Sick and wounded information input method and system based on mobile terminal

Abstract

The invention provides a sick and wounded information input method and system based on a mobile terminal. The method comprises the following steps: collecting basic information of the sick and wounded through a user mobile terminal, and generating a basic information framework of the sick and wounded in combination with geographic position and timestamp data; based on the basic information framework of the sick and wounded, comparing with historical cases stored in a cloud database by applying a medical map reinforcement learning algorithm, and performing causal relationship analysis on historical case texts by adopting a causal reasoning technology to generate a preliminary diagnosis suggestion; on the basis of the preliminary diagnosis suggestion, carrying out detailed feature recognition processing by applying a self-supervised learning algorithm, and carrying out transcription and sentiment analysis on the voice description of the sick and wounded by adopting a multi-modal fusion technology to generate a multimedia symptom record; and based on the multimedia symptom record, creating a sick and wounded electronic medical record update record, and generating a sick and wounded information input scheme. According to the technical scheme provided by the invention, the sick and wounded information input efficiency and comprehensiveness are improved, and the medical service quality is improved.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of medical information technology, and in particular to a method and system for entering information of injured and sick persons based on a mobile terminal. Background Art

[0002] With the rapid development of medical informatization and mobile Internet technology, modern medical services have put forward higher requirements for the rapid and accurate entry of information on the injured and sick. Especially in scenarios such as emergency treatment, telemedicine and daily outpatient clinics, medical staff need to obtain and process a large amount of information on the injured and sick in a short period of time to ensure timely and effective diagnosis and treatment. In order to meet this demand, the system not only needs to efficiently collect basic information on the injured and sick, but also needs to combine geographic location and timestamp data to generate a basic information framework, and be able to intelligently analyze this information to provide preliminary diagnostic recommendations. In addition, the system should also support multimodal data analysis, such as sentiment analysis of voice descriptions, to generate more comprehensive multimedia symptom records, and ultimately achieve automatic updating and push of electronic medical records.

[0003] At present, many medical institutions have adopted information entry systems based on mobile terminals, collecting basic information of the injured and sick through mobile devices (such as smartphones and tablets) and uploading it to cloud databases for storage and management. Existing systems usually combine simple data form filling and partial automation processing functions, which can improve the speed and accuracy of information entry to a certain extent. However, these systems mainly rely on predefined data templates and rule engines, lacking intelligence and flexibility.

[0004] Although the existing solutions have improved the process of entering the information of the wounded and sick to a certain extent, there are still many shortcomings. First, the traditional method relies on manual input, which is time-consuming and prone to errors. Especially in emergency situations, it cannot quickly respond to and process a large amount of information, and the efficiency of entering the information of the wounded and sick is low. Secondly, the intelligence level of the existing system is limited, and it cannot make full use of historical case data for intelligent matching and causal analysis, resulting in inaccurate diagnostic recommendations. Finally, the existing multimodal data analysis capabilities are weak, and it fails to fully integrate multiple data sources such as voice and images, ignoring the value of unstructured information such as emotional cues, which limits the comprehensive evaluation capabilities of the system. Therefore, there is an urgent need for a more intelligent and efficient method for entering the information of the wounded and sick to improve the quality and efficiency of medical services. Summary of the invention

[0005] The embodiments of the present application provide a method and system for entering information of injured and sick persons based on a mobile terminal, so as to solve the problem of low efficiency in entering information of injured and sick persons in the prior art.

[0006] In a first aspect, an embodiment of the present application provides a method for entering information of a sick and injured person based on a mobile terminal, comprising:

[0007] Collect basic information of the injured and sick through the user's mobile terminal, and generate a basic information framework of the injured and sick by combining the geographic location and timestamp data;

[0008] Based on the basic information framework of the injured and sick, the medical graph-enhanced learning algorithm is used to compare with the historical cases stored in the cloud database, the knowledge graph is used to enhance the learning process, and the causal reasoning technology is used to analyze the causal relationship of the historical case text, extract the key factors causing the disease, and generate preliminary diagnosis suggestions;

[0009] Based on the preliminary diagnostic suggestions, a self-supervised learning algorithm is used to perform detailed feature recognition processing, automatically learn useful feature representations, and use multimodal fusion technology to transcribe and perform sentiment analysis on the patient's voice descriptions, identify sentiment clues, and generate multimedia symptom records;

[0010] Based on the multimedia symptom record, an updated record of the electronic medical record of the patient is created and simultaneously pushed to relevant medical institutions, allowing medical personnel to review and confirm through mobile terminals and generate a plan for entering the patient's information.

[0011] Optionally, based on the basic information framework of the injured and sick, a medical graph enhancement learning algorithm is used to compare with historical cases stored in a cloud database, a knowledge graph is used to enhance the learning process, and causal reasoning technology is used to perform causal relationship analysis on historical case texts, extract key pathogenic factors, and generate preliminary diagnosis suggestions, including:

[0012] Based on the basic information framework of the wounded and sick, semantic parsing and context understanding processing are performed to capture surface meanings and implicit associations, and generate background portraits of the wounded and sick;

[0013] Based on the background portrait of the injured and sick, the medical graph enhanced learning algorithm is used to perform comprehensive intelligent matching processing on the injured and sick information, and compared with the historical cases stored in the cloud database, and the medical entity and relationship information in the knowledge graph is used to enhance the learning process to generate a matching result list;

[0014] Based on the matching result list, causal reasoning technology is used to conduct in-depth analysis of the retrieved historical case texts, explore the causal mechanism between events, identify the key causes of specific diseases, and generate a report on key pathogenic factors;

[0015] Based on the report on key pathogenic factors, an expert system rule engine is used to perform optimization processing and generate preliminary diagnostic recommendations.

[0016] Optionally, based on the background portrait of the patient, the medical graph enhancement learning algorithm is used to perform comprehensive intelligent matching processing on the patient information, compare it with the historical cases stored in the cloud database, and enhance the learning process using the medical entities and relationship information in the knowledge graph to generate a matching result list, including:

[0017] Based on the background portraits of the injured and sick, combined with external medical resources, multi-source data fusion processing is performed to generate a comprehensive information matrix of the injured and sick;

[0018] Based on the comprehensive patient information matrix, the medical graph enhanced learning algorithm is used to perform comprehensive intelligent matching processing on the patient information, and compared with the historical cases stored in the cloud database. The learning process is enhanced by the medical entity and relationship information in the knowledge graph to identify similar historical cases and generate a preliminary matching list.

[0019] Based on the preliminary matching list, a dynamic weight adjustment technology is used to dynamically adjust the matching weight of each historical case to generate a dynamic adjustment result;

[0020] Based on the dynamic adjustment results, a complete evidence chain from symptoms to diagnosis is constructed through an evidence chain construction method to generate a matching result list.

[0021] Optionally, based on the matching result list, causal reasoning technology is used to deeply analyze the retrieved historical case texts, explore the causal mechanism between events, identify the key causes of specific diseases, and generate a key pathogenic factor report, including:

[0022] Based on the matching result list, semantic analysis and structural processing are performed through natural language analysis to mark possible causal relationships and generate causal relationship annotation text;

[0023] Based on the causal association annotated text, causal reasoning technology is used to deeply analyze the causal mechanism between events, identify and verify the causal links between different medical events, and generate a causal chain model;

[0024] Based on the causal chain model, further identification processing is performed to screen factors with high frequency of occurrence and strong causal indication effects, and a list of key pathogenic factors is generated;

[0025] Based on the list of key pathogenic factors, a multi-dimensional risk assessment process is performed to quantify the risk level and generate a key pathogenic factor report.

[0026] Optionally, based on the preliminary diagnosis suggestion, a self-supervised learning algorithm is used to perform detailed feature recognition processing, automatically learn useful feature representation, adopt multimodal fusion technology to transcribe and perform sentiment analysis on the voice description of the patient, identify sentiment clues, and generate a multimedia symptom record, including:

[0027] Based on the preliminary diagnostic suggestions, the real-time symptom images are standardized and enhanced to ensure image quality consistency and generate standardized symptom images;

[0028] Based on the standardized symptom images, a self-supervised learning algorithm is used to perform detailed feature recognition processing, automatically learn useful feature representations, accurately capture the symptom characteristics of the injured and sick, and generate a detailed symptom feature map;

[0029] Based on the detailed symptom feature map, multimodal fusion technology is used to convert the voice signal into text form, and sentiment analysis technology is used to analyze the sentiment clues to generate a symptom text sentiment description;

[0030] Based on the emotional description of the symptom text, visualization tools are used to display the development and change process of the patient's symptoms and generate a multimedia symptom record.

