A multi-source data intelligent collection method and system for occupational health monitoring

Abstract

This invention discloses an intelligent acquisition method and system for multi-source occupational health monitoring, relating to the field of occupational health monitoring technology. The method includes the following steps: collecting multi-source occupational health data within a target monitoring area; extracting features from the multi-source occupational health data; detecting anomalies in the multi-source occupational health data and locating target abnormal data items; constructing a data quality assessment model based on the duration of the anomalies in the target abnormal data items and the industry type parameters of the target monitoring area; and selecting key monitoring targets with abnormal health risks from among the target workers based on the multi-dimensional health feature vectors extracted from multi-source occupational health data and the scene risk feature assessment results generated from the quantified data quality index of the target abnormal data items in the target monitoring area. This invention achieves accurate screening of workers with abnormal health risks by combining multi-dimensional health feature vectors extracted from multi-source occupational health data with scene risk feature assessment results generated from the quantified data quality index of the target abnormal data items in the target monitoring area.

Description

Technical Field

[0001] This invention relates to the technical field of occupational health monitoring, and in particular to an intelligent method and system for multi-source data acquisition in occupational health monitoring. Background Technology

[0002] With the continuous improvement of standardized requirements for occupational health management, accurate health risk identification and screening of key monitoring targets in scenarios such as workshops of industrial and mining enterprises and key occupational disease prevention and control areas have become core research directions in the field of occupational health. However, in related technologies, traditional occupational health monitoring methods often rely on single detection items (such as only dust concentration and noise intensity) to assess risks, or on manually compiling scattered physical examination data and environmental monitoring data. This results in one-sided and fragmented health risk assessments, making it difficult to locate core high-risk work areas and achieve full-cycle tracking and accurate assessment of workers' health status, thus leading to low accuracy in identifying key monitoring targets.

[0003] Specifically, existing technologies suffer from the following core shortcomings: First, the one-sidedness of analysis based on a single data source / single indicator, over-reliance on a particular type of testing item, and inability to capture early signs of health deterioration, hidden occupational exposure, and other risks; second, the heterogeneity of multi-source data leads to integration difficulties, with inconsistent data formats and standards generated by different institutions and equipment, resulting in low differentiation between valid and abnormal data and easily distorted assessment results; third, the separation of individual health information from work scenario information, without combining job exposure, environmental risk, and other scenario information for collaborative analysis, making it difficult to accurately pinpoint the root cause of health abnormalities; fourth, the lack of analysis of individual spatiotemporal patterns and group correlations, relying on single test results to judge risks, and inability to identify gradual health deterioration or group health hazards; fifth, the lack of a closed-loop system for early warning response and intervention, only completing data aggregation and analysis without forming a standardized early warning and intervention mechanism, resulting in low response efficiency; and sixth, insufficient in-depth data mining and prediction capabilities, making it difficult to support forward-looking prevention and control decisions.

[0004] Currently, no effective solutions have been proposed to address the problems in occupational health monitoring, such as one-sided health risk assessment, difficulty in integrating multi-source data, and low accuracy in identifying key monitoring targets. Summary of the Invention

[0005] The purpose of this section is to outline some aspects of embodiments of the present invention and to briefly describe some preferred embodiments. Simplifications or omissions may be made in this section, as well as in the abstract and title of this application, to avoid obscuring the purpose of these documents; however, such simplifications or omissions should not be construed as limiting the scope of the invention.

[0006] In view of the problems existing in the prior art, the present invention is proposed.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a method for intelligent acquisition of multi-source data for occupational health monitoring, characterized by comprising the following steps: Collect multi-source occupational health data within the target monitoring area, including worker physical examination datasets, work environment monitoring datasets, and equipment operation datasets. Feature extraction is performed on the multi-source occupational health data to obtain a multi-dimensional health feature vector corresponding to the target worker; Anomaly detection is performed on the multi-source occupational health data to locate target abnormal data items; Based on the duration of the anomalies in the target anomaly data items and the industry type parameters of the target monitoring area, a data quality assessment model is constructed to obtain a data quality index. By combining the data quality index and the data anomaly coefficients for each time period, a scenario risk characteristic assessment result is generated; Based on the multi-dimensional health feature vector and scenario risk feature assessment results, key monitoring targets with abnormal health risks are selected from the target workers; In response to the identification of key monitoring targets, retrieve the historical health record data of those targets; Based on the historical health record data, the frequency of on-the-job stay and the fluctuation range of health indicators of key monitoring subjects within a preset time period are statistically analyzed. Based on the frequency of job stays and the fluctuation range of health indicators, a spatiotemporal health trend visualization model is constructed. Health trajectory information is obtained by reconstructing the health monitoring data of key monitoring subjects. The health trajectory information is matched with the health trajectories of known high-risk workers to construct a group health association network; Based on the aforementioned spatiotemporal health trend visualization model and population health association network, the system outputs target abnormal health judgment conclusions for key monitoring subjects.

[0008] As a preferred embodiment of the intelligent acquisition method for multi-source data in occupational health monitoring according to the present invention, the step of constructing the data quality assessment model includes: Determine the data type of the target abnormal data item and assign a corresponding data type weight; Introduce a preset time decay factor; The data quality index is calculated based on the duration of the anomaly, data type weight, industry type parameter, and time decay factor.

[0009] As a preferred embodiment of the occupational health monitoring multi-source data intelligent acquisition method of the present invention, the step of assigning corresponding data type weights includes: Preset initial classification weights for different data types; construct a weight optimization objective function based on the initial classification weights and preset data quality assessment indicators; collect historical data of the target monitoring area as training samples; The training samples are input into the weight optimization objective function for iterative optimization to obtain the optimized classification weights; according to the data type of the target abnormal data item, the corresponding optimized classification weight is matched as the data type weight.

[0010] As a preferred embodiment of the intelligent acquisition method for multi-source data in occupational health monitoring according to the present invention, the step of generating scenario risk characteristic assessment results includes: The multi-source occupational health data is divided into multiple data subsets according to a preset time period; The number of abnormal data records and the total number of records in each data subset are counted, and the data anomaly coefficient for each time period is calculated. The data quality index and the data anomaly coefficient for each time period are weighted and fused to obtain the scenario risk characteristic assessment result.

