Data analysis system for predicting respiratory diseases of children

Abstract

The invention discloses a data analysis system for predicting respiratory diseases of children, and relates to the technical field of health data analysis, and a data acquisition platform comprises a data cleaning model used for acquiring heterogeneous data of the respiratory diseases of children; the data cleaning model identifies the data inconsistency of the child respiratory diseases, analyzes unstructured clinical patient feedback, and extracts semantic information; the prediction model construction module is used for constructing a hybrid prediction model based on the medical environment data and the semantic information; and the scheme making module is used for inputting the respiratory disease condition of the child into the prediction model building module, analyzing the influence of different climate changes on the air quality and customizing a prevention and treatment scheme. According to the invention, real-time early warning of air quality is realized, parents and medical institutions are helped to take preventive measures in time, the onset risk of respiratory diseases of children is reduced, the intelligent level of prediction is improved, and a solid technical support is provided for personalized health management.

Description

Technical Field

[0001] The present invention relates to the technical field of health data analysis, in particular to a data analysis system for predicting respiratory diseases in children. Background Art

[0002] At present, with the acceleration of population growth and urbanization, children's health problems, especially the incidence of respiratory diseases, have attracted widespread attention. Modern families are paying more and more attention to children's health. In addition, the impact of factors such as environmental pollution and climate change has led to an increase in the incidence of children's respiratory diseases such as asthma and pneumonia. In order to meet this challenge, all sectors of society are actively seeking innovative solutions to improve the efficiency of disease prevention and treatment. As an emerging discipline, healthcare informatics uses advanced information and communication technologies to manage and analyze complex medical data, providing strong support for public health and personal health management. Especially in the context of the rapid development of big data and artificial intelligence technologies, the application potential of healthcare informatics has been unprecedentedly released, becoming an important means to improve the quality of medical services and enhance patient experience.

[0003] Currently, existing prediction methods for children's respiratory diseases mainly rely on doctors' experience and judgment and limited historical case data, which often leads to insufficient accuracy and timeliness of diagnosis. Traditionally, the data collected by medical institutions are mostly structured data. Due to the lack of effective data analysis tools and techniques, a large amount of valuable medical data has not been fully mined, which limits the understanding of the development of the disease and the improvement of early warning capabilities. These factors jointly restrict the effectiveness of existing technologies in practical applications, especially when faced with complex and changeable children's respiratory diseases, traditional prediction methods seem to be unable to cope with the situation. Summary of the invention

[0004] In view of the problems existing in the existing data analysis system for predicting children's respiratory diseases, the present invention is proposed.

[0005] In response to the problems of data integration difficulties and insufficient optimization of prediction models, the present invention solves the problems of data integration difficulties and insufficient optimization of prediction models in the prior art by constructing a multi-source heterogeneous data integration platform, parsing unstructured data and extracting semantic information, combining deep learning and transfer learning to design a hybrid prediction model, and introducing a reinforcement learning mechanism, thereby achieving more accurate prediction of children's respiratory diseases and personalized prevention and treatment plans.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: In a first aspect, the present invention provides a data analysis system for predicting pediatric respiratory diseases, comprising: a data collection platform for collecting heterogeneous data of pediatric respiratory diseases; a data cleaning model, wherein the data cleaning model includes identifying data inconsistencies of pediatric respiratory diseases, parsing unstructured clinical patient feedback, and extracting semantic information; and collecting medical environment data based on the extracted semantic information; A prediction model building module is used to build a hybrid prediction model based on the medical environment data and the semantic information; the hybrid prediction model is used to capture the long-term dependency between the medical environment data and the semantic information, and to share knowledge; a plan formulation module is used to input the respiratory disease conditions of children into the prediction model building module, analyze the impact of different climate changes on air quality, and customize prevention and treatment plans.