[0031] Optionally, based on the standardized symptom images, a self-supervised learning algorithm is used to perform detailed feature recognition processing, automatically learn useful feature representations, accurately capture the symptom characteristics of the injured and sick, and generate a detailed symptom feature map, including:

[0032] Based on the standardized symptom images, pre-training processing is performed to extract general feature representations, ensure that the learning process is efficient and stable, and generate pre-trained feature representations;

[0033] Based on the pre-trained feature representation, a self-supervised learning algorithm is used to perform feature enhancement processing, identify detailed features, automatically learn useful feature representations, and generate enhanced feature representations;

[0034] Based on the enhanced feature representation, the symptom characteristics of the injured and sick are captured from different levels and angles through multi-scale feature extraction technology to generate a multi-level symptom feature set;

[0035] Based on the multi-level symptom feature set, feature association analysis is performed, the strength of association and potential impact between features are marked, and a detailed symptom feature map is generated.

[0036] Optionally, based on the multimedia symptom record, an electronic medical record update record of the patient is created, which is simultaneously pushed to relevant medical institutions, allowing medical personnel to review and confirm through a mobile terminal, and generating a patient information entry plan, including:

[0037] Based on the multimedia symptom record, parsing visual and text information, extracting key symptom descriptions and preliminary diagnostic suggestions, performing standardized format conversion, and generating structured diagnostic information;

[0038] Based on the structured diagnostic information, the existing electronic medical record data of the patient is updated and processed, and the latest medical assessment is integrated to generate an updated electronic medical record version;

[0039] Based on the updated electronic medical record version, the updated electronic medical record version is synchronously pushed to the cloud server and the database of the relevant medical institution through a secure data transmission protocol to generate a synchronous push record;

[0040] Based on the synchronous push records, a review and confirmation process is constructed to allow medical personnel to access the latest patient information through mobile terminals for review and confirmation, and to generate a plan for entering patient information.

[0041] In a second aspect, an embodiment of the present application provides a system for entering information of injured and sick persons based on a mobile terminal, comprising:

[0042] The collection module is used to collect basic information of the injured and sick through the user's mobile terminal, and generate a basic information framework of the injured and sick by combining the geographic location and timestamp data;

[0043] An analysis module is used to compare the basic information framework of the injured and sick with the historical cases stored in the cloud database using the medical graph enhancement learning algorithm, use the knowledge graph to enhance the learning process, and use causal reasoning technology to perform causal relationship analysis on the historical case text, extract key factors causing the disease, and generate preliminary diagnosis suggestions;

[0044] A processing module, which is used to perform detailed feature recognition processing based on the preliminary diagnosis suggestion, use a self-supervised learning algorithm, automatically learn useful feature representations, use multimodal fusion technology to transcribe and perform sentiment analysis on the voice description of the patient, identify sentiment clues, and generate a multimedia symptom record;

[0045] A creation module is used to create an updated electronic medical record of the patient based on the multimedia symptom record, and push it to relevant medical institutions simultaneously, allowing medical personnel to review and confirm through mobile terminals and generate an information entry plan for the patient.

[0046] In a third aspect, an embodiment of the present application provides a computing device, comprising a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a method for entering information of injured and sick persons based on a mobile terminal as described in the first aspect.

[0047] In a fourth aspect, an embodiment of the present application provides a computer storage medium storing a computer program. When the computer program is executed by a computer, a method for entering information of injured and sick persons based on a mobile terminal as described in the first aspect is implemented.

[0048] In an embodiment of the present application, basic information of the injured and sick is collected through a user's mobile terminal, and a basic information framework of the injured and sick is generated in combination with geographic location and timestamp data; based on the basic information framework of the injured and sick, a medical graph-enhanced learning algorithm is used to compare with historical cases stored in a cloud database, the learning process is enhanced by a knowledge graph, and causal reasoning technology is used to perform causal relationship analysis on historical case texts, extract key pathogenic factors, and generate preliminary diagnosis suggestions; based on the preliminary diagnosis suggestions, a self-supervised learning algorithm is used to perform detailed feature recognition processing, automatically learn useful feature representations, and use multimodal fusion technology to transcribe and sentimentally analyze the voice descriptions of the injured and sick, identify emotional clues, and generate multimedia symptom records; based on the multimedia symptom records, an electronic medical record update record of the injured and sick is created and pushed to relevant medical institutions simultaneously, allowing medical personnel to review and confirm through mobile terminals, and generate a plan for entering information on the injured and sick. The basic information of the injured and sick is collected through the user's mobile terminal, and the geographic location and timestamp data are combined to generate a basic information framework for the injured and sick. This method ensures the real-time and accuracy of the information, and provides a solid foundation for subsequent analysis; the medical graph enhancement learning algorithm and causal reasoning technology are used to compare and analyze historical cases, extract key factors causing the disease, and generate preliminary diagnosis suggestions, which not only improves the speed of diagnosis, but also enhances the accuracy of diagnosis; the self-supervised learning algorithm and multimodal fusion technology are used to transcribe and sentimentally analyze the voice descriptions of the injured and sick, identify emotional clues, and generate multimedia symptom records. This method can more comprehensively capture the symptom characteristics of the injured and sick and provide richer information support; create electronic medical record update records for the injured and sick, and push them to relevant medical institutions simultaneously, allowing medical personnel to review and confirm through mobile terminals, which ensures the timely sharing of information and improves the efficiency of medical services.

[0049] Furthermore, through semantic parsing and context understanding, background portraits of the injured and sick are generated, capturing surface meanings and implicit associations, providing a more accurate basis for subsequent matching. Secondly, the medical graph enhancement learning algorithm is used for comprehensive intelligent matching processing, combining the medical entities and relationship information in the knowledge graph to generate a list of matching results, ensuring the intelligence and efficiency of the matching process. Furthermore, through causal reasoning technology, historical case texts are deeply analyzed, the causal mechanism between events is explored, the key causes of specific diseases are identified, and a report on key pathogenic factors is generated, which improves the depth and breadth of diagnosis. Finally, preliminary diagnostic suggestions are generated through expert system rule engine optimization processing, ensuring the professionalism and authority of the diagnostic suggestions.

[0050] Furthermore, the real-time symptom images are standardized and enhanced to ensure the consistency of image quality and generate standardized symptom images, providing a high-quality data basis for subsequent feature recognition. Secondly, the self-supervised learning algorithm is used for detailed feature recognition processing, automatically learning useful feature representations, accurately capturing the symptom characteristics of the injured and sick, and generating detailed symptom feature maps, which enhances the robustness and discrimination of feature representation. Thirdly, multimodal fusion technology is used to convert voice signals into text form, and sentiment analysis technology is combined to parse sentiment clues and generate sentiment descriptions of symptom texts, enriching the content of symptom records and providing more dimensional information support. Finally, visualization tools are used to display the development and change process of the symptoms of the injured and sick, and multimedia symptom records are generated, so that medical personnel can intuitively understand the development of the injured and sick, support more accurate clinical decision-making, and thus improve the quality of diagnosis and treatment and patient satisfaction.

[0051] These and other aspects of the present application will become more clearly understood in the description of the following embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0053] Figure 1 A flowchart of a method for entering information of injured and sick persons based on a mobile terminal provided in an embodiment of the present application;

[0054] Figure 2 A schematic diagram of the structure of a system for entering information of injured and sick persons based on a mobile terminal provided in an embodiment of the present application;

[0055] Figure 3 A schematic diagram of the structure of a computing device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0056] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0057] In some of the processes described in the specification and claims of this application and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this article or executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish between different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., do not represent the order of precedence, and do not limit the "first" and "second" to be different types.

[0058] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.

[0059] Figure 1 A flowchart of a method for entering information of a patient based on a mobile terminal is provided for an embodiment of the present application, such as Figure 1 As shown, the method includes:

[0060] 101. Collect basic information of the injured and sick through the user's mobile terminal, and generate a basic information framework of the injured and sick by combining the geographic location and timestamp data;

[0061] In this step, the basic information of the injured and sick includes personal information such as name, age, gender, contact information, as well as medical information such as symptom description and medical history. This information is used to build a basic file for the injured and sick.

[0062] The geographic location data is the current location coordinates obtained through the GPS module of the mobile terminal. It is used to determine the specific location where the injury or illness occurred, which helps to quickly respond in emergency situations.

[0063] Timestamp data is data that records the time point of information collection. It is used to track the development of the disease and ensure the traceability of information.

[0064] The basic information framework of the injured and sick is a structured representation that integrates basic information, geographic location and timestamp data. It serves as the input for subsequent intelligent processing to ensure the integrity and consistency of all relevant information.

[0065] In an embodiment of the present application, it is assumed that a traffic accident occurs in a city and multiple people are injured. After the emergency personnel arrive at the scene, they use a mobile terminal equipped with a special medical application to quickly enter the personal information and preliminary symptom description of each injured person; first, the emergency personnel enter the name, age, gender and other basic information of the injured person through the application interface; secondly, the system automatically obtains the current location coordinates and records the timestamp to ensure that each piece of information has a clear time and space mark; thirdly, the application uses the built-in GPS module and network time protocol to ensure the accuracy of the geographic location and time data; finally, all collected information is uploaded to the cloud server in real time to build a basic information framework for the injured and sick, in preparation for subsequent processing.