[0011] As a preferred embodiment of the intelligent acquisition method for multi-source data in occupational health monitoring according to the present invention, wherein: when the abnormal health judgment conclusion of the target indicates that the key monitoring object is a high-risk person, the following steps are performed: Identify core health risk data points from the multi-source occupational health data; Based on a preset threshold for the number of associated data entries, extract associated health risk data groups containing the core health risk data points from the multi-source occupational health data; Based on the associated health risk data set, a traceable health risk evidence set is generated.

[0012] As a preferred embodiment of the intelligent acquisition method for multi-source data in occupational health monitoring according to the present invention, after generating the traceable health risk evidence set, the method further includes: Feature analysis is performed on the traceable health risk evidence set to extract abnormal health risk parameters of key monitoring targets; Retrieve historical health risk event records for the target monitoring area; The abnormal health risk parameters are compared and analyzed with the historical health risk event log to generate a graded risk warning score. Based on the aforementioned graded risk warning scores, corresponding health intervention and treatment plans are matched; Based on the aforementioned graded risk warning scores and health intervention and treatment plans, a standardized occupational health risk monitoring report is generated.

[0013] As a preferred embodiment of the intelligent acquisition method for multi-source data of occupational health monitoring described in this invention, the multi-dimensional health feature vector includes three core feature dimensions: physiological indicator features, occupational exposure features, and health trend features. The step of feature extraction from multi-source occupational health data specifically includes: calculating an abnormal health risk assessment value based on the physiological indicator features, occupational exposure features, and health trend features; Based on the industry type of the target monitoring area, assign differentiated dynamic weights to the health risk anomaly assessment value and the scenario risk characteristic assessment result; The results of the abnormal health risk assessment and the scenario risk characteristic assessment are weighted and fused to obtain the identification results of key monitoring targets; The results of the identification of key monitoring targets are used to indicate key monitoring targets with abnormal health risks selected from the target workers.

[0014] The data acquisition system applied to the above-mentioned intelligent acquisition method for multi-source occupational health monitoring includes: The data acquisition unit is used to collect multi-source occupational health data within the target monitoring area. The multi-source occupational health data includes worker physical examination datasets, work environment monitoring datasets, and equipment operation datasets. The feature extraction unit is used to extract features from the multi-source occupational health data to obtain a multi-dimensional health feature vector corresponding to the target worker. An anomaly detection unit is used to detect anomalies in the multi-source occupational health data, locate target abnormal data items, and construct a data quality assessment model based on the duration of the anomaly of the target abnormal data items and the industry type parameters of the target monitoring area to obtain a data quality index. The scenario assessment unit is used to combine the data quality index and the data anomaly coefficients for each time period to generate scenario risk characteristic assessment results. The object screening unit is used to screen key monitoring objects with abnormal health risks from the target workers based on the multi-dimensional health feature vector and scenario risk feature assessment results. The risk assessment unit, in response to the identification of a key monitoring target, retrieves the target's historical health record data. Based on the historical health record data, it statistically analyzes the frequency of the key monitoring target's on-the-job stay and the fluctuation range of health indicators within a preset time period, constructs a spatiotemporal health trend visualization model, reconstructs the health monitoring data of the key monitoring target to obtain health trajectory information, matches the health trajectory information with the health trajectories of known high-risk workers to construct a group health association network, and outputs a target abnormal health assessment conclusion for the key monitoring target based on the spatiotemporal health trend visualization model and the group health association network. The data acquisition unit, feature extraction unit, anomaly detection unit, scenario evaluation unit, target screening unit, and risk assessment unit are sequentially connected and collaboratively complete the entire process of intelligent data acquisition and analysis from multi-source occupational health data collection to the assessment of abnormal health of key monitoring targets.

[0015] The present invention also discloses a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described intelligent acquisition method for multi-source data in occupational health monitoring.

[0016] The present invention also discloses a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described method for intelligent acquisition of multi-source data for occupational health monitoring.

[0017] The beneficial effects of this invention are: 1. The method of this invention uses multi-dimensional health feature vectors extracted based on multi-source occupational health data, combined with the data quality index generated by quantifying the abnormal data items in the target monitoring area, to generate scenario risk feature assessment results. These two become the two core bases for screening key monitoring targets. Through the synergistic analysis of the two, the accurate screening of workers with abnormal health risks is achieved. 2. The method of this invention breaks away from the reliance of traditional occupational health monitoring on manual data collection and single indicator evaluation, and constructs a multi-dimensional correlation assessment model of "people-data-scenario", which breaks down the correlation barrier between individual abnormal health signals and scenario risks.

[0018] 3. This invention effectively avoids the problems of difficulty in locating core high-risk areas and missed identification of high-risk personnel caused by fragmented data and isolated analysis, significantly reduces the false alarm rate of risk assessment, and fundamentally solves a series of core technical problems in traditional occupational health monitoring, such as one-sided analysis of single data sources, difficulty in integrating heterogeneous multi-source data, low differentiation between effective and abnormal data, and separation of individual and scenario information. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein: Figure 1 This is a schematic diagram of the process for an intelligent multi-source data acquisition method for occupational health monitoring proposed in this invention; Figure 2 This is a schematic diagram of the overall structure of an intelligent multi-source data acquisition system for occupational health monitoring proposed in this invention. Detailed Implementation

[0020] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0021] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0022] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0023] Reference Figure 1 As an embodiment of the present invention, a method for intelligent acquisition of multi-source data for occupational health monitoring is provided, the method comprising the following steps: Step 1: Collect multi-source occupational health data within the target monitoring area. The multi-source occupational health data includes worker physical examination datasets, work environment monitoring datasets, and equipment operation datasets.

[0024] Specifically, the multi-source occupational health data of the target monitoring area is collected by dedicated multi-source sensing terminals. These terminals are suitable for different monitoring scenarios such as workshops of industrial and mining enterprises and key occupational disease prevention and control areas. They include intelligent physical examination terminal arrays, work environment monitoring stations with integrated dust / noise dual detection modules, and production equipment with built-in operating parameter acquisition units. At the same time, they can be directly connected to the enterprise health management system to realize the linkage retrieval of existing data.