[0007] As a preferred solution of the data analysis system for predicting children's respiratory diseases described in the present invention, wherein: the construction of the data collection platform for children's respiratory diseases includes designing a multi-source heterogeneous data integration module to obtain structured electronic medical record test results, and the design of the multi-source heterogeneous data integration module includes using an adapter to obtain the data source characteristics of children's respiratory diseases, Convert the data collected by the data collection platform into an internal format that is consistent with the characteristics of the data source, establish a rule engine in the internal format, detect inconsistencies in the data in the internal format, use machine learning models to identify and mark abnormal data points, trigger a manual review process for abnormal data points, and retain normal data points for attention in subsequent analysis.

[0008] As a preferred solution of the data analysis system for predicting respiratory diseases in children of the present invention, wherein: the integration of the heterogeneous data includes constructing an adaptive classification architecture according to a machine learning model, and classifying the data as structured, semi-structured or unstructured according to the inconsistency of the data; The structuring includes searching using SQL query language, optimizing the query through indexes, and when the column header or row data of inconsistent data is found, the first step is to determine that it is structured data; Use the rule engine to check whether the column header of each column or the row data of each row conforms to the normal data type. When all column headers or row data are successfully verified, they are confirmed as structured data; The unstructured data includes searching using SQL query language, not finding any predefined format or structure, classifying the unstructured data using a machine learning model, and adjusting the classification results in combination with structured or semi-structured data; The semi-structured data includes using natural language processing technology to identify label attributes within the inconsistency of the data, building a semantic graph based on the identified label attributes, displaying the association relationship between the data, and enabling the machine learning model to learn the semi-structured data pattern through the graph; When encountering different data patterns, predict whether it belongs to semi-structured data by comparing it with known patterns.

[0009] As a preferred solution of the data analysis system for predicting respiratory diseases in children of the present invention, the parsing of unstructured clinical patient feedback includes applying a rule engine to match the classification results of data inconsistencies and checking whether different data patterns have the same label attributes; The machine learning model matches the labels in different data patterns. When two data patterns contain AA labels, the machine learning model will consider them to have the same label attributes. At this time, the machine learning model will generate a matching report and prompt the user to perform a test. For uncertain matching results, the machine learning model will prompt the user to perform manual matching, and the user can confirm or modify the matching results by selecting an operation; When the user selects yes, the machine learning model answers the matching question. When the user selects no, the machine learning model dynamically adjusts the matching rules. At this time, the user points out the misclassification of the label, and the machine learning model will update the rules. The extraction of semantic information includes preliminary cleaning of unstructured clinical patient feedback, removing redundant information, using natural language processing technology to segment the unstructured clinical patient feedback, and designing an interactive segmentation interface.

[0010] As a preferred solution of the data analysis system for predicting respiratory diseases in children described in the present invention, wherein: the collection of medical environment data based on the extracted semantic information includes feature extraction of the interactive word segmentation interface, and the feature extraction includes time series features, spatial features and text features; The time series features include extracting the timestamp and periodic change features of the interactive word segmentation interface, and capturing the dependency between the feature extraction and time of the interactive word segmentation interface; The spatial features include extracting geographical location and environmental parameter features of the interactive word segmentation interface and analyzing the differences between different regions; The text features include extracting word frequency, sentence length and punctuation usage features from unstructured text, and using NLP technology to identify medical terms and symptom descriptions; Associate the extracted semantic information with medical environment data, match the patient's symptom description with laboratory test results and environmental parameters to form a comprehensive health record, and build a knowledge graph in the medical field based on the comprehensive health record; The design of hybrid prediction models includes using learning algorithms to capture long-term dependencies in time series data, combining learning algorithms with machine learning models to form hybrid prediction models; The transfer learning technology of the machine learning model includes bringing the hybrid prediction model into a distributed training framework, allocating the model's training tasks to the reinforcement learning mechanism computing nodes, accelerating the training process, aggregating the results through a communication mechanism, and optimizing the hybrid prediction model after knowledge sharing.

[0011] As a preferred embodiment of the data analysis system for predicting respiratory diseases in children of the present invention, the customized prevention and treatment plan includes identifying risk factors through an optimized hybrid prediction model, the risk factors include identifying air quality based on collected medical environment data, and the air quality includes climate pattern identification and air quality impact assessment; The climate pattern recognition includes integrating air quality data and geographic information data from meteorological data, building a multi-source data fusion platform, and capturing the relationship between climate change and air quality through machine learning models; Capturing the relationship between climate change and air quality involves using time series analysis and an optimized hybrid forecasting model to predict future climate patterns and air quality changes.