[0066] 102. Based on the basic information framework of the injured and sick, the medical graph enhancement learning algorithm is used to compare with the historical cases stored in the cloud database, the knowledge graph is used to enhance the learning process, and the causal reasoning technology is used to analyze the causal relationship of the historical case text, extract the key factors causing the disease, and generate preliminary diagnosis suggestions;

[0067] In this step, the medical graph reinforcement learning algorithm is a learning algorithm that combines professional knowledge in the medical field. It is used to match the information of injured and sick people with historical cases in the cloud database, find similar case patterns, and improve the accuracy and efficiency of matching.

[0068] A cloud database refers to a large database stored on a cloud server, which contains a large number of historical cases and their related information, provides rich reference data, and supports the generation of intelligent matching and diagnostic recommendations.

[0069] The knowledge graph is a structured knowledge base formed by constructing a network of medical entities (such as diseases, symptoms, drugs, etc.) and their relationships. It enhances the relevance of matching results and helps the system identify relevant disease characteristics more accurately.

[0070] Causal reasoning technology is a logical reasoning method used to analyze the cause-effect relationship in historical case texts, identify the key causes of specific diseases, and enhance the depth and breadth of diagnosis.

[0071] Key pathogenic factors are important factors or conditions that lead to specific diseases extracted from historical case texts, including but not limited to environmental factors, genetic factors, lifestyle habits, etc. They are crucial for understanding the cause of the disease and formulating treatment plans.

[0072] The preliminary diagnostic recommendation is a diagnostic opinion generated based on the report of key pathogenic factors. It provides important reference for medical personnel and helps them make preliminary judgments quickly.

[0073] In an embodiment of the present application, assuming that after the emergency center receives the basic information framework uploaded from the accident site, the system immediately starts the intelligent matching process. First, the system searches for similar historical cases in the cloud database based on the basic information framework of the injured and sick, and uses the medical graph enhancement learning algorithm for intelligent matching; secondly, the system enhances the relevance of the matching results through the medical entities and relationship information in the knowledge graph to ensure that the most relevant cases are found; thirdly, the system uses causal reasoning technology to deeply analyze the retrieved historical case texts, identify the key causes of specific diseases, and generate a report on key pathogenic factors; finally, based on these analysis results, the system optimizes the processing through the expert system rule engine to generate detailed preliminary diagnosis suggestions for reference by medical personnel.

[0074] Optionally, the method in step 102 is based on the basic information framework of the injured and sick, uses a medical graph to enhance the learning algorithm, compares with historical cases stored in a cloud database, uses the knowledge graph to enhance the learning process, and uses causal reasoning technology to perform causal relationship analysis on the historical case text, extract key pathogenic factors, and generate preliminary diagnosis suggestions, including: based on the basic information framework of the injured and sick, perform semantic parsing and context understanding processing, capture surface meanings and implicit associations, and generate background portraits of the injured and sick; based on the background portraits of the injured and sick, use a medical graph to enhance the learning algorithm, perform comprehensive intelligent matching processing on the information of the injured and sick, compare with historical cases stored in a cloud database, use the medical entity and relationship information in the knowledge graph to enhance the learning process, and generate a matching result list; based on the matching result list, use causal reasoning technology to perform in-depth analysis on the retrieved historical case text, explore the causal mechanism between events, identify the key causes of specific diseases, and generate a report on key pathogenic factors; based on the report on key pathogenic factors, use an expert system rule engine to perform optimization processing to generate preliminary diagnosis suggestions.

[0075] In this step, semantic parsing and context understanding processing include using natural language processing (NLP) technology to conduct in-depth analysis of the text descriptions provided by the injured and sick. It not only captures the surface meaning, but also identifies implicit associations, such as potential connections between symptoms and the historical background of the disease.

[0076] The background portrait of the injured and sick is a comprehensive description file generated through semantic parsing and context understanding. It contains the basic information of the injured and sick, symptom descriptions and their potential connections. It provides a detailed basis for subsequent intelligent matching and helps medical staff better understand the specific conditions of the injured and sick.

[0077] The matching result list refers to a list of similar historical cases found by the system in the cloud database based on the background portrait of the injured and sick. Each entry contains not only the basic information of the case, but also includes the high similarities with the current situation of the injured and sick, supporting further causal analysis.

[0078] Causal analysis uses logical reasoning methods to parse the causal relationships in historical case texts to identify the key causes of specific diseases. It not only focuses on the direct causal chain, but also considers indirect factors such as environmental influences, genetic factors, lifestyle habits, etc., thereby providing a more comprehensive etiology analysis.

[0079] The expert system rule engine is an automated decision-making tool based on predefined rules, combined with the knowledge base of medical experts, to optimize preliminary diagnostic recommendations.

[0080] Cloud database refers to a large database stored on a cloud server, containing a large number of historical cases and their related information.

[0081] In the embodiment of the present application, it is assumed that a traffic accident occurred on a highway and multiple people were injured. After the emergency personnel arrived at the scene, they used a mobile terminal equipped with a special medical application to quickly enter the personal information and preliminary symptom description of each injured person. First, the system performs semantic analysis and context understanding on the basic information framework of the injured and sick, captures the surface meaning and implicit association, and generates a detailed background portrait of the injured and sick; secondly, based on the generated background portrait, the system uses the medical atlas enhanced learning algorithm to perform comprehensive intelligent matching processing on the information of the injured and sick, and compares it with the historical cases stored in the cloud database to generate a list of matching results; thirdly, the system uses causal analysis technology to deeply analyze the retrieved historical case texts, explore the causal mechanism between events, identify the key causes of specific diseases, and generate a report on key pathogenic factors; finally, based on the report on key pathogenic factors, the system uses the expert system rule engine for optimization processing to generate the final preliminary diagnosis suggestions for reference by medical personnel to guide subsequent treatment work.

[0082] Optionally, based on the background portrait of the injured and sick, the medical graph enhancement learning algorithm is used to perform comprehensive intelligent matching processing on the information of the injured and sick, and the information is compared with historical cases stored in the cloud database, and the learning process is enhanced by using the medical entities and relationship information in the knowledge graph to generate a matching result list, including: based on the background portrait of the injured and sick, combined with external medical resources, multi-source data fusion processing is performed to generate a comprehensive information matrix of the injured and sick; based on the comprehensive information matrix of the injured and sick, the medical graph enhancement learning algorithm is used to perform comprehensive intelligent matching processing on the information of the injured and sick, and the information is compared with historical cases stored in the cloud database, and similar historical cases are identified through the learning process of medical entities and relationship information in the knowledge graph to generate a preliminary matching list;

[0083] Based on the matching result list, causal reasoning technology is used to conduct in-depth analysis of the retrieved historical case texts, explore the causal mechanism between events, identify the key causes of specific diseases, and generate a report on key pathogenic factors, including: based on the matching result list, semantic analysis and structured processing are performed through natural language analysis, possible causal associations are marked, and causal association annotation texts are generated; based on the causal association annotation text, causal reasoning technology is used to conduct in-depth analysis of the causal mechanism between events, identify and verify the causal links between different medical events, and generate a causal chain model; based on the causal chain model, further identification processing is performed to screen factors with high frequency of occurrence and strong causal indication effects, and a list of key pathogenic factors is generated; based on the list of key pathogenic factors, multi-dimensional risk assessment processing is performed to quantify the risk level and generate a report on key pathogenic factors.

[0084] In this step, external medical resources include data from other medical institutions, public health databases, or academic literature, which are used to supplement and verify existing information. These resources help the system acquire more comprehensive medical knowledge and improve diagnostic accuracy.

[0085] Multi-source data fusion processing refers to the integration of data from different sources (such as background portraits of the injured and sick, external medical resources, etc.) to generate a comprehensive information matrix. This process ensures that all relevant information is fully considered and improves the comprehensiveness and depth of data analysis.

[0086] The comprehensive patient information matrix is ​​a structured data set that contains all relevant information about the patients, including but not limited to personal information, symptom descriptions, historical medical records, and external medical resources.

[0087] The causal association annotation text is a labeled text generated by natural language analysis and structured processing of the historical case text in the matching result list. It marks the possible causal association and provides a basis for subsequent causal reasoning.

[0088] The causal chain model is a model generated by in-depth analysis of the causal mechanisms between events through causal reasoning technology. It shows the causal links between different medical events and helps identify the key causes of specific diseases.

[0089] The list of key pathogenic factors is the factors with high frequency of occurrence and strong causal indication effect screened out from the causal chain model.

[0090] Multidimensional risk assessment is a process of quantitative analysis of key pathogenic factors, evaluating the risk level of each factor, helping doctors understand which factors have a significant impact on the development of the disease, and thus developing personalized treatment plans.