[0025] Among the various data collected by the multi-source sensing terminals, the worker health examination dataset is the core carrier for subsequent analysis. It forms a full occupational health record of workers from employment to departure, collected once a day. Meanwhile, heterogeneous data such as real-time monitoring data of the work environment and operating parameters of production equipment serve as important supplementary dimensions for individual health risk assessment, together forming a complete multi-source occupational health data system.

[0026] Step 2: Extract features from the multi-source occupational health data to obtain a multi-dimensional health feature vector corresponding to the target worker; wherein, by integrating heterogeneous data collected by multi-source sensing terminals within the target monitoring area, a three-dimensional health profile of the target worker is constructed.

[0027] Specifically, the process of extracting effective health features from multi-source heterogeneous data requires a comprehensive analysis of workers' health status: first, feature mining of physiological indicators is carried out on the physical examination dataset to extract core physiological indicators such as lung capacity, blood pressure, and blood routine; at the same time, the work environment monitoring dataset is linked to simultaneously analyze work exposure-related features such as dust exposure concentration and noise exposure duration. Based on the time-series data of workers' annual physical examinations, a health trend analysis algorithm is used for in-depth processing to obtain trend characteristics that reflect health changes, such as the annual decline rate of vital capacity and the amplitude of blood pressure fluctuations. Then, combined with the job operation information of the equipment operation dataset, job exposure parameters such as the average dust concentration and the proportion of actual working time for each job position are calculated. By aligning and correlating feature vectors extracted from different data sources across spatiotemporal dimensions, and using a weighted fusion model to mitigate assessment biases from single data sources—such as random deviations from single physical examination indicators or partial gaps in environmental monitoring data—a composite feature set integrating physiological indicators, occupational exposure, and health trends is ultimately formed. This feature set accurately characterizes the health status of individual workers and provides a structured data foundation for the subsequent selection of key monitoring targets. This multi-dimensional feature extraction method achieves comprehensive coverage of workers' health status. Even if some data sources have missing data or partial biases, accurate assessment of health status can be completed based on features from other complete dimensions. This fundamentally solves the technical pain points of traditional monitoring, which relies solely on single testing items for risk assessment and cannot promptly capture early signs of health deterioration such as a slow decline in lung capacity or hidden occupational exposure risks such as long-term exposure to low-concentration noise.

[0028] Step 3: Perform anomaly detection on the multi-source occupational health data and locate the target abnormal data items; Based on the duration of the anomalies in the target abnormal data items and the industry type parameters of the target monitoring area, a data quality assessment model is constructed to obtain a data quality index.

[0029] Specifically, this step focuses on the quality control of multi-source occupational health data. It conducts systematic anomaly analysis on the full amount of data continuously collected within the target monitoring area, accurately identifying data anomalies caused by various factors such as data collection equipment errors, manual data entry errors, and abnormal on-site working conditions. At the same time, it quantitatively assesses the degree of anomalies in the overall data quality. The results of this quantitative assessment become the core data quality basis for subsequent scenario risk level assessment and precise selection of key monitoring targets.

[0030] The specific implementation process for data anomaly detection and target anomaly data item location is as follows: Relying on the pre-built industry standard database, which has been entered into the core reference data such as occupational disease diagnosis standards, human health index thresholds, and work environment monitoring limits of various industries, combined with the health and environmental data of workers throughout their occupational life cycle collected by multi-source sensing terminals, machine learning anomaly detection algorithms such as isolated forest and autoencoder are used to perform dimension-by-dimensional analysis on batch data, and to accurately locate abnormal records that deviate from the normal threshold and have logical contradictions from the dataset. These records are the target anomaly data items. The target abnormal data items are mainly divided into three categories: abnormal physical examination data, abnormal environmental data, and abnormal equipment data. For example, if a worker's lung capacity decreases by more than 10% for three consecutive years, it is considered abnormal physical examination data. If the dust concentration in the work area exceeds the national standard for eight consecutive hours, it is classified as abnormal environmental data. The testing process needs to balance accuracy and real-time performance. On the one hand, it should avoid misjudging individual physiological differences of workers as data anomalies. On the other hand, it should control the processing time of a single batch of data to within 1 minute to meet the actual application needs of on-site occupational health real-time monitoring.

[0031] After locating the target abnormal data item, the duration of its abnormal state is accurately statistically analyzed. Specifically, a time-series tracking algorithm is used to track the changes in the indicators of each abnormal data item throughout the entire cycle. When the indicator value of an abnormal data item exceeds a preset threshold (such as dust concentration exceeding the national standard by 10%), the record of the abnormal start time is automatically triggered. When the indicator value falls back to the threshold range or the corresponding work area stops production, the record of the abnormal end time is recorded. The difference between the two time points is the actual duration of the abnormal data item. Taking the dust concentration exceeding the standard in a mining workshop as an example, the algorithm can accurately count the duration of continuous exceedance for 8 hours. At the same time, a short-term interference filtering threshold of 1 hour is set. Instantaneous abnormalities with a duration of less than 1 hour are judged as invalid abnormalities, thereby eliminating misjudgments caused by instantaneous fluctuations in indicators due to brief equipment failures, and ensuring the continuity and accuracy of the abnormal duration statistics.

[0032] The industry type parameter mentioned here is a quantitative indicator obtained by combining the industry attributes of the target monitoring area, the actual on-site work intensity, and historical health risk statistics through predefined or machine learning training. Its core function is to define the normal fluctuation boundary of data indicators within the industry and provide an industry-appropriate basis for judging data anomalies. This parameter needs to be set differently based on the industry risk level. It is obtained by relying on multi-dimensional data statistics: First, the occurrence rate of abnormal data in different industries is calculated to reflect the proportion of natural data anomalies under normal production conditions in the industry. For example, the occurrence rate of abnormal data in the mining industry is 15%, which means that about 15 out of every 100 data points will fluctuate naturally due to the complex working environment. Then, the maximum allowable duration of normal anomalies in each industry is calculated. At the same time, the work intensity coefficient is dynamically adjusted in combination with the current actual work intensity of the industry. Finally, multiple indicators such as the occurrence rate of abnormal data, the threshold for the duration of anomalies, and the work intensity coefficient are integrated to determine the industry type parameter.