[0012] As a preferred embodiment of the data analysis system for predicting respiratory diseases in children of the present invention, the analysis of the impact of different climate changes on air quality includes introducing a weather event prediction model to evaluate the impact of events on air quality; When the weather event prediction model is introduced, an early warning of air quality is issued, and the weather event prediction model is combined with air quality data and geographic information data; When the PM2.5 concentration in the air quality data rises, an early warning mechanism will be triggered, and the weather event prediction model will monitor future climate patterns and air quality changes, and complete data analysis for the prediction of children's respiratory diseases.

[0013] In a second aspect, the present invention provides a data analysis method for predicting children's respiratory diseases, which includes: building a data collection platform for children's respiratory diseases, identifying data inconsistencies in children's respiratory diseases, parsing unstructured clinical patient feedback, and extracting semantic information; collecting medical environment data based on the extracted semantic information, designing a hybrid prediction model, and optimizing the hybrid prediction model after knowledge sharing by introducing a reinforcement learning mechanism; in the optimized knowledge-sharing hybrid prediction model, customizing a prevention and treatment plan according to the child's respiratory disease condition.

[0014] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the processor executes the computer program, any step of the above-mentioned data analysis system for predicting childhood respiratory diseases is implemented.

[0015] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the above-mentioned data analysis system for predicting respiratory diseases in children is implemented.

[0016] The beneficial effects of the present invention are as follows: the present invention significantly improves the prediction accuracy and personalized prevention effect, and solves the problem of data inconsistency by constructing a multi-source heterogeneous data integration platform, integrating structured, semi-structured and unstructured data, ensuring that the data is complete and reliable, and using natural language processing technology to parse clinical feedback, extract semantic information, and enrich data depth. The designed hybrid prediction model combines deep learning to capture long-term dependencies, and realizes knowledge sharing through transfer learning, which improves the generalization ability of the model. The reinforcement learning mechanism is introduced to optimize the model so that it can be adaptively adjusted to ensure the real-time and accuracy of the prediction results. In addition, the impact of climate change on air quality is analyzed, and combined with the weather event prediction model, real-time early warning of air quality is achieved, which helps parents and medical institutions take preventive measures in time and reduce the risk of respiratory diseases in children. In general, the present invention improves the level of intelligent prediction and provides solid technical support for personalized health management. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them: Figure 1 The detailed flow chart of the data analysis system for predicting children's respiratory diseases.

[0018] Figure 2 Diagram of the internal structure of the computer equipment of the data analysis system used for the prediction of children's respiratory diseases. DETAILED DESCRIPTION

[0019] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0020] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0021] 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 term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0022] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0023] At the same time, in the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the directions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0024] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0025] Example 1, reference Figure 1 , which is the first embodiment of the present invention, provides a data analysis system for predicting respiratory diseases in children, including: Data collection platform, prediction model building module, and program formulation module; A data collection platform for collecting heterogeneous data on pediatric respiratory diseases; comprising a data cleaning model, the data cleaning model comprising identifying data inconsistencies on pediatric respiratory diseases, parsing unstructured clinical patient feedback, and extracting semantic information; and collecting medical environment data based on the extracted semantic information; A prediction model building module, used to build a hybrid prediction model based on the medical environment data and the semantic information; using the hybrid prediction model to capture the long-term dependency between the medical environment data and the semantic information, and to share knowledge; The program formulation module is used to input the children's respiratory disease conditions into the prediction model building module, analyze the impact of different climate changes on air quality, and customize prevention and treatment plans.