[0091] In the embodiment of the present application, first, the system performs multi-source data fusion processing based on the background portrait of the injured and sick in combination with external medical resources to generate a comprehensive information matrix of the injured and sick; secondly, based on the comprehensive information matrix of the injured and sick, the system uses the medical graph enhancement learning algorithm to perform comprehensive intelligent matching processing on the information of the injured and sick, and compares it with the historical cases stored in the cloud database, and enhances the learning process through the medical entity and relationship information in the knowledge graph, identifies similar historical cases, and generates a preliminary matching list; thirdly, based on the matching result list, the system performs semantic analysis and structured processing through natural language analysis, marks possible causal associations, and generates causal association annotation text; finally, based on the causal association annotation text, the system uses causal reasoning technology to conduct in-depth analysis of the causal mechanism between events, identify and verify the causal links between different medical events, generate a causal chain model, and further screen out a list of key pathogenic factors, conduct multi-dimensional risk assessment processing, quantify the risk level, and generate a report on key pathogenic factors.

[0092] In the embodiment of the present application, it is assumed that a serious natural disaster occurs in a remote mountainous area and many villagers are injured. The rescue team quickly arrives at the scene with a portable mobile terminal and immediately initiates the emergency response procedure. First, the rescue team used mobile terminals to quickly input the personal information and preliminary symptom description of each patient and uploaded them to the central server; secondly, based on the background portrait of the patient and combined with external medical resources (such as the nearest hospital database, public health records, etc.), the system performed multi-source data fusion processing to generate a comprehensive patient information matrix; thirdly, the system used the medical graph enhancement learning algorithm to conduct comprehensive and intelligent matching processing of the patient information, compared it with the historical cases stored in the cloud database, and enhanced the learning process through the medical entity and relationship information in the knowledge graph to identify similar historical cases and generate a preliminary matching list; finally, based on the matching result list, the system performed semantic analysis and structured processing through natural language analysis, marked possible causal associations, and generated causal association annotation text; then, the system used causal reasoning technology to conduct in-depth analysis of the causal mechanism between events, identify and verify the causal links between different medical events, generate a causal chain model, and further screen out a list of key pathogenic factors, conduct multi-dimensional risk assessment processing, quantify the risk level, and generate a report on key pathogenic factors for reference by medical personnel to guide subsequent rescue work.

[0093] This application takes into account that in the prior art, since the traditional patient information matching method relies on simple keyword comparison and predefined rules, there are problems of low matching efficiency and insufficient accuracy, especially when facing a large number of complex cases, it is difficult to provide personalized and accurate diagnosis suggestions. Therefore, the embodiment of the invention proposes this optional solution, which introduces dynamic weight adjustment technology and evidence chain construction method, combined with entity and relationship information in the medical knowledge graph, to solve the above technical problems and improve the accuracy and intelligence level of patient information matching.

[0094] Based on the preliminary matching list, a dynamic weight adjustment technology is used to dynamically adjust the matching weight of each historical case to generate a dynamic adjustment result;

[0095] Based on the dynamic adjustment results, a complete evidence chain from symptoms to diagnosis is constructed through an evidence chain construction method to generate a matching result list.

[0096] Optionally, based on the comprehensive patient information matrix, a medical graph enhancement learning algorithm is used to perform comprehensive intelligent matching processing on the patient information, and the information is compared with historical cases stored in the cloud database. The learning process is enhanced through the medical entity and relationship information in the knowledge graph to identify similar historical cases and generate a preliminary matching list, including:

[0097] Based on the comprehensive patient information matrix, semantic analysis and keyword extraction are performed;

[0098] Combine geographic location and timestamp data, map them to medical entities and relationship networks, and construct structured patient information representation to generate similarity scores;

[0099] The similarity score is calculated using the following formula:

[0100]

[0101] Among them, S ij is the similarity score between the patient i and the historical case j; λ is the time difference influencing factor; TimeDiff(I i ,H j ) is the time difference between the timestamp data of patient i and historical case j, ranging from [0,1]; w 1 is the symptom similarity weight; w 2 is the context similarity weight; Sim symptom (I i ,H j ) is the symptom similarity between patient i and historical case j calculated based on the medical graph reinforcement learning algorithm; Sim context (I i ,H j ) is the contextual similarity after considering the geographic location and timestamp data; b is the bias term;

[0102] Based on the similarity score, combined with the entity and relationship information in the medical knowledge graph, the strength of the association between the injured and the patient and the historical cases is evaluated, a nonlinear adjustment factor is introduced, and the similarity score is dynamically adjusted to generate a comprehensive matching score;

[0103] The comprehensive matching score is calculated using the following formula:

[0104]

[0105] Among them, M j is the comprehensive matching score of historical case j; i is the index of the wounded and sick personnel, ranging from 1 to N; N is the number of the wounded and sick personnel; α is the weight factor of the similarity score; S ij is the similarity score between the wounded and sick personnel i and historical case j; β is the non-linear adjustment index of the similarity score; Sim entity (I i ,H j ) is the entity similarity between the wounded and sick personnel i and historical case j calculated through the medical entity information in the knowledge graph; Sim relation (I i ,H j ) is the relationship similarity between the wounded and sick personnel i and historical case j calculated through the medical relationship information in the knowledge graph; w 3 is the entity similarity weight; w 4 is the relationship similarity weight; c is the bias term; γ is the non-linear adjustment index of the comprehensive matching score; μ is the geographical difference influence factor; GeoDiff(I i ,H j ) is the geographical difference of the geographical location data between the wounded and sick personnel i and historical case j, ranging from [0, 1];

[0106] Based on the comprehensive matching score, all historical cases are sorted, screened by the diversity constraint method, and a secondary review is performed to remove redundant and irrelevant cases, generating a preliminary matching list.

[0107] This method aims to optimize the matching weight of historical cases through the dynamic weight adjustment technology, generating a dynamically adjusted result; adopts the evidence chain construction method to establish a complete evidence chain from symptoms to diagnosis, ensuring the logical coherence and reliability of the matching result; uses complex formulas to calculate the similarity score and the comprehensive matching score, ensuring that the information transmission intensity matches the closeness of the relationship between the wounded and sick personnel and historical cases; finally, minimizes the influence of redundant and irrelevant cases through the non-linear adjustment factor, continuously adjusts the weight and other parameters to converge to the optimal state, optimizes the overall matching process configuration, and prevents overfitting; improves the matching accuracy of the wounded and sick personnel information, ensures that the preliminary diagnosis suggestion can reflect the real situation of the condition, and quantifies the influence of different factors through complex formulas.

[0108] In the similarity score, the time difference influence term λ·TimeDiff(I i ,H j ): The specific reason is to consider the difference in the timestamp data of the wounded and sick personnel and historical cases, and use the sine function to smooth the influence of the time difference; the symptom similarity weighting term w 1 ·Sim symptom(I i ,H j ): The specific reason is that the symptom similarity is calculated based on the medical graph reinforcement learning algorithm, and the weight w 1 reflects the importance of the symptom in the match; the context similarity weighted term w 2 ·Sim context (I i ,H j ): The specific reason is that the context similarity after considering the geographical location and time data, the weight w 2 It reflects the importance of contextual information; bias term b: the specific reason is to adjust the baseline value of the formula to ensure the rationality of the output.

[0109] Among them, the time difference influencing factor λ is determined by experimental parameter adjustment; the time difference TimeDiff (I i ,H j ) is directly calculated from the timestamp data of the injured and the historical cases; the symptom similarity weight w 1 Obtained by training a self-supervised learning model; Symptom similarity Sim symptom (I i ,H j ) is calculated by the medical graph reinforcement learning algorithm; the context similarity weight w 2 Determined by experimental parameter adjustment; context similarity Sim context (I i ,H j ) is calculated by combining geographic location and timestamp data; the bias term b is determined by experimental parameter adjustment;

[0110] In the comprehensive matching score, the similarity score is strengthened The specific reason is to evaluate the strength of association between the injured and the historical cases. The weight factor α reflects the importance of the similarity score. The nonlinear adjustment index term The specific reason is to make nonlinear adjustments to the similarity score to reflect the importance differences of different score intervals; the entity similarity weighting term w 3 ·Sim entity (I i ,H j ): The specific reason is that the similarity is calculated through the medical entity information in the knowledge graph, and the weight w 3 Reflects the importance of entity similarity; the relation similarity weighted term w 4 ·Sim relation (I i ,H j ): The specific reason is that the similarity is calculated through the medical relationship information in the knowledge graph, and the weight w 4Reflects the importance of relationship similarity; Bias term c: To adjust the baseline value of the formula to ensure the rationality of the output; Importance difference nonlinear adjustment term The specific reason is that the comprehensive similarity is adjusted nonlinearly to reflect the importance difference of different scoring intervals; the geographical difference term μ·GeoDiff(I i ,H j ): The specific reason is that the geographical differences between the injured and the historical cases are taken into consideration, and the cosine function is used to smooth the impact of geographical differences;