[0033] After completing the statistics on the duration of anomalies, the weighting of data types, the determination of industry type parameters, and the selection of time decay factors, the data quality index is calculated by integrating the above indicators through a quantitative model. This index is a direct quantitative representation of the overall data quality anomaly degree in the target monitoring area. As a core indicator for determining the validity of data in occupational health monitoring, the data quality index can accurately distinguish between valid anomalies that truly reflect workers' health risks or on-site environmental problems, and invalid anomalies caused by collection errors or data entry errors, by combining the severity of anomalies reflected by the duration of anomalies and the industry adaptability reflected by the industry type parameters. This improves the accuracy of subsequent key monitoring target selection work from the data source.

[0034] Step 4: Combine the data quality index and the data anomaly coefficients for each time period to generate a scenario risk characteristic assessment result; specifically, the data quality index is fused with the data anomaly coefficients for each time period corresponding to all target abnormal data items in the target monitoring area to generate a scenario risk characteristic assessment result, providing scenario-based contextual information for the subsequent selection of key monitoring targets.

[0035] Based on the multi-dimensional health feature vector and scenario risk feature assessment results, key monitoring targets with abnormal health risks are selected from the target workers. Specifically, the multi-dimensional health feature vector of each worker is dynamically weighted and fused with the scenario risk feature assessment results of the worker's work scenario to generate a corresponding individual health risk comprehensive index. Using a preset health risk threshold as the judgment standard, the individual health risk comprehensive index is compared with the threshold to accurately select key monitoring targets with abnormal health risks from all target workers. The health status of these key monitoring targets will show unexpected and non-gradual deterioration characteristics. Their work scenarios are often accompanied by high-risk work environments or have already shown signs of group health hazards. Taking a frontline worker in the mining industry as an example, their multi-dimensional health feature vector shows that their lung capacity has decreased by 12% for three consecutive years, and their noise exposure time exceeds the industry average by 50%, which is a typical case of abnormal individual health characteristics. At the same time, the scene risk feature assessment result of their workplace is high risk. After dynamic weighted fusion calculation, the worker's comprehensive health risk index is far higher than the preset threshold, and they will be directly marked as a key monitoring target and included in the subsequent special health management and tracking scope. This screening logic realizes a deep synergy between individual health characteristics and scene risk, effectively making up for the single deficiency of traditional assessment that only looks at individual indicators or only looks at scene environment, and greatly improving the accuracy and reliability of the selection of key monitoring targets.

[0036] Step 5: In response to the identification of key monitoring targets, retrieve the historical health record data of the target; the historical health record data includes the worker's annual physical examination records, job change records, health intervention records, etc., which are used to support subsequent spatiotemporal health trend analysis and the construction of a group health correlation network.

[0037] Based on the historical health record data, the frequency of job stays and the fluctuation range of health indicators of key monitoring subjects within a preset period are statistically analyzed. Specifically, the frequency of job stays of key monitoring subjects within a preset period (such as the past 3 years) is statistically analyzed (e.g., 1000 hours of cumulative stay in high dust positions), and the fluctuation range of health indicators (e.g., 12% annual decline in vital capacity and 15% fluctuation range in blood pressure) is calculated.

[0038] Based on the frequency of job stays and the fluctuation range of health indicators, a spatiotemporal health trend visualization model is constructed. This model uses time as the horizontal axis and job stay frequency and health indicators as the vertical axis to visualize the changing trend of the health status of key monitoring objects over time and job changes, providing an intuitive basis for health risk prediction.

[0039] Step Six: Reconstruct the health monitoring data of key monitoring subjects to obtain health trajectory information; match the health trajectory information with the health trajectories of known high-risk workers to construct a group health association network; specifically, by comparing the health trajectory information of key monitoring subjects with those of known high-risk workers using a health trajectory matching algorithm (such as DTW dynamic time warping), identify groups with highly overlapping health trajectories, and construct a group health association network for the purpose of identifying potential group health risks.

[0040] Based on the aforementioned spatiotemporal health trend visualization model and population health correlation network, the system outputs target abnormal health judgment conclusions for key monitoring subjects. These conclusions include the health risk level, potential occupational disease types, and recommended intervention measures for the key monitoring subjects, providing precise support for occupational health management decisions.

[0041] Through the above steps, the multi-dimensional health feature vector extracted based on multi-source occupational health data, combined with the data quality index generated by quantifying the abnormal data items in the target monitoring area, becomes the two core bases for screening key monitoring targets. Through the collaborative analysis of the two, the accurate screening of workers with abnormal health risks is achieved. This process completely eliminates the reliance of traditional occupational health monitoring on manual data collection and single-indicator assessment, and builds a multi-dimensional correlation assessment model of "people-data-scenario", breaking down the correlation barrier between individual abnormal health signals and scenario risks. It effectively avoids problems such as difficulty in locating core high-risk areas and missed identification of high-risk personnel caused by fragmented data and isolated analysis, significantly reduces the false alarm rate of risk assessment, and fundamentally solves a series of core technical problems in traditional occupational health monitoring, such as one-sided analysis of single data sources, difficulty in integrating heterogeneous multi-source data, low differentiation between effective and abnormal data, and separation of individual and scenario information.

[0042] In one embodiment, the step of constructing the data quality assessment model includes: Determine the data type of the target abnormal data item and assign a corresponding data type weight; Introduce a preset time decay factor; The data quality index is calculated based on the duration of the anomaly, data type weight, industry type parameter, and time decay factor.