[0026] Specifically, the data collection platform includes designing a multi-source heterogeneous data integration module to obtain structured electronic medical record test results, and the design of a multi-source heterogeneous data integration module includes using adapters to obtain data source characteristics of children's respiratory diseases; Convert the data collected by the data collection platform into an internal format that is consistent with the characteristics of the data source, establish a rule engine in the internal format, detect inconsistencies in the data in the internal format, use machine learning models to identify and mark abnormal data points, trigger a manual review process for abnormal data points, and retain normal data points for attention in subsequent analysis.

[0027] Integration of heterogeneous data involves building an adaptive classification architecture based on machine learning models to classify data as structured, semi-structured, or unstructured based on their inconsistency; Structuring includes using SQL query language for retrieval and optimizing queries through indexes. When inconsistent data column headers or row data are found, the first step is to determine whether the data is structured. Use the rule engine to check whether the column header of each column or the row data of each row conforms to the normal data type. When all column headers or row data are successfully verified, they are confirmed as structured data; Unstructured includes searching using SQL query language, without finding any predefined format or structure, using machine learning models to classify unstructured data, and adjusting the classification results in combination with structured or semi-structured data; Semi-structured data includes using natural language processing technology to identify label attributes within the inconsistency of data, building semantic graphs based on the identified label attributes, displaying the associations between data, and enabling machine learning models to learn semi-structured data patterns through graphs; When encountering different data patterns, predict whether it belongs to semi-structured data by comparing it with known patterns.

[0028] Furthermore, parsing unstructured clinical patient feedback includes applying a rule engine to match the classification results based on the inconsistency of the data and checking whether different data patterns have the same label attributes; The machine learning model matches the labels in different data patterns. When two data patterns contain AA labels, the machine learning model will consider them to have the same label attributes. At this time, the machine learning model will generate a matching report and prompt the user to perform a test. For uncertain matching results, the machine learning model will prompt the user to perform manual matching, and the user can confirm or modify the matching results by selecting an operation; When the user selects yes, the machine learning model answers the matching question. When the user selects no, the machine learning model dynamically adjusts the matching rules. At this time, the user points out the misclassification of the label, and the machine learning model will update the rules. Extracting semantic information includes preliminary cleaning of unstructured clinical patient feedback, removing redundant information, using natural language processing technology to segment unstructured clinical patient feedback, and designing an interactive segmentation interface.

[0029] The prediction model building module collects medical environment data based on the extracted semantic information, designs a hybrid prediction model, combines deep learning algorithms to capture long-term dependencies in time series data, uses transfer learning technology to enable the hybrid prediction model to share knowledge, and optimizes the hybrid prediction model after knowledge sharing by introducing a reinforcement learning mechanism.

[0030] Among them, collecting medical environment data based on the extracted semantic information includes feature extraction of the interactive word segmentation interface, and the feature extraction includes time series features, spatial features and text features; Time series features include extracting the timestamp and periodic change features of the interactive word segmentation interface, and capturing the dependency between feature extraction and time of the interactive word segmentation interface; Spatial features include extracting geographic location and environmental parameter features of the interactive word segmentation interface and analyzing differences between different regions; Text features include extracting word frequency, sentence length, and punctuation usage features from unstructured text, and using NLP technology to identify medical terms and symptom descriptions; Associate the extracted semantic information with medical environment data, match the patient's symptom description with laboratory test results and environmental parameters to form a comprehensive health record, and build a knowledge graph in the medical field based on the comprehensive health record; The design of hybrid prediction models includes using learning algorithms to capture long-term dependencies in time series data, combining learning algorithms with machine learning models to form hybrid prediction models; The transfer learning technology of machine learning models includes bringing the hybrid prediction model into the distributed training framework, assigning the model's training tasks to the reinforcement learning mechanism computing nodes, accelerating the training process, aggregating the results through the communication mechanism, and optimizing the hybrid prediction model after knowledge sharing.

[0031] In the optimized knowledge-sharing hybrid prediction model, the plan-making module customizes prevention and treatment plans based on children's respiratory diseases, analyzes the impact of different climate changes on air quality, and completes data analysis for children's respiratory disease prediction.