[0111] Among them, the similarity score weight factor α is determined by experimental parameter adjustment; the similarity score S ij Calculated by the previous formula; the nonlinear adjustment index β is determined by experimental parameter adjustment; the entity similarity weight w 3 Obtained by training a self-supervised learning model; entity similarity Sim entity (I i ,H j ) is calculated through the medical entity information in the knowledge graph; the relationship similarity weight w 4 Obtained by training a self-supervised learning model; relation similarity Sim relation (I i ,H j ) is calculated through the medical relationship information in the knowledge graph; the bias term c is determined by experimental parameter adjustment; the nonlinear adjustment index γ is determined by experimental parameter adjustment; the geographical difference impact factor μ is determined by experimental parameter adjustment; the geographical difference GeoDiff (I i ,H j ) calculated directly from geographic location data of the sick and injured and historical cases;

[0112] Assume that a food poisoning incident occurs during a major sports event. On-site medical volunteers respond quickly and start collecting information using mobile terminals equipped with specialized medical applications. Assume that there are 5 injured and sick people (I 1 ,I 2 ,I 3 ,I 4 ,I 5 ), which needs to be compared with the historical cases stored in the cloud database (H 1 ,H 2 ,…,H m ) for comparison; assuming that the time difference influence factor λ = 0.8; the symptom similarity weight w 1 =0.7; context similarity weight w 2 =0.3; bias term b = -0.5;

[0113]

[0114] Assume that the similarity scoring weight factor α = 0.6; the non - linear adjustment index β = 1.2; the entity similarity weight w 3 = 0.5; the relationship similarity weight w 4 = 0.5; the bias term c = - 0.3; the non - linear adjustment index γ = 1.5; the geographical difference influence factor μ = 0.7;

[0115]

[0116] Assume that the set threshold is 4.0. Since the calculated result 4.2 is greater than the set threshold, it indicates that the matching scheme for the casualty information has high effectiveness and accuracy, which can ensure that the preliminary diagnosis suggestions can reflect the true situation of the condition. This is because the higher comprehensive matching score reflects that the system can effectively identify similar cases in the current situation without affecting the matching speed and accuracy. Through the above steps, the accuracy and scientific nature of the casualty information matching are ensured, the reliability and effectiveness of the rescue operation are improved, and the accuracy and response speed of the entire health monitoring system are enhanced.

[0117] 103. Based on the preliminary diagnosis suggestions, use the self - supervised learning algorithm to perform detailed feature recognition processing, automatically learn useful feature representations, adopt multi - modal fusion technology, transcribe and perform sentiment analysis on the casualty's voice description, identify sentiment clues, and generate multimedia symptom records;

[0118] In this step, the self - supervised learning algorithm is a machine learning method that can learn useful feature representations from the data itself without labeled data. It is used to capture the subtle changes in the casualty's symptoms and improve the robustness and distinctiveness of feature recognition.

[0119] The detailed feature recognition processing is the processing of standardized symptom images to ensure the consistency of image quality. It generates detailed symptom feature maps, provides more accurate symptom descriptions, and assists doctors in making more accurate diagnoses.

[0120] The multi - modal fusion technology is a technology that integrates multiple types of data (such as text, voice, and images). It supports a more comprehensive assessment of the condition and enables the system to comprehensively consider information from different sources.

[0121] The voice description transcription is the process of converting the casualty's voice description into text form through speech recognition technology, which ensures the effective utilization of voice information and enriches the content of the symptom record.

[0122] The sentiment analysis is a method of parsing the sentiment clues in the voice description through natural language processing technology. It reveals the emotional state of the patient when expressing and helps doctors better understand the patient's subjective feelings.

[0123] Multimedia symptom records are comprehensive records generated by combining symptom characteristic maps and emotional clues. They show the development and change process of the symptoms of the injured and sick, provide an intuitive display of the disease progression, and support doctors to make more comprehensive assessments.

[0124] In an embodiment of the present application, it is assumed that the emergency personnel continue to monitor the condition of the injured and sick in the ambulance, and update the symptom images and other relevant information in real time through the mobile terminal. First, the system standardizes and enhances the symptom images collected in real time according to the preliminary diagnosis suggestions to ensure the consistency of image quality; secondly, the system uses a self-supervised learning algorithm to automatically identify and extract features in the image to generate a detailed symptom feature map; thirdly, the system transcribes the voice description of the injured and sick into text through speech recognition technology, and combines sentiment analysis tools to parse its emotional clues to generate a symptom text emotional description; finally, the system uses visualization tools to display the development and change process of the symptoms of the injured and sick, and generates a multimedia symptom record so that medical personnel can intuitively understand the progression of the disease.

[0125] Optionally, in step 103, based on the preliminary diagnosis suggestion, a self-supervised learning algorithm is used to perform detailed feature recognition processing, automatically learn useful feature representations, and multimodal fusion technology is used to transcribe and sentimentally analyze the voice description of the patient, identify emotional clues, and generate a multimedia symptom record, including: based on the preliminary diagnosis suggestion, standardizing and enhancing the real-time symptom image to ensure image quality consistency and generate a standardized symptom image; based on the standardized symptom image, a self-supervised learning algorithm is used to perform detailed feature recognition processing, automatically learn useful feature representations, accurately capture the symptom characteristics of the patient, and generate a detailed symptom feature map; based on the detailed symptom feature map, multimodal fusion technology is used to convert the voice signal into text form, and emotional clues are parsed in combination with emotional analysis technology to generate a symptom text emotional description; based on the symptom text emotional description, visualization tools are used to display the development and change process of the patient's symptoms to generate a multimedia symptom record.

[0126] Among them, based on the standardized symptom image, a self-supervised learning algorithm is used to perform detailed feature recognition processing, automatically learn useful feature representations, accurately capture the symptom characteristics of the injured and sick, and generate a detailed symptom feature map, including: based on the standardized symptom image, pre-training processing is performed to extract general feature representations to ensure that the learning process is efficient and stable, and a pre-trained feature representation is generated; based on the pre-trained feature representation, a self-supervised learning algorithm is used to perform feature enhancement processing, detailed features are recognized, useful feature representations are automatically learned, and enhanced feature representations are generated; based on the enhanced feature representation, multi-scale feature extraction technology is used to capture the symptom characteristics of the injured and sick from different levels and angles, and a multi-level symptom feature set is generated; based on the multi-level symptom feature set, feature correlation analysis is performed, the correlation strength and potential impact between features are marked, and a detailed symptom feature map is generated.

[0127] In this step, real-time symptom images refer to symptom photos or videos taken in real time by mobile terminal devices during the patient's medical treatment or emergency treatment. These images are used to capture the patient's current physical condition and symptom manifestations.

[0128] Standardization and enhancement processing is the process of preprocessing real-time symptom images to ensure the quality consistency and reliability of all images, including adjusting brightness, contrast, cropping and other operations to generate standardized symptom images.

[0129] Detailed feature recognition processing is an in-depth analysis of standardized image data to identify subtle features that are critical to a specific task, which is crucial to understanding the evolution of symptoms in the injured and sick.

[0130] The detailed symptom feature map is generated by a self-supervised learning algorithm and contains a multi-level set of symptom characteristics of the injured and patients. It displays feature information at different levels and angles, helping doctors to assess the condition more comprehensively.

[0131] Multimodal fusion technology refers to the technology that integrates data from different types of sensors or input sources (such as images, audio, text) for comprehensive analysis. This technology can improve the accuracy of diagnosis and provide medical personnel with more comprehensive information.

[0132] The symptom text sentiment description is a text description generated by combining the results of sentiment analysis. It reflects the emotional state of the injured and the patient and their description of the symptoms. It provides more dimensional information support for medical personnel.

[0133] The visualization tool is a graphical interface used to display the development and changes of symptoms of the injured and sick. It intuitively presents the progression of the disease through charts, animations, etc., making it easier for doctors to track and understand.

[0134] The multimedia symptom record is a comprehensive record generated by combining standardized symptom images, detailed symptom characteristic maps and symptom text emotional descriptions. It shows the development and change process of the symptoms of the injured and provides an intuitive display of the disease progression.

[0135] In the embodiment of the present application, first, the system standardizes and enhances the real-time symptom images according to the preliminary diagnosis suggestions, ensures the consistency of image quality, and generates standardized symptom images; secondly, based on the standardized symptom images, the system uses a self-supervised learning algorithm to perform detailed feature recognition processing, automatically learns useful feature representations, accurately captures the symptom characteristics of the injured and sick, and generates a detailed symptom feature map; thirdly, based on the detailed symptom feature map, the system uses multimodal fusion technology to convert voice signals into text form, combines sentiment analysis technology to parse sentiment clues, and generates a symptom text sentiment description; finally, based on the symptom text sentiment description, the system uses visualization tools to display the development and change process of the symptoms of the injured and sick, and generates a multimedia symptom record.