[0043] The data quality assessment model in this embodiment is not a simple parameter aggregation, but a dynamic assessment system that balances risk priority, real-time performance, and industry adaptability. The core logic of the model revolves around the "essential impact of data anomalies": First, different anomalies are weighted by data type to determine their risk level; for example, anomalies in physical examination data that directly reflect workers' health are given higher weight, while instantaneous fluctuations in environmental monitoring are given lower weight. Next, a time decay factor is used to calibrate the timeliness of anomalies, making recent anomalies more prominent in the assessment results and preventing historical anomalies from dominating the assessment conclusions for a long time. Finally, industry type parameters are used as scenario correction factors to match the assessment standards with the industry's production characteristics, ultimately forming a quantitative index that accurately characterizes the degree of data quality anomalies. This design makes the data quality index no longer an isolated value, but an assessment tool that can dynamically respond to changes in the scenario and accurately reflect the authenticity and validity of the data.

[0044] One method for quantifying the aforementioned data quality index is represented by the following mathematical model: ; In the above model, Used to represent a data quality index. This represents the number of currently identified target anomalous data items. Used to indicate the first Data type weights for each target anomaly data item; Used to indicate the first The normalized duration of anomalies corresponding to each target anomaly data item ( = / , For the first The duration of the anomaly corresponding to each target anomaly data item. (The preset maximum duration threshold for anomalies). Used to represent industry type weight (i.e., the industry type parameter mentioned above). Used to represent the time decay factor (used to suppress the excessive influence of historical outlier data).

[0045] Through the above embodiments, firstly, the risk priority of different abnormal data items is distinguished by data type weights, avoiding assessment bias caused by homogenizing all abnormal data; secondly, the time decay factor, through a non-linear decay model, ensures that the impact of recently occurring abnormal data on the index is significantly higher than that of historical data, ensuring that the data quality assessment results reflect the current real-time status; finally, industry-specific parameters are combined to achieve industry-adaptation of risk assessment, making the data quality anomaly assessment standards of different industries more aligned with actual production scenarios. The synergistic effect of these three elements enables the data quality index to have dynamic weight adjustment capabilities, highlighting key data anomalies in high-risk industries while filtering out occasional data fluctuations in low-risk scenarios. Ultimately, this improves the accuracy of selecting key monitoring targets, while reducing redundant data interference through the time-series decay mechanism and optimizing the computational resource allocation efficiency of the health monitoring system.

[0046] In one embodiment, the step of assigning weights to the corresponding data types includes: Preset initial classification weights for different data types; construct a weight optimization objective function based on the initial classification weights and preset data quality assessment indicators; collect historical data of the target monitoring area as training samples; The training samples are input into the weight optimization objective function for iterative optimization to obtain the optimized classification weights; according to the data type of the target abnormal data item, the corresponding optimized classification weight is matched as the data type weight.

[0047] The core input for weight optimization is real training data from the target monitoring area. This data needs to be multi-dimensionally labeled, not only clarifying the type of each abnormal record (e.g., abnormal physical examination data, abnormal environmental data), but also labeling its corresponding degree of abnormality and the actual deviation it causes to the overall quality assessment, providing accurate feedback for weight iteration. The optimization process aims to minimize quality assessment deviation, using machine learning algorithms such as gradient descent and Adam to iteratively adjust the initial weights. The algorithm automatically identifies the actual impact of different data types on the assessment results, gradually increasing the weights of high-impact types and decreasing the weights of low-impact types until the model's output weights highly match the actual situation. After optimization, the optimized classification weights for each data type are embedded in the system. When abnormal data of the corresponding type is detected, the optimized weights are directly called, ensuring the accuracy of data quality assessment and scenario adaptability.

[0048] The starting point for weight optimization is to predefine the initial classification weights for each data type. For example, the initial weight for abnormal physical examination data, which directly reflects the health status of workers, is set to 0.3, and the initial weight for abnormal environmental monitoring data is set to 0.2, serving as the basis for model learning. Subsequently, historical occupational health data from the target monitoring area is collected as training samples. Each data point is labeled with its true type (e.g., "abnormal physical examination data - decreased lung capacity") and the corresponding quality assessment result (e.g., "causing a 15% risk assessment bias"), forming a fully labeled training set. Combining core quality assessment indicators (e.g., data accuracy, risk assessment bias rate), the initial classification weights are correlated with the assessment indicators. Essentially, this constructs an objective function of "weighted data impact × quality loss," used to quantify the rationality of the weight settings and provide direction for subsequent optimization.

[0049] After constructing the objective function, the labeled training data is input into the model in batches to initiate iterative optimization with the goal of minimizing quality loss. During optimization, the algorithm automatically adjusts the weights of each data type. For example, if the actual impact of environmental data anomalies on risk assessment is found to be higher than initially expected, its weight will be increased; conversely, it will be decreased, until the weights output by the model highly match the actual situation of the scenario. After optimization, the optimized classification weights of each data type will be embedded in the system. When anomaly data of the corresponding type is detected, the optimized weights will be directly applied to ensure the accuracy of data quality assessment.

[0050] This data-driven weight optimization mechanism completely breaks free from the limitations of traditional experience-based weight settings, achieving scenario-based adaptive weighting for different data types. Through iterative correction using historical data, the optimized weights accurately align with the industry characteristics and data distribution features of the target monitoring area, avoiding quality assessment biases caused by fixed weights. Ultimately, this mechanism effectively improves the accuracy of data quality index calculation, providing more reliable data support for the subsequent selection of key monitoring targets and fundamentally solving the assessment distortion problem caused by the heterogeneity of multi-source data.

[0051] In one embodiment, the step of generating the scenario risk characteristic assessment result includes: The multi-source occupational health data is divided into multiple data subsets according to a preset time period; for example, day shift, night shift, weekend, etc., and each dataset corresponds to a continuous work cycle.

[0052] The number of abnormal data records and the total number of records in each data subset are counted, and the data anomaly coefficient for each time period is calculated. The data quality index and the data anomaly coefficient for each time period are weighted and fused to obtain the scenario risk characteristic assessment result.