[0032] Among them, customized prevention and treatment plans include identifying risk factors through optimized hybrid prediction models, risk factors include identifying air quality based on collected medical environment data, and air quality includes climate pattern identification and air quality impact assessment; Climate pattern recognition includes integrating air quality data and geographic information data from meteorological data, building a multi-source data fusion platform, and capturing the relationship between climate change and air quality through machine learning models; Capturing the relationship between climate change and air quality involves using time series analysis and optimized hybrid forecasting models to predict future climate patterns and air quality changes.

[0033] Further, the analysis of the impacts of different climate changes on air quality includes the introduction of weather event prediction models to assess the impact of events on air quality; When the weather event prediction model is introduced, an early warning of air quality is issued, and the weather event prediction model is combined with air quality data and geographic information data; When the PM2.5 concentration in the air quality data rises, an early warning mechanism will be triggered, and the weather event prediction model will monitor future climate patterns and air quality changes, and complete data analysis for the prediction of children's respiratory diseases.

[0034] Furthermore, in the optimized knowledge-sharing hybrid prediction model, the system customizes prevention and treatment plans based on children's respiratory diseases and analyzes the impact of climate change on air quality. Specifically, when the average PM2.5 concentration in the past week was detected to be 40 micrograms per cubic meter, exceeding the safety standard of 25 micrograms per cubic meter recommended by the World Health Organization, the system recommends that parents reduce their children's outdoor activities and use air purifiers. At the same time, the system found that the concentrations of sulfur dioxide and nitrogen oxides in the area have also increased, further exacerbating the risk of air pollution. For children with a history of asthma, when the PM2.5 concentration exceeds 35 micrograms per cubic meter, the system reminds parents to prepare necessary medications and take their children for regular physical examinations. The system integrates data from meteorological stations, air quality monitoring stations and geographic information systems, builds a multi-source data fusion platform, and monitors temperature, humidity, wind speed, precipitation, and PM2.5, PM10, ozone, sulfur dioxide and nitrogen oxides in real time. Through time series analysis, the system predicts that the PM2.5 concentration in a city will gradually increase in the next seven days, especially on the fourth and fifth days, when the PM2.5 concentration may reach 50 micrograms per cubic meter or more, exceeding the safety standard. In addition, the system predicts that the temperature will drop during the same period, which may lead to an increase in demand for coal-fired heating, further affecting air quality. The data on children's respiratory diseases and air quality impacts are shown in Table 1 below: Table 1 Data table of children's respiratory diseases and air quality impact Data Types Numeric unit describe Average PM2.5 concentration 40 Micrograms per cubic meter The average PM2.5 concentration in a city in the past week exceeded the safety standard recommended by the World Health Organization. World Health Organization safety standards 25 Micrograms per cubic meter PM2.5 safety standards recommended by the World Health Organization PM2.5 concentration threshold 35 Micrograms per cubic meter When PM2.5 concentration exceeds this value, parents are advised to take preventive measures Predicted PM2.5 concentration 50 Micrograms per cubic meter The system predicts the PM2.5 concentration in a city that may reach levels on the fourth and fifth days in the next seven days. Forecast period 7 sky The system predicts changes in air quality in the next seven days Temperature changes decline - The system predicts that temperatures will drop during the same period, which may lead to an increase in demand for coal-fired heating Table 1 shows the specific data involved in the system's analysis of children's respiratory diseases and the impact of air quality. These data provide a scientific basis for formulating prevention and treatment plans and help parents and medical institutions take effective protective measures in a timely manner.

[0035] In a preferred embodiment, a data analysis system for predicting children's respiratory diseases includes a construction module, which constructs a data collection platform for children's respiratory diseases, identifies data inconsistencies of children's respiratory diseases, parses unstructured clinical patient feedback, and extracts semantic information; an optimization module, which collects medical environment data based on the extracted semantic information, designs a hybrid prediction model, and optimizes the hybrid prediction model after knowledge sharing by introducing a reinforcement learning mechanism; and a prediction module, which customizes a prevention and treatment plan according to the child's respiratory disease condition in the optimized knowledge-sharing hybrid prediction model.