[0136] In the embodiment of the present application, it is assumed that a chemical leak occurs in a large factory, and many workers are exposed to a harmful environment and experience discomfort symptoms. The rescue team responds quickly and starts information collection using a mobile terminal equipped with a special medical application. First, the rescue team uses the mobile terminal to take real-time symptom images of each injured and sick person, and simultaneously records the injured and sick person's verbal description of their symptoms; secondly, the built-in software of the mobile terminal standardizes the received symptom images, adjusts the image parameters to ensure the consistency of quality, and uses a self-supervised learning algorithm to automatically learn useful feature representations from them to generate a detailed symptom feature map; thirdly, the software transcribes the voice description of the injured and sick person into text, and uses sentiment analysis technology to identify the emotional clues therein, generates a symptom text emotional description, and further enriches the content of the multimedia symptom record; finally, all the collected information, including standardized symptom images, detailed symptom feature maps, and symptom text emotional descriptions, are organized into easy-to-understand multimedia symptom records through visualization tools for reference by medical personnel to guide subsequent rescue work.

[0137] This application takes into account that in the prior art, due to the problems of insufficient feature representation and more redundant information in traditional feature extraction methods when processing complex patient information, the accuracy and personalization of diagnostic recommendations are insufficient. Therefore, the embodiment of the invention proposes this optional solution, which introduces self-supervised learning algorithms and multimodal feature fusion technology to optimize the feature representation process and improve the accuracy and robustness of feature extraction to solve the above technical problems and ensure that the preliminary diagnostic recommendations can more accurately reflect the true condition of the disease.

[0138] Optionally, based on the pre-trained feature representation, a self-supervised learning algorithm is used to perform feature enhancement processing, identify detailed features, automatically learn useful feature representations, and generate enhanced feature representations, including:

[0139] Based on the pre-trained feature representation, a series of pseudo labels are generated for optimization, the most critical part of the symptom description in the pre-trained feature is highlighted through an attention mechanism, and similar features are grouped through clustering analysis to generate an intermediate feature representation;

[0140] The intermediate feature representation is calculated using the following formula:

[0141]

[0142] Among them, F int is the intermediate feature representation; i and j are the indexes of the context features, from 1 to M; M is the number of context features; w 1 ' is the feature similarity weight; w 2 ' is the contrast loss weight; is a pre-trained feature, indicating the similarity with the i-th context feature; F pre is the pre-trained feature representation; is the i-th context feature; is the pre-trained feature, which represents the contrast loss between the i-th feature after data enhancement; is the i-th feature after data enhancement;

[0143] Based on the intermediate feature representation, matching and fusing with multiple modal features are performed, and the weights of different modal features are dynamically adjusted by introducing a nonlinear adjustment factor, and the fusion quality is evaluated using a reconstruction loss function to generate a feature fusion representation;

[0144] The feature fusion representation is calculated using the following formula:

[0145]

[0146] Among them, F fusion is the feature fusion representation; i is the index of the multimodal feature, from 1 to N; N is the number of multimodal features; α i is the weight factor of the i-th multimodal feature; γ i is the nonlinear adjustment index of the i-th multimodal feature; w 3 ' is the feature similarity weight; w 4 ' is the reconstruction loss weight; is an intermediate feature, indicating the similarity with the i-th multimodal feature; is the i-th multimodal feature; is an intermediate feature, representing the reconstruction loss between the i-th feature reconstructed by the autoencoder; is the i-th reconstructed feature; c is the bias term;

[0147] Based on the feature fusion representation, reconstruction is performed through an autoencoder to extract higher-level abstract features, secondary optimization is performed in combination with context information, a verification mechanism is introduced to perform difference comparison, redundant and irrelevant feature components are removed, and an enhanced feature representation is generated.

[0148] This method aims to use pre-trained feature representation and perform feature enhancement processing through self-supervised learning algorithms, identify and process detailed features, automatically learn useful feature representations, and generate enhanced feature representations; use pseudo-label optimization and attention mechanisms to highlight key symptom descriptions, and group similar features through clustering analysis to generate intermediate feature representations; calculate intermediate feature representations through complex formulas to ensure that the intensity of information transmission matches the closeness of the relationship between features; introduce nonlinear adjustment factors to minimize the impact of redundant and irrelevant features, continuously adjust weights and other parameters to converge to the optimal state, optimize the overall feature fusion configuration, and prevent overfitting; finally, reconstruct and extract higher-level abstract features through autoencoder, perform secondary optimization based on contextual information, generate enhanced feature representations, and ensure the accuracy and scientificity of diagnostic recommendations.

[0149] In the intermediate feature representation, the feature similarity weighted term The specific reason is to measure the similarity between the pre-trained features and the context features, the weight w' 1 Reflects the importance of similarity; contrast loss weighted term The specific reason is to measure the contrast loss between the pre-trained features and their features after data enhancement, the weight w' 2 reflects the importance of contrast loss; the exponential decay term in the denominator The specific reason is to smooth the influence of similarity and ensure the rationality of the output;

[0150] Among them, the feature similarity weight w' 1 Determined by experimental parameter adjustment; Similarity Obtained by calculating the similarity between the pre-trained features and the context features; contrast loss weight w' 2 Determined by experimental parameter adjustment; contrast loss It is obtained by calculating the contrast loss between the pre-trained features and their features after data enhancement; the pre-trained features F pre Obtained through pre-training model; context features Extract from the wounded and sick information matrix; enhanced features Obtained by performing data augmentation on pre-trained features;

[0151] In the feature fusion representation, the feature similarity weighted term The specific reason is to measure the similarity between the intermediate features and the multimodal features, the weight w' 3 Reflects the importance of similarity; reconstruction loss weighted term The specific reason is to measure the reconstruction loss between the intermediate features and their reconstructed features, the weight w' 4 reflects the importance of reconstruction loss; bias term c: the specific reason is to smooth the impact of reconstruction loss and ensure the rationality of output; nonlinear adjustment exponential term exp(-γ i ): The specific reason is to introduce a nonlinear adjustment factor to minimize the impact of redundant and irrelevant features; the weight factor α i : The specific reason is to dynamically adjust the weights of different modal features to ensure the fusion quality;

[0152] Among them, the feature similarity weight w' 3 Determined by experimental parameter adjustment; Similarity Obtained by calculating the similarity between the intermediate features and the multimodal features; reconstruction loss weight w' 4 Determined by experimental parameter adjustment; reconstruction loss It is obtained by calculating the reconstruction loss between the intermediate feature and its reconstructed feature; the intermediate feature F int Calculated by the previous formula; multimodal features Extract from the information matrix of the injured and sick; reconstructed features Obtained by autoencoder reconstruction of intermediate features; nonlinear adjustment index γ i Determined by experimental parameter adjustment; weight factor α i Determined by experimental parameter adjustment; The bias term c is determined by experimental parameter adjustment;

[0153] Suppose a serious natural disaster occurs in a western city, and many people are injured and sent to different hospitals. It is necessary to use a mobile terminal equipped with a special medical application to start information collection; suppose there are 5 context features Need to be combined with pre-trained features F pre For comparison; assuming feature similarity weight w' 1 =0.7; contrast loss weight w' 2 =0.3;

[0154]

[0155] Assume there are 3 multimodal features Need to be related to the intermediate feature F int Fusion; Assume feature similarity weight w' 3=0.6; reconstruction loss weight w' 4 =0.4; nonlinear adjustment index γ i =0.5; weight factor α i =0.3; bias term c = -0.2;

[0156]

[0157] Assuming that the threshold is set to 0.9, since the calculated result 0.92 is greater than the set threshold, it shows that the feature fusion scheme of the patient information has high effectiveness and accuracy, which can ensure that the preliminary diagnosis recommendation can more accurately reflect the actual condition of the disease. This is because the higher feature fusion score reflects that the system can effectively extract and fuse key features under the current circumstances without affecting the quality and accuracy of feature representation. Through the above steps, the system can not only effectively extract and fuse key features in the patient information, but also significantly improve the accuracy and personalization of diagnostic recommendations. This ensures that medical personnel can obtain a more reliable and detailed analysis of the condition, so as to make more scientific treatment decisions.

[0158] 104. Based on the multimedia symptom record, create an updated record of the electronic medical record of the patient, and push it to the relevant medical institutions simultaneously, allowing medical personnel to review and confirm through mobile terminals, and generate a plan for entering the patient's information.

[0159] In this step, the electronic medical record update record converts the multimedia symptom record into a formal document in a standard format. It ensures that all medical personnel involved in the treatment can obtain the latest information on the injured and sick in a timely manner and maintain the consistency and timeliness of the information.

[0160] Synchronous push is the process of pushing updated records to the cloud server through a secure data transmission protocol. It ensures that information can be efficiently transmitted between different medical institutions and supports real-time collaboration.

[0161] Relevant medical institutions refer to all medical institutions involved in the treatment of the wounded and sick, including emergency centers, hospitals, clinics, etc. They obtain the latest information on the wounded and sick through synchronous push to ensure the continuity of medical services.

[0162] Review and confirmation is the process in which medical personnel log into the system through mobile terminals to review and confirm the latest medical information of the injured and sick. It ensures the integrity and accuracy of the information and provides a reliable basis for subsequent treatment.