[0053] Specifically, the statistical calculation of the data anomaly coefficient needs to be aligned with the time characteristics of the work scenario. For example, dividing the work into 8-hour shifts and 24-hour days, the number of records of abnormal physical examination data, environmental data, and equipment data within each time period is counted, and then divided by the total number of records in that time period to obtain the anomaly coefficient for that time period. The data anomaly coefficient is a quantitative indicator of the frequency of data anomalies within the scenario. The higher the coefficient, the more prominent the data quality problem within that time period, and the higher the risk level of the scenario. The generation of the scenario risk characteristic assessment results is not a simple coefficient summation, but rather a deep integration of the data anomaly coefficient and the data quality index through weighted fusion, multiplication, or composite functions to comprehensively reflect the overall degree of data quality anomalies within the scenario. Taking a chemical plant workshop as an example, the number of abnormal records during the day shift can be calculated separately ( ) Total number of records during the day shift ( The percentage of ) and the number of abnormal records during night shifts ( ) in the total number of night shift records ( The proportion in ) and then the two are combined with the data quality index ( The risk characteristics of the scenario are then weighted and fused to obtain the assessment results. The quantitative calculation of the scenario risk characteristics assessment results can be represented by the following model: ; in, Used to represent the results of scenario risk characteristic assessment; , These are used to represent the weight values ​​corresponding to the day shift and night shift, respectively; For data quality index; , These represent the number of abnormal records during the day shift and the total number of records during the day shift, respectively. , These represent the number of abnormal records during the night shift and the total number of records during the night shift, respectively. , It can be dynamically adjusted based on industry workload, historical risk data, and other factors. Through the above embodiments, focusing on scenario risk characteristic assessment, by quantifying the spatiotemporal distribution of data anomalies and combining it with data quality, the system ultimately achieves accurate identification and assessment of scenario risks, thereby effectively improving the accuracy of key monitoring target selection and solving the assessment distortion problem caused by the heterogeneity of multi-source data.

[0054] In one embodiment, if the target abnormal health assessment conclusion indicates that the key monitoring target is a high-risk individual, the following steps are performed: Identify core health risk data points from the multi-source occupational health data; Based on a preset threshold for the number of associated data entries, extract associated health risk data groups containing the core health risk data points from the multi-source occupational health data; Based on the associated health risk data set, a traceable health risk evidence set is generated.

[0055] Specifically, the identification of core health risk data points is based on "core manifestations of abnormal health risks," which are key data records that directly reflect high-risk health states. Examples include continuously exceeding lung function test standards, peak exposure data strongly correlated with occupational diseases, or key time points recording sudden changes in health indicators. The threshold for the amount of associated data can be flexibly configured according to the industry's risk level. For example, in the chemical industry, the preset threshold is 20 records before and after the core data point. The system will extract physical examination records, environmental monitoring data, and job operation records within this range according to the time dimension, forming a set of associated health risk data that can completely reconstruct the development trajectory of health risks.

[0056] The generation of a traceable health risk evidence set requires standardized data processing and feature annotation. During processing, data is ordered chronologically by collection time. Annotations cover the risk characteristics of core data points, the source and time of each data item, and include job information and work background details for key monitoring targets. This evidence set can be directly used for internal traceability analysis in enterprise occupational health management and can also serve as a compliance basis for occupational health regulatory departments. Its complete chain-like data structure ensures the authenticity of health risk assessments and provides verifiable core support for subsequent health interventions and liability determination.

[0057] In some application scenarios, after constructing the traceable health risk evidence set, this solution also sets up a supporting implementation process for health risk classification, early warning, and intervention. After generating the traceable health risk evidence set, it also includes: Feature analysis is performed on the traceable health risk evidence set to extract abnormal health risk parameters of key monitoring targets; Retrieve historical health risk event records for the target monitoring area; The abnormal health risk parameters are compared and analyzed with the historical health risk event log to generate a graded risk warning score. Based on the aforementioned graded risk warning scores, corresponding health intervention and treatment plans are matched; Based on the aforementioned graded risk warning scores and health intervention and treatment plans, a standardized occupational health risk monitoring report is generated.

[0058] Specifically, the analysis of abnormal health risk parameters is based on the "core quantitative representation of health risk," including quantitative indicators such as the annual decline rate of lung function indicators, the multiple by which noise exposure duration exceeds the standard, and the similarity to the health trajectory of historically high-risk individuals. The scoring mechanism compares the current abnormal parameters with historical averages and industry thresholds. For example, a Level 1 warning is triggered when the annual decline rate of lung function exceeds 1.5 times the industry average and the similarity to the health trajectory is higher than 60%; a Level 2 warning is triggered when four or more health indicators are abnormal. This scoring mechanism uses a multi-dimensional weighted calculation model, comprehensively considering factors such as the severity of abnormal parameters, the correlation with historical risks, and the intensity of industry work, to ensure the accuracy and scenario adaptability of the graded risk warning scores. The health intervention and treatment plan is precisely matched with the graded risk warning scores. For example, a Level 1 warning corresponds to immediately initiating a special health verification and notifying the occupational health management department to intervene; a Level 2 warning corresponds to continuously tracking health indicators and adjusting protective measures for work positions; and a Level 3 warning corresponds to inclusion in routine health monitoring and periodic review. The standardized occupational health risk monitoring report integrates core risk data points, related health risk data groups, abnormal parameter analysis results, intervention and treatment recommendations, and other content, and is output in a unified format. This report can be directly used for internal occupational health management decisions within enterprises and can also serve as a compliance basis for regulatory department verification. It effectively solves the problems of vague warning classification, delayed intervention response, and non-standard report formats in traditional monitoring, improving the efficiency and accuracy of health risk control.

[0059] In one embodiment, the multi-dimensional health feature vector includes three core feature dimensions: physiological indicator features, job exposure features, and health trend features. Specifically, the extraction of multi-dimensional health characteristics revolves around "the full-cycle representation of workers' health status": First, physiological indicator characteristics are extracted by constructing an individual health baseline through quantitative analysis of core physiological indicators such as lung function, blood routine, and blood pressure; second, occupational exposure characteristics are extracted by characterizing the impact of occupational scenarios on health by combining environmental data such as dust concentration, noise duration, and exposure to chemical toxins; and third, health trend characteristics are extracted by capturing early health deterioration signals such as the rate of decline in vital capacity and the amplitude of blood pressure fluctuations through time-series analysis of annual physical examination data.