[0036] The computer device may be a terminal, which includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. Figure 2As shown, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0037] In summary, the present invention significantly improves the prediction accuracy and personalized prevention effect. By building a multi-source heterogeneous data integration platform, integrating structured, semi-structured and unstructured data, solving the problem of data inconsistency, ensuring data integrity and reliability, and using natural language processing technology to parse clinical feedback, extract semantic information, and enrich data depth, the designed hybrid prediction model combines deep learning to capture long-term dependencies, and realizes knowledge sharing through transfer learning, which improves the generalization ability of the model. The reinforcement learning mechanism is introduced to optimize the model so that it can be adaptively adjusted to ensure the real-time and accuracy of the prediction results. In addition, the impact of climate change on air quality is analyzed, and combined with the weather event prediction model, real-time warning of air quality is achieved, which helps parents and medical institutions take preventive measures in time and reduce the risk of respiratory diseases in children. In general, the present invention improves the intelligent level of prediction and provides solid technical support for personalized health management.

[0038] Example 2, reference Figure 1 , which is the second embodiment of the present invention, and this embodiment provides a data analysis system for predicting children's respiratory diseases. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through simulation experiments.

[0039] The extracted semantic information is associated with the medical environment data to form a comprehensive health record, which contains symptom descriptions of 500 patients, 300 laboratory test results and 200 environmental parameters. These data are used to build a knowledge graph in the medical field. The designed hybrid prediction model uses the LSTM algorithm to capture the long-term dependencies in time series data, and combines the XGBoost machine learning model to form a hybrid prediction model. During the training process, 10,000 historical data and 2,000 verification data were used, and the accuracy of the model reached 92%. The transfer learning technology brought the hybrid prediction model into the distributed training framework, and the training tasks of the model were distributed to 10 computing nodes. Each node processed 1,000 data, which accelerated the training process. The results of each node were summarized through the communication mechanism, and the accuracy of the optimized hybrid prediction model on the test set was increased to 95%. The experimental results show that this method can effectively improve the accuracy and efficiency of predicting respiratory diseases in children. The experimental data of the present invention are shown in Table 2 below: Table 2 Experimental data table of the present invention Phases / Components describe Key data Data Linking and Integrated Health Records Associate the extracted semantic information with the medical environment data to form a comprehensive health record 500 patient symptom descriptions, 300 laboratory test results, 200 environmental parameters Knowledge graph construction Constructing a knowledge graph in the medical field based on comprehensive health records - Hybrid prediction model design The LSTM algorithm is used to capture the long-term dependencies in time series data, and the XGBoost machine learning model is combined to form a hybrid prediction model. LSTM + XGBoost Training Data Using historical data for model training 10,000 historical data Verify data Evaluate model performance using validation data 2000 verification data Initial Accuracy Model accuracy before optimization 92% Distributed training framework Transfer learning technology brings hybrid prediction models into distributed training framework 10 computing nodes, each node processes 1000 pieces of data Communication Mechanism Aggregate the results of each node through the communication mechanism - Optimized accuracy The accuracy of the optimized hybrid prediction model on the test set 95% Table 2 summarizes the key steps and related data from data association to final model optimization, making the information more intuitive and easy to read. Since the original text does not mention the specific amount of test data, this item is not included in the table. If more detailed information is needed, the table content can be further refined or additional columns can be added to cover more details. The comparison between the present invention and the prior art is shown in Table 3 below: Table 3 Comparison table between the present invention and the prior art Comparison Items Prior art The present invention Advantages of the Invention Data collection and integration Relying on a single data source makes it difficult to integrate heterogeneous data from multiple sources; data cleaning and preprocessing processes are complex and inefficient A multi-source heterogeneous data integration module has been built to efficiently convert and unify data formats, and automatically detect and mark abnormal data points Supports efficient integration of multi-source heterogeneous data, improving data quality and processing efficiency Semantic Information Extraction Difficulty in effectively parsing unstructured text, such as clinical patient feedback, leading to insufficient use of semantic information Apply rule engines and machine learning models to match tag attributes, and combine natural language processing technology for word segmentation and feature extraction Improved understanding of unstructured text and enhanced accuracy and usability of semantic information Predictive model design Mainly rely on traditional statistical methods or single algorithms, which are difficult to capture long-term dependencies in time series A hybrid prediction model was designed, combining LSTM and XGBoost, which can more effectively capture long-term dependencies in time series data. Hybrid model design improves prediction accuracy, especially for time-dependent respiratory disease prediction Transfer learning applications Lack of effective knowledge sharing mechanism and low model training efficiency Introducing transfer learning technology, bringing the hybrid prediction model into the distributed training framework, accelerating the training process Knowledge sharing is achieved through transfer learning, which greatly shortens the model training time and improves the training efficiency Computing resource utilization Single-node training is usually used, and computing resource utilization is low Distribute the training tasks to 10 computing nodes, each node processes 1,000 pieces of data, and aggregate the results through the communication mechanism The distributed training framework significantly improves the utilization of computing resources and speeds up model optimization Prediction Accuracy The accuracy is usually around 85%, and there is room for improvement The optimized hybrid prediction model has an accuracy of 95% on the test set. Significantly improved prediction accuracy, providing a more reliable basis for the prevention and treatment of children's respiratory diseases Table 3 shows the main differences between the present invention and the prior art, highlighting the significant advantages of the present invention in multiple key areas, including data processing, semantic information extraction, prediction model design, transfer learning application, computing resource utilization, prediction accuracy, and climate change and air quality analysis. These advantages together constitute the core competitiveness of the present invention, making it have obvious technological leadership in the field of children's respiratory disease prediction.

