Multi-agent framework insurance past symptom key information structured extraction system

Abstract

The invention discloses a multi-agent architecture-based previous insurance symptom key information structured extraction system, which comprises a previous insurance symptom key information extraction system and a multi-agent architecture-based previous insurance symptom key information extraction system, the previous insurance symptom key information extraction system comprises a data preprocessing module, a task allocation and management module, a disease diagnosis and recognition module, a treatment record extraction module, an information integration and structuring module and a feedback optimization module. Each module is independently responsible for a specific function, the flexibility and expandability of the system are ensured, natural language processing and machine learning are applied, semantic information of a medical text can be deeply understood, key information can be accurately recognized and extracted, information sharing and cooperative work are achieved through an efficient communication mechanism, and the system is high in practicability. Complex processing tasks are coped together, and the cooperative mode not only improves the processing efficiency, but also enhances the overall performance and robustness of the system.

Description

Technical Field

[0001] The present invention relates to the technical field of medical systems, and more specifically to a structured extraction system for key information of insurance history with a multi-agent architecture. Background Art

[0002] In today's highly digitized insurance industry, it is critical to quickly and accurately process large volumes of medical records to identify pre-existing condition information relevant to insurance applications.

[0003] In the past, the extraction of key information of past medical history had a low degree of automation and needed to rely on manual review, which had low accuracy and increased the burden of manual review. Therefore, a new technical solution is needed to solve the problem. Summary of the invention

[0004] The purpose of the present invention is to provide a structured extraction system for key information of insurance medical records with a multi-agent architecture, which solves the problem that the extraction of key information of medical records has a low degree of automation, requires manual review, has low accuracy, and increases the burden of manual review.

[0005] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: a multi-agent architecture insurance history key information structured extraction system, comprising: insurance history key information extraction system, the insurance history key information extraction system including a data preprocessing module, a task allocation and management module, a disease diagnosis and identification module, a treatment record extraction module, an information integration and structuring module and a feedback optimization module;

[0006] The data preprocessing module is responsible for receiving raw medical data such as electronic medical records, diagnosis books, examination reports, etc., and performing preprocessing such as cleaning, formatting, and denoising to unify the data format in preparation for subsequent intelligent agent processing;

[0007] The task allocation and management module dynamically allocates tasks to different agents according to task requirements and data characteristics, and is responsible for coordination and communication between agents;

[0008] The disease diagnosis and identification module uses natural language processing and machine learning technology to identify key information such as the specific disease name, type, and diagnosis time from the text;

[0009] The treatment record extraction module extracts information related to disease treatment from medical records, such as treatment methods, medication records, and surgical conditions;

[0010] The information integration and structuring module integrates the information extracted by each module to form a structured record of medical history data, and outputs the structured data in a database, JSON, XML or other format for use by insurance companies or other systems;

[0011] The feedback optimization module collects user feedback, evaluates system performance, and continuously optimizes the processing capabilities and accuracy of each module and agent.

[0012] As a preferred embodiment of the present invention, the data preprocessing module includes a data acquisition module and a preprocessing module. The data acquisition module collects data including medical institution systems, public databases and resources, and data directly provided by patients. The preprocessing module includes data cleaning, text segmentation and text standards.

[0013] As a preferred implementation mode of the present invention, the task allocation and management module includes task allocation, resource management and communication modules.

[0014] As a preferred embodiment of the present invention, the disease diagnosis and identification module includes named entity recognition, timestamp extraction, disease classification and disease map establishment, and the disease map establishment includes symptom map, treatment method map, drug map, examination item map, complication map and cause map.

[0015] As a preferred embodiment of the present invention, the treatment record extraction module includes medication information extraction, treatment effect evaluation and information integration and verification. The medication information extraction includes drug name identification and dosage and medication cycle. The treatment effect evaluation includes symptom change identification and quantitative indicator extraction.

[0016] As a preferred implementation mode of the present invention, the information integration and structuring module includes data association, data verification and result output.

[0017] As a preferred implementation mode of the present invention, the feedback and optimization module includes user feedback collection, performance evaluation and model optimization.

[0018] Compared with the prior art, the present invention has the following beneficial effects:

[0019] The insurance history key information extraction system of the present invention includes a data preprocessing module, a task allocation and management module, a disease diagnosis and identification module, a treatment record extraction module, an information integration and structuring module and a feedback optimization module. The data preprocessing module is responsible for receiving original medical data such as electronic medical records, diagnosis books, examination reports, etc., and performing preprocessing work such as cleaning, formatting, and denoising to unify the data format and prepare for subsequent intelligent agent processing. The task allocation and management module dynamically allocates tasks to different intelligent agents according to task requirements and data characteristics, and is responsible for coordination and communication between intelligent agents. The disease diagnosis and identification module uses natural language processing and machine learning technology to identify key information such as specific disease name, type, and diagnosis time from the text. The treatment record extraction module extracts information related to disease treatment from the medical record. , such as treatment methods, medication records, and surgical conditions. The information integration and structuring module integrates the information extracted by each module to form a structured record of past medical data, and outputs the structured data in database, JSON, XML and other formats for use by insurance companies or other systems. The feedback optimization module collects user feedback, evaluates system performance, and continuously optimizes the processing capabilities and accuracy of each module and intelligent agent. Each module of this system is independently responsible for specific functions, ensuring the flexibility and scalability of the system. The use of natural language processing and machine learning can deeply understand the semantic information of medical texts, accurately identify and extract key information, and realize information sharing and collaborative work through efficient communication mechanisms to jointly cope with complex processing tasks. The collaborative mode not only improves processing efficiency, but also enhances the overall performance and robustness of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a schematic diagram of the overall process of the present invention;