[0060] Compared to traditional occupational health monitoring methods that rely on a single test item (such as dust concentration only), this solution integrates multi-dimensional features to accurately capture risk signs that are difficult to identify with traditional technologies, such as early health deterioration and hidden occupational exposure, effectively improving the comprehensiveness and foresight of health risk assessment.

[0061] The step of feature extraction from multi-source occupational health data specifically includes: calculating an abnormal health risk assessment value based on the physiological indicator features, occupational exposure features, and health trend features; Based on the industry type of the target monitoring area, differentiated dynamic weights are assigned to the health risk anomaly assessment value and the scenario risk characteristic assessment results. Specifically, this solution can achieve accurate identification of key monitoring targets, with the core logic revolving around a collaborative assessment of "individual health anomalies + scenario risk adaptation." The calculation of the health risk anomaly assessment value uses multi-dimensional health characteristics as the core input. By weighted fusion of physiological indicator characteristics, occupational exposure characteristics, and health trend characteristics, a quantitative health risk anomaly assessment value is generated. The calculation formula is as follows: ; The weight constraints satisfy: 1 ​​= + + This ensures that the contribution of each dimension of characteristics to the evaluation results is reasonably allocated.

[0062] In the above formula, This indicates an abnormal health risk assessment value; It represents physiological indicator characteristics and is quantified through indicators such as lung function and blood routine. This information represents the characteristics of the work exposure, quantified using data such as dust concentration and noise duration. It represents health trend characteristics and is quantified by time-series indicators such as the annual decline rate of vital capacity; These are the weights of features in each dimension, which can be dynamically adjusted according to industry characteristics. For example, in the chemical industry, the weight of operational exposure features can be increased to more accurately reflect the health risks in chemical scenarios.

[0063] The results of the abnormal health risk assessment and the scenario risk characteristic assessment are weighted and fused to obtain the identification results of key monitoring targets; Furthermore, based on the industry type of the target monitoring area, differentiated dynamic weights are assigned to the health risk anomaly assessment value and the scenario risk characteristic assessment result. The key monitoring object identification result is obtained through weighted fusion calculation, realizing the collaborative assessment of individual health anomalies and scenario risks.

[0064] The calculation process for the identification results of key monitoring targets can be expressed by the following formula: ; in, This indicates the results of the identification of key monitoring targets; , These are the weights corresponding to the abnormal health risk assessment value and the scenario risk characteristic assessment result, respectively, which can be dynamically adjusted according to industry risk level, work intensity, etc. The results of the scenario risk characteristic assessment are generated by fusing the data quality index and the data anomaly coefficient.

[0065] The identification results of key monitoring targets are used to indicate key monitoring targets with abnormal health risks selected from the target workers. Through the above embodiments, this solution takes into account both the individual health status of workers and the risks of the work environment, ultimately achieving accurate identification of key monitoring targets with abnormal health risks. This effectively solves the core technical problems of traditional occupational health monitoring, such as the one-sidedness of a single data source and the separation between individuals and scenarios, and significantly improves the accuracy and reliability of health risk assessment.

[0066] In addition, refer to Figure 2 In addition to the above-mentioned intelligent acquisition method for multi-source data in occupational health monitoring, this invention also discloses an intelligent acquisition system for multi-source data in occupational health monitoring, which includes: The data acquisition unit is used to collect multi-source occupational health data within the target monitoring area. The multi-source occupational health data includes worker physical examination datasets, work environment monitoring datasets, and equipment operation datasets.

[0067] The feature extraction unit is used to extract features from the multi-source occupational health data to obtain a multi-dimensional health feature vector corresponding to the target worker.

[0068] An anomaly detection unit is used to detect anomalies in the multi-source occupational health data, locate target abnormal data items, and construct a data quality assessment model based on the duration of the anomaly of the target abnormal data items and the industry type parameters of the target monitoring area to obtain a data quality index.

[0069] The scenario assessment unit is used to combine the data quality index and the data anomaly coefficients for each time period to generate scenario risk characteristic assessment results.

[0070] The object screening unit is used to screen key monitoring objects with abnormal health risks from the target workers based on the multi-dimensional health feature vector and scenario risk feature assessment results.

[0071] The risk assessment unit, in response to the identification of a key monitoring target, retrieves the target's historical health record data. Based on the historical health record data, it statistically analyzes the frequency of the key monitoring target's on-the-job stay and the fluctuation range of health indicators within a preset time period, constructs a spatiotemporal health trend visualization model, reconstructs the health monitoring data of the key monitoring target to obtain health trajectory information, matches the health trajectory information with the health trajectories of known high-risk workers to construct a group health association network, and outputs a target abnormal health assessment conclusion for the key monitoring target based on the spatiotemporal health trend visualization model and the group health association network. The data acquisition unit, feature extraction unit, anomaly detection unit, scenario evaluation unit, target screening unit, and risk assessment unit are sequentially connected and collaboratively complete the entire process of intelligent data acquisition and analysis from multi-source occupational health data collection to the assessment of abnormal health of key monitoring targets.

[0072] This embodiment also provides a computer device applicable to a method for intelligent acquisition of multi-source data for occupational health monitoring, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the method for intelligent acquisition of multi-source data for occupational health monitoring as proposed in the above embodiment.