[0040] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A data analysis system for predicting respiratory diseases in children, characterized by: include: Data collection platform for collecting heterogeneous data on children's respiratory diseases; It includes a data cleaning model, which includes identifying data inconsistencies of children's respiratory diseases, parsing unstructured clinical patient feedback, and extracting semantic information; collecting medical environment data based on the extracted semantic information; A prediction model building module, used to build a hybrid prediction model based on the medical environment data and the semantic information; using the hybrid prediction model to capture the long-term dependency between the medical environment data and the semantic information, and to share knowledge; The program formulation module is used to input the children's respiratory disease conditions into the prediction model building module, analyze the impact of different climate changes on air quality, and customize prevention and treatment plans.

2. The data analysis system for predicting children's respiratory diseases according to claim 1, characterized in that: The data acquisition platform includes designing a multi-source heterogeneous data integration module to obtain structured electronic medical record test results, and the designing of the multi-source heterogeneous data integration module includes using an adapter to obtain data source characteristics of children's respiratory diseases; Convert the data collected by the data collection platform into an internal format that is consistent with the characteristics of the data source, establish a rule engine in the internal format, detect inconsistencies in the data in the internal format, use machine learning models to identify and mark abnormal data points, trigger a manual review process for abnormal data points, and retain normal data points for attention in subsequent analysis.

3. The data analysis system for predicting children's respiratory diseases according to claim 2, characterized in that: The integration of the heterogeneous data includes building an adaptive classification framework based on a machine learning model to classify the data into structured, semi-structured or unstructured according to its inconsistency; The structuring includes searching using SQL query language, optimizing the query through indexes, and when the column header or row data of inconsistent data is found, the first step is to determine that it is structured data; Use the rule engine to check whether the column header of each column or the row data of each row conforms to the normal data type. When all column headers or row data are successfully verified, they are confirmed as structured data; The unstructured data includes searching using SQL query language, not finding any predefined format or structure, classifying the unstructured data using a machine learning model, and adjusting the classification results in combination with structured or semi-structured data; The semi-structured method includes using natural language processing technology to identify label attributes within the inconsistency of data, constructing a semantic graph based on the identified label attributes, displaying the association relationship between data, and enabling the machine learning model to learn the semi-structured data pattern through the graph; When encountering different data patterns, predict whether it belongs to semi-structured data by comparing it with known patterns.