[0163] The plan for entering information on the injured and sick is a formal plan generated based on the feedback from medical staff. It guides subsequent treatment and nursing work, ensures the quality and continuity of medical services, and ensures that each link is properly handled.

[0164] In the embodiment of the present application, assuming that the ambulance generates a multimedia symptom record when arriving at the hospital, first, the system converts the generated multimedia symptom record into an electronic medical record update record in a standard format, and pushes it to the cloud server through a secure data transmission protocol; secondly, the system notifies the emergency department of the hospital so that it can access the latest information immediately; thirdly, the medical staff logs into the system through the mobile terminal to review and confirm the latest condition information of the injured and sick to ensure the completeness and accuracy of the information; finally, the system generates a formal information entry plan for the injured and sick based on the feedback from the medical staff to ensure that each link is properly handled and the quality and continuity of medical services are guaranteed. This not only improves the efficiency of treatment, but also ensures the accuracy and timeliness of information transmission.

[0165] Optionally, in step 104, based on the multimedia symptom record, an updated record of the electronic medical record of the injured or sick is created, and the record is pushed synchronously to relevant medical institutions, allowing medical personnel to review and confirm through mobile terminals, and generating an information entry plan for the injured or sick, including: based on the multimedia symptom record, parsing visual and text information, extracting key symptom descriptions and preliminary diagnostic suggestions, performing standardized format conversion, and generating structured diagnostic information; based on the structured diagnostic information, updating and processing the existing electronic medical record data of the injured or sick, integrating the latest medical assessment, and generating an updated electronic medical record version; based on the updated electronic medical record version, synchronously pushing it to the cloud server and the database of relevant medical institutions through a secure data transmission protocol to generate a synchronous push record; based on the synchronous push record, constructing a review and confirmation process, allowing medical personnel to access the latest information of the injured or sick through mobile terminals for review and confirmation, and generating an information entry plan for the injured or sick.

[0166] In this step, visual and text information parsing refers to analyzing the image and text data in the multimedia symptom records to extract key symptom descriptions and preliminary diagnostic suggestions.

[0167] Key symptom descriptions are specific symptom manifestations extracted from multimedia symptom records, such as redness, swelling, pain, etc. They are the basis for doctors to assess the condition and help formulate treatment plans.

[0168] The preliminary diagnostic recommendation is a diagnostic opinion generated based on the report of key pathogenic factors, providing an important reference for medical personnel.

[0169] Standardized format conversion refers to converting the extracted key symptom descriptions and preliminary diagnostic recommendations into a unified standard format to ensure information interoperability between different medical institutions.

[0170] Structured diagnostic information is generated by conversion of a standardized format and is a structured data set containing all important symptoms and diagnostic recommendations of the injured or sick, which is used to update existing electronic medical record data.

[0171] The latest medical assessment refers to a comprehensive evaluation of the current health status of the injured or sick person, combined with the latest collected symptom information and diagnosis results. It is used to generate an updated version of the electronic medical record.

[0172] The updated electronic medical record version is a new version of the electronic medical record generated after integrating the latest medical assessment, reflecting the latest condition and diagnosis information of the injured or sick.

[0173] The secure data transmission protocol refers to a communication protocol that uses encryption and other security measures to ensure that data is not tampered with or leaked during transmission. It is used to synchronize data to cloud servers and related medical institution databases.

[0174] The synchronization push record is a log file generated by the system after completing data transmission. It records information such as the push time, content, and recipient. It is used to track and verify the data synchronization process.

[0175] The review and confirmation process allows medical personnel to access the latest information of injured and sick patients through mobile terminals for review and confirmation. It ensures the integrity and accuracy of the information and provides a reliable basis for subsequent treatment.

[0176] In the embodiment of the present application, first, the system parses visual and text information based on multimedia symptom records, extracts key symptom descriptions and preliminary diagnostic suggestions, performs standardized format conversion, and generates structured diagnostic information; secondly, based on the structured diagnostic information, the system updates and processes the existing electronic medical record data of the injured and sick, integrates the latest medical assessments, and generates an updated electronic medical record version; thirdly, based on the updated electronic medical record version, the system synchronously pushes it to the cloud server and the relevant medical institution database through a secure data transmission protocol, and generates a synchronous push record; finally, based on the synchronous push record, the system constructs a review and confirmation process, allowing medical personnel to access the latest information on the injured and sick through mobile terminals for review and confirmation, and generates an information entry plan for the injured and sick.

[0177] Suppose that a sudden infectious disease outbreak occurs in a remote mountainous area, and many villagers develop unexplained fever and respiratory symptoms. Local medical resources are limited and telemedicine support is urgently needed. First, on-site medical volunteers use mobile terminals equipped with special medical applications to collect symptom images of each patient in real time, and simultaneously record the patient's verbal description of their symptoms. Second, the built-in software of the mobile terminal standardizes the received symptom images, adjusts the image parameters to ensure quality consistency, and parses visual and text information, extracts key symptom descriptions and preliminary diagnostic suggestions, performs standardized format conversion, and generates structured diagnostic information. Third, based on the generated structured diagnostic information, the system updates and processes the existing electronic medical record data of the patient, integrates the latest medical assessment, and generates an updated electronic medical record version. Finally, the system pushes the updated electronic medical record version to the cloud server and the database of relevant medical institutions through a secure data transmission protocol, and generates a synchronization push record. Based on the synchronization push record, the system builds a review and confirmation process, allowing professional medical personnel far away in the city to access the latest patient information through mobile terminals, review and confirm, generate a patient information entry plan, and guide subsequent treatment work.

[0178] In summary, steps 101 to 104 cover the complete process from the initial collection of patient information to feature enhancement and fusion, aiming to provide an efficient and accurate intelligent diagnosis assistance system to meet the needs of rapid response and personalized treatment plans in medical scenarios. This series of steps ensures that each patient can obtain the most suitable diagnostic advice by optimizing the feature representation and matching process, while improving the efficiency and quality of medical services, and adapting to the requirements for intelligent and data-driven solutions in modern medical environments.

[0179] Figure 2 A schematic diagram of the structure of a system for entering information of injured and sick persons based on a mobile terminal is provided for an embodiment of the present application, such as Figure 2 As shown, the device comprises:

[0180] The collection module 21 is used to collect basic information of the injured and sick through the user's mobile terminal, and generate a basic information framework of the injured and sick in combination with the geographical location and timestamp data;

[0181] The analysis module 22 is used to use the medical graph enhancement learning algorithm based on the basic information framework of the injured and sick, compare with the historical cases stored in the cloud database, use the knowledge graph to enhance the learning process, adopt causal reasoning technology, perform causal relationship analysis on the historical case text, extract key factors causing the disease, and generate preliminary diagnosis suggestions;

[0182] The processing module 23 is used to perform detailed feature recognition processing based on the preliminary diagnosis suggestion, use a self-supervised learning algorithm, automatically learn useful feature representations, use multimodal fusion technology to transcribe and perform sentiment analysis on the patient's voice description, identify sentiment clues, and generate a multimedia symptom record;

[0183] The creation module 24 is used to create an updated record of the electronic medical record of the patient based on the multimedia symptom record, and push it to the relevant medical institutions simultaneously, allowing medical personnel to review and confirm through the mobile terminal, and generate an information entry plan for the patient.

[0184] Figure 2 The mobile terminal-based patient information entry system can be executed Figure 1 The implementation principle and technical effect of the method for entering the information of the injured and sick based on a mobile terminal described in the embodiment shown are not described in detail. The specific manner in which each module and unit performs operations in the above embodiment of the system for entering the information of the injured and sick based on a mobile terminal has been described in detail in the embodiment of the method, and will not be described in detail here.

[0185] In one possible design, Figure 2 The mobile terminal-based patient information entry system of the illustrated embodiment can be implemented as a computing device, such as Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32;

[0186] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32 .

[0187] The processing component 32 is used to: collect basic information of the injured and sick through the user's mobile terminal, and generate a basic information framework of the injured and sick in combination with the geographic location and timestamp data; based on the basic information framework of the injured and sick, use the medical graph enhancement learning algorithm to compare with the historical cases stored in the cloud database, use the knowledge graph to enhance the learning process, use causal reasoning technology to perform causal relationship analysis on the historical case text, extract key factors causing the disease, and generate preliminary diagnosis suggestions; based on the preliminary diagnosis suggestions, use the self-supervised learning algorithm to perform detailed feature recognition processing, automatically learn useful feature representations, use multimodal fusion technology to transcribe and sentimentally analyze the voice description of the injured and sick, identify emotional clues, and generate multimedia symptom records; based on the multimedia symptom records, create electronic medical record update records for the injured and sick, and simultaneously push them to relevant medical institutions, allowing medical personnel to review and confirm through mobile terminals, and generate information entry plans for the injured and sick.

[0188] The processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component may also be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components to perform the above method.

[0189] The storage component 31 is configured to store various types of data to support operations at the terminal. The storage component 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 read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0190] Of course, the computing device may also include other components, such as input / output interfaces, display components, communication components, etc.