[0073] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0074] This embodiment also provides a storage medium storing a computer program. When executed by a processor, the program implements the intelligent acquisition method for multi-source occupational health monitoring as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0075] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for intelligent acquisition of multi-source data for occupational health monitoring, characterized in that, Includes the following steps: Collect multi-source occupational health data within the target monitoring area, including worker physical examination datasets, work environment monitoring datasets, and equipment operation datasets. Feature extraction is performed on the multi-source occupational health data to obtain a multi-dimensional health feature vector corresponding to the target worker; Anomaly detection is performed on the multi-source occupational health data to locate target abnormal data items; Based on the duration of the anomalies in the target anomaly data items and the industry type parameters of the target monitoring area, a data quality assessment model is constructed to obtain a data quality index. By combining the data quality index and the data anomaly coefficients for each time period, a scenario risk characteristic assessment result is generated; Based on the multi-dimensional health feature vector and scenario risk feature assessment results, key monitoring targets with abnormal health risks are selected from the target workers; In response to the identification of key monitoring targets, retrieve the historical health record data of those targets; Based on the historical health record data, the frequency of on-the-job stay and the fluctuation range of health indicators of key monitoring subjects within a preset time period are statistically analyzed. Based on the frequency of job stays and the fluctuation range of health indicators, a spatiotemporal health trend visualization model is constructed. Health trajectory information is obtained by reconstructing the health monitoring data of key monitoring subjects. The health trajectory information is matched with the health trajectories of known high-risk workers to construct a group health association network; Based on the aforementioned spatiotemporal health trend visualization model and population health association network, the system outputs target abnormal health judgment conclusions for key monitoring subjects.

2. The method for intelligent acquisition of multi-source data for occupational health monitoring according to claim 1, characterized in that: The steps for constructing the data quality assessment model include: Determine the data type of the target abnormal data item and assign a corresponding data type weight; Introduce a preset time decay factor; The data quality index is calculated based on the duration of the anomaly, data type weight, industry type parameter, and time decay factor.

3. The method for intelligent acquisition of multi-source data for occupational health monitoring according to claim 2, characterized in that: The step of assigning weights to the corresponding data types includes: Preset initial classification weights for different data types; construct a weight optimization objective function based on the initial classification weights and preset data quality assessment indicators; collect historical data of the target monitoring area as training samples; The training samples are input into the weight optimization objective function for iterative optimization to obtain the optimized classification weights; according to the data type of the target abnormal data item, the corresponding optimized classification weight is matched as the data type weight.

4. The method for intelligent acquisition of multi-source data for occupational health monitoring according to claim 3, characterized in that: The steps for generating scenario risk characteristic assessment results include: The multi-source occupational health data is divided into multiple data subsets according to a preset time period; The number of abnormal data records and the total number of records in each data subset are counted, and the data anomaly coefficient for each time period is calculated. The data quality index and the data anomaly coefficient for each time period are weighted and fused to obtain the scenario risk characteristic assessment result.

5. The method for intelligent acquisition of multi-source data for occupational health monitoring according to claim 1, characterized in that: If the abnormal health assessment conclusion indicates that the key monitoring target is a high-risk individual, the following steps shall be performed: Identify core health risk data points from the multi-source occupational health data; Based on a preset threshold for the number of associated data entries, extract associated health risk data groups containing the core health risk data points from the multi-source occupational health data; Based on the associated health risk data set, a traceable health risk evidence set is generated.

6. The method for intelligent acquisition of multi-source data for occupational health monitoring according to claim 5, characterized in that: After generating the traceable health risk evidence set, the process also includes: Feature analysis is performed on the traceable health risk evidence set to extract abnormal health risk parameters of key monitoring targets; Retrieve historical health risk event records for the target monitoring area; The abnormal health risk parameters are compared and analyzed with the historical health risk event log to generate a graded risk warning score. Based on the aforementioned graded risk warning scores, corresponding health intervention and treatment plans are matched; Based on the aforementioned graded risk warning scores and health intervention and treatment plans, a standardized occupational health risk monitoring report is generated.

7. The method for intelligent acquisition of multi-source data for occupational health monitoring according to claim 1, characterized in that: The multi-dimensional health feature vector includes three core feature dimensions: physiological indicator features, job exposure features, and health trend features. The step of feature extraction from multi-source occupational health data specifically includes: calculating an abnormal health risk assessment value based on the physiological indicator features, occupational exposure features, and health trend features; Based on the industry type of the target monitoring area, assign differentiated dynamic weights to the health risk anomaly assessment value and the scenario risk characteristic assessment result; The results of the abnormal health risk assessment and the scenario risk characteristic assessment are weighted and fused to obtain the identification results of key monitoring targets; The results of the identification of key monitoring targets are used to indicate key monitoring targets with abnormal health risks selected from the target workers.

8. An intelligent acquisition system for multi-source occupational health monitoring, applied to the intelligent acquisition method for multi-source occupational health monitoring as described in claims 1-8, characterized in that: The system includes: The data acquisition unit is used to collect multi-source occupational health data within the target monitoring area. The multi-source occupational health data includes worker physical examination datasets, work environment monitoring datasets, and equipment operation datasets. The feature extraction unit is used to extract features from the multi-source occupational health data to obtain a multi-dimensional health feature vector corresponding to the target worker. An anomaly detection unit is used to detect anomalies in the multi-source occupational health data, locate target abnormal data items, and construct a data quality assessment model based on the duration of the anomaly of the target abnormal data items and the industry type parameters of the target monitoring area to obtain a data quality index. The scenario assessment unit is used to combine the data quality index and the data anomaly coefficients for each time period to generate scenario risk characteristic assessment results. The object screening unit is used to screen key monitoring objects with abnormal health risks from the target workers based on the multi-dimensional health feature vector and scenario risk feature assessment results. The risk assessment unit, in response to the identification of a key monitoring target, retrieves the target's historical health record data. Based on the historical health record data, it statistically analyzes the frequency of the key monitoring target's on-the-job stay and the fluctuation range of health indicators within a preset time period, constructs a spatiotemporal health trend visualization model, reconstructs the health monitoring data of the key monitoring target to obtain health trajectory information, matches the health trajectory information with the health trajectories of known high-risk workers to construct a group health association network, and outputs a target abnormal health assessment conclusion for the key monitoring target based on the spatiotemporal health trend visualization model and the group health association network. The data acquisition unit, feature extraction unit, anomaly detection unit, scenario evaluation unit, target screening unit, and risk assessment unit are sequentially connected and collaboratively complete the entire process of intelligent data acquisition and analysis from multi-source occupational health data collection to the assessment of abnormal health of key monitoring targets.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the intelligent acquisition method for multi-source occupational health monitoring as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the intelligent acquisition method for multi-source data of occupational health monitoring as described in any one of claims 1 to 7.

Images