights or contributions in the model into concrete biological mechanism explanations through the chain causal pathway.

10. The method for risk assessment of microplastics in aquatic environments under multiple environmental factors according to claim 1, characterized in that, Step S4 includes: A risk level classification standard is determined, and the risk level is divided into three levels: low risk, medium risk and high risk. The threshold of the microplastic risk quantification value corresponding to each level is determined by the ternary value of the microplastic risk quantification value of statistical historical data or current global data. The initial risk level of each sampling point is determined based on the microplastic risk quantification value. A toxic polymer adjustment mechanism is implemented; if the polymer hazard index H at a sampling point is greater than 1000, or if at least two polymers from polyvinyl chloride, polyurethane, acrylonitrile-butadiene-styrene copolymer, and polycarbonate are detected at the sampling point, the final risk level of the sampling point is determined to be one level higher than its initial risk level; when the initial risk level is high risk, the final risk level remains high risk. Based on the spatial coordinates of each sampling point and the corresponding final risk level, a risk level spatial distribution vector file is generated.

Images