[0191] The input / output interface provides an interface between the processing component and the peripheral interface module, which may be an output device, an input device, etc.

[0192] The communication component is configured to facilitate, among other things, wired or wireless communications between the computing device and other devices.

[0193] Among them, the computing device can be a physical device or an elastic computing host provided by a cloud computing platform, etc. In this case, the computing device can refer to a cloud server, and the above-mentioned processing components, storage components, etc. can be basic server resources rented or purchased from the cloud computing platform.

[0194] The present application also provides a computer storage medium storing a computer program, wherein the computer program can achieve the above-mentioned Figure 1 The illustrated embodiment is a method for entering information of injured and sick persons based on a mobile terminal.

[0195] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0196] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0197] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0198] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for entering information of injured and sick persons based on a mobile terminal, characterized in that: include: Collect basic information of the injured and sick through the user's mobile terminal, and generate a basic information framework of the injured and sick by combining the geographic location and timestamp data; Based on the basic information framework of the injured and sick, the medical graph-enhanced learning algorithm is used to compare with the historical cases stored in the cloud database, the knowledge graph is used to enhance the learning process, and the causal reasoning technology is used to analyze the causal relationship of the historical case text, extract the key factors causing the disease, and generate preliminary diagnosis suggestions; Based on the preliminary diagnostic suggestions, a self-supervised learning algorithm is used to perform detailed feature recognition processing, automatically learn useful feature representations, and use multimodal fusion technology to transcribe and perform sentiment analysis on the patient's voice descriptions, identify sentiment clues, and generate multimedia symptom records; Based on the multimedia symptom record, an updated record of the electronic medical record of the patient is created and simultaneously pushed to relevant medical institutions, allowing medical personnel to review and confirm through mobile terminals and generate a plan for entering the patient's information.

2. The method according to claim 1, characterized in that Based on the basic information framework of the injured and sick, the medical graph enhancement learning algorithm is used to compare with the historical cases stored in the cloud database, the knowledge graph is used to enhance the learning process, and the causal reasoning technology is used to analyze the causal relationship of the historical case text, extract the key factors causing the disease, and generate preliminary diagnosis suggestions, including: Based on the basic information framework of the wounded and sick, semantic parsing and context understanding processing are performed to capture surface meanings and implicit associations, and generate background portraits of the wounded and sick; Based on the background portrait of the injured and sick, the medical graph enhanced learning algorithm is used to perform comprehensive intelligent matching processing on the injured and sick information, and compared with the historical cases stored in the cloud database, and the medical entity and relationship information in the knowledge graph is used to enhance the learning process to generate a matching result list; Based on the matching result list, causal reasoning technology is used to conduct in-depth analysis of the retrieved historical case texts, explore the causal mechanism between events, identify the key causes of specific diseases, and generate a report on key pathogenic factors; Based on the report on key pathogenic factors, an expert system rule engine is used to perform optimization processing and generate preliminary diagnostic recommendations.

3. The method according to claim 2, characterized in that Based on the background portrait of the patient, the medical graph enhancement learning algorithm is used to perform comprehensive intelligent matching processing on the patient information, compare it with the historical cases stored in the cloud database, and use the medical entity and relationship information in the knowledge graph to enhance the learning process and generate a matching result list, including: Based on the background portraits of the injured and sick, combined with external medical resources, multi-source data fusion processing is performed to generate a comprehensive information matrix of the injured and sick; Based on the comprehensive patient information matrix, the medical graph enhanced learning algorithm is used to perform comprehensive intelligent matching processing on the patient information, and compared with the historical cases stored in the cloud database. The learning process is enhanced by the medical entity and relationship information in the knowledge graph to identify similar historical cases and generate a preliminary matching list. Based on the preliminary matching list, a dynamic weight adjustment technology is used to dynamically adjust the matching weight of each historical case to generate a dynamic adjustment result; Based on the dynamic adjustment results, a complete evidence chain from symptoms to diagnosis is constructed through an evidence chain construction method to generate a matching result list.

4. The method according to claim 2, characterized in that: Based on the matching result list, causal reasoning technology is used to deeply analyze the retrieved historical case texts, explore the causal mechanism between events, identify the key causes of specific diseases, and generate a report on key pathogenic factors, including: Based on the matching result list, semantic analysis and structural processing are performed through natural language analysis to mark possible causal relationships and generate causal relationship annotation text; Based on the causal association annotated text, causal reasoning technology is used to deeply analyze the causal mechanism between events, identify and verify the causal links between different medical events, and generate a causal chain model; Based on the causal chain model, further identification processing is performed to screen factors with high frequency of occurrence and strong causal indication effects, and a list of key pathogenic factors is generated; Based on the list of key pathogenic factors, a multi-dimensional risk assessment process is performed to quantify the risk level and generate a key pathogenic factor report.

5. The method according to claim 1, characterized in that Based on the preliminary diagnosis suggestion, the self-supervised learning algorithm is used to perform detailed feature recognition processing, automatically learn useful feature representation, and adopt multimodal fusion technology to transcribe and analyze the voice description of the injured and sick, identify emotional clues, and generate multimedia symptom records, including: Based on the preliminary diagnostic suggestions, the real-time symptom images are standardized and enhanced to ensure image quality consistency and generate standardized symptom images; Based on the standardized symptom images, a self-supervised learning algorithm is used to perform detailed feature recognition processing, automatically learn useful feature representations, accurately capture the symptom characteristics of the injured and sick, and generate a detailed symptom feature map; Based on the detailed symptom feature map, multimodal fusion technology is used to convert the voice signal into text form, and sentiment analysis technology is used to analyze the sentiment clues to generate a symptom text sentiment description; Based on the emotional description of the symptom text, visualization tools are used to display the development and change process of the patient's symptoms and generate a multimedia symptom record.

6. The method according to claim 5, characterized in that Based on the standardized symptom images, a self-supervised learning algorithm is used to perform detailed feature recognition processing, automatically learn useful feature representations, accurately capture the symptom characteristics of the injured and sick, and generate a detailed symptom feature map, including: Based on the standardized symptom images, pre-training processing is performed to extract general feature representations, ensure that the learning process is efficient and stable, and generate pre-trained feature representations; Based on the pre-trained feature representation, a self-supervised learning algorithm is used to perform feature enhancement processing, identify detailed features, automatically learn useful feature representations, and generate enhanced feature representations; Based on the enhanced feature representation, the symptom characteristics of the injured and sick are captured from different levels and angles through multi-scale feature extraction technology to generate a multi-level symptom feature set; Based on the multi-level symptom feature set, feature association analysis is performed, the strength of association and potential impact between features are marked, and a detailed symptom feature map is generated.

7. The method according to claim 1, characterized in that Based on the multimedia symptom record, the electronic medical record update record of the patient is created and pushed to the relevant medical institution simultaneously, allowing medical personnel to review and confirm through the mobile terminal, and generating the patient information entry plan, including: Based on the multimedia symptom record, parsing visual and text information, extracting key symptom descriptions and preliminary diagnostic suggestions, performing standardized format conversion, and generating structured diagnostic information; Based on the structured diagnostic information, the existing electronic medical record data of the patient is updated and processed, and the latest medical assessment is integrated to generate an updated electronic medical record version; Based on the updated electronic medical record version, the updated electronic medical record version is synchronously pushed to the cloud server and the database of the relevant medical institution through a secure data transmission protocol to generate a synchronous push record; Based on the synchronous push records, a review and confirmation process is constructed to allow medical personnel to access the latest patient information through mobile terminals for review and confirmation, and to generate a plan for entering patient information.

8. A system for entering information of injured and sick persons based on a mobile terminal, characterized in that: include: The collection module is used to collect basic information of the injured and sick through the user's mobile terminal, and generate a basic information framework of the injured and sick by combining the geographic location and timestamp data; An analysis module is used to compare the basic information framework of the injured and sick with the historical cases stored in the cloud database using the medical graph enhancement learning algorithm, use the knowledge graph to enhance the learning process, and use causal reasoning technology to perform causal relationship analysis on the historical case text, extract key factors causing the disease, and generate preliminary diagnosis suggestions; A processing module, which is used to perform detailed feature recognition processing based on the preliminary diagnosis suggestion, use a self-supervised learning algorithm, automatically learn useful feature representations, use multimodal fusion technology to transcribe and perform sentiment analysis on the voice description of the patient, identify sentiment clues, and generate a multimedia symptom record; A creation module is used to create an updated electronic medical record of the patient based on the multimedia symptom record, and push it to relevant medical institutions simultaneously, allowing medical personnel to review and confirm through mobile terminals and generate an information entry plan for the patient.

9. A computing device, characterized in that It comprises a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a method for entering information of injured and sick persons based on a mobile terminal as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that: A computer program is stored, and when the computer program is executed by a computer, a method for entering information of injured and sick persons based on a mobile terminal as described in any one of claims 1 to 7 is implemented.

Images