4. The data analysis system for predicting children's respiratory diseases according to claim 3, characterized in that: The parsing of unstructured clinical patient feedback includes applying a rule engine to match the classification results based on the inconsistency of the data, and checking whether different data patterns have the same label attributes; The machine learning model matches the labels in different data patterns. When two data patterns contain AA labels, the machine learning model will consider them to have the same label attributes. At this time, the machine learning model will generate a matching report and prompt the user to perform a test. For uncertain matching results, the machine learning model will prompt the user to perform manual matching, and the user can confirm or modify the matching results by selecting an operation; When the user selects yes, the machine learning model answers the matching question. When the user selects no, the machine learning model dynamically adjusts the matching rules. At this time, the user points out the misclassification of the label, and the machine learning model will update the rules. The extraction of semantic information includes preliminary cleaning of unstructured clinical patient feedback, removing redundant information, using natural language processing technology to segment the unstructured clinical patient feedback, and designing an interactive segmentation interface.

5. The data analysis system for predicting children's respiratory diseases according to claim 4, characterized in that: The collecting of medical environment data based on the extracted semantic information includes extracting features from an interactive word segmentation interface, wherein the feature extraction includes time series features, spatial features, and text features; The time series features include extracting the timestamp and periodic change features of the interactive word segmentation interface, and capturing the dependency between the feature extraction and time of the interactive word segmentation interface; The spatial features include extracting geographical location and environmental parameter features of the interactive word segmentation interface and analyzing the differences between different regions; The text features include extracting word frequency, sentence length and punctuation usage features from unstructured text, and using NLP technology to identify medical terms and symptom descriptions; Associate the extracted semantic information with medical environment data, match the patient's symptom description with laboratory test results and environmental parameters to form a comprehensive health record, and build a knowledge graph in the medical field based on the comprehensive health record; The design of hybrid prediction models includes using learning algorithms to capture long-term dependencies in time series data, combining learning algorithms with machine learning models to form hybrid prediction models; The transfer learning technology of the machine learning model includes bringing the hybrid prediction model into a distributed training framework, allocating the model's training tasks to the reinforcement learning mechanism computing nodes, accelerating the training process, aggregating the results through a communication mechanism, and optimizing the hybrid prediction model after knowledge sharing.

6. The data analysis system for predicting children's respiratory diseases according to claim 5, characterized in that: The customized preventive treatment plan includes identifying risk factors through an optimized hybrid prediction model, the risk factors include identifying air quality based on collected medical environment data, the air quality includes climate pattern identification and air quality impact assessment; The climate pattern recognition includes integrating air quality data and geographic information data from meteorological data, building a multi-source data fusion platform, and capturing the relationship between climate change and air quality through machine learning models; Capturing the relationship between climate change and air quality involves using time series analysis and an optimized hybrid forecasting model to predict future climate patterns and air quality changes.

7. The data analysis system for predicting children's respiratory diseases according to claim 6, characterized in that: The analysis of the impacts of different climate changes on air quality includes introducing a weather event prediction model to assess the impact of events on air quality; When the weather event prediction model is introduced, an early warning of air quality is issued, and the weather event prediction model is combined with air quality data and geographic information data; When the PM2.5 concentration in the air quality data rises, an early warning mechanism will be triggered, and the weather event prediction model will monitor future climate patterns and air quality changes, and complete data analysis for the prediction of children's respiratory diseases.

8. A data analysis method for predicting respiratory diseases in children, based on the data analysis system for predicting respiratory diseases in children according to any one of claims 1 to 7, characterized in that: include: Build a data collection platform for children's respiratory diseases, identify data inconsistencies for children's respiratory diseases, parse unstructured clinical patient feedback, and extract semantic information; Collect medical environment data based on the extracted semantic information, design a hybrid prediction model, and optimize the hybrid prediction model after knowledge sharing by introducing a reinforcement learning mechanism; In the optimized knowledge-sharing hybrid prediction model, prevention and treatment plans are customized according to the children's respiratory diseases.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the data analysis system for predicting children's respiratory diseases according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the data analysis system for predicting respiratory diseases in children described in any one of claims 1 to 7 are implemented.

Images