[0021] Figure 2 This is a flow chart of the data preprocessing module of the present invention;

[0022] Figure 3 This is a flowchart of the task allocation and management module of the present invention;

[0023] Figure 4 This is a flow chart of the disease diagnosis and identification module of the present invention;

[0024] Figure 5 This is a flow chart of the treatment record extraction module of the present invention;

[0025] Figure 6 The information integration and structured module process intention of the present invention;

[0026] Figure 7 It is a schematic diagram of the feedback and optimization module flow of the present invention. DETAILED DESCRIPTION

[0027] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0028] See also Figure 1-7 , the present invention provides a technical solution: a multi-agent architecture insurance history key information structured extraction system, including: insurance history key information extraction system, the insurance history key information extraction system includes a data preprocessing module, a task allocation and management module, a disease diagnosis and identification module, a treatment record extraction module, an information integration and structuring module and a feedback optimization module;

[0029] The data preprocessing module is responsible for receiving raw medical data such as electronic medical records, diagnosis books, examination reports, etc., and performing preprocessing such as cleaning, formatting, and denoising to unify the data format in preparation for subsequent intelligent agent processing;

[0030] The task allocation and management module dynamically allocates tasks to different agents according to task requirements and data characteristics, and is responsible for coordination and communication between agents;

[0031] The disease diagnosis and identification module uses natural language processing and machine learning technology to identify key information such as the specific disease name, type, and diagnosis time from the text;

[0032] The treatment record extraction module extracts information related to disease treatment from medical records, such as treatment methods, medication records, and surgical conditions;

[0033] The information integration and structuring module integrates the information extracted by each module to form a structured record of medical history data, and outputs the structured data in a database, JSON, XML or other format for use by insurance companies or other systems;

[0034] The feedback optimization module collects user feedback, evaluates system performance, and continuously optimizes the processing capabilities and accuracy of each module and agent.

[0035] Further improvements, such as Figure 2As shown: the data preprocessing module includes a data acquisition module and a preprocessing module. The data acquisition module collects data from medical institution systems, public databases and resources, and data directly provided by patients. The preprocessing module includes data cleaning, text segmentation, and text standards. The data acquisition module collects data from medical institution systems. It is necessary to establish cooperative relationships with major hospitals, clinics and other medical institutions to obtain data interface permissions for their electronic medical record systems, diagnostic report systems, and examination report systems. The data source usually includes comprehensive information such as the patient's basic information, diagnostic records, treatment process, and examination results.

[0036] Public databases and resources include public database resources from the National Health Commission, medical research institutions, etc. These resources may contain standardized data on common and frequently occurring diseases, which can be used to assist and supplement private data sources;

[0037] In special cases, patients may also be allowed to directly upload or mail paper medical records, diagnosis books and other documents and digitize them;

[0038] Data cleaning removes invalid information, non-text data, and content with inconsistent formats; deletes non-medical content such as headers, footers, comments, and copyright notices in the text, standardizes the use of punctuation, such as using English punctuation uniformly, removes redundant punctuation or converts it to a consistent format, removes non-ASCII characters and emoticons in the text, and removes duplicate paragraphs or sentences if there are duplicate paragraphs or sentences in the document;

[0039] Text segmentation uses a special segmentation tool in the medical field to segment the text. For Chinese text, a Chinese segmentation algorithm is used to divide sentences into vocabulary units. Commonly used segmentation tools include jieba, HanLP, etc. They support custom dictionaries to improve the accuracy of segmentation.

[0040] The word segmentation of English text is relatively simple, usually based on spaces and punctuation marks, but more complex algorithms may be needed to identify abbreviations and hyphenated words;

[0041] Text standards convert synonyms, abbreviations, and irregular terms into standard terms, and convert vocabulary in the text into standard forms, such as expanding abbreviations into full names and unifying different forms of the same word; converting all words in the text into lowercase or first letter capitalized formats to reduce the complexity of model learning.

[0042] Further improvements, such as Figure 3 As shown: the task allocation and management module includes task allocation, resource management and communication modules;

[0043] Format numbers in text into a unified format, such as converting dates, times, numbers, etc. into standardized representations.

[0044] The task assignment module analyzes the data characteristics, divides the tasks into different subtasks such as disease diagnosis and identification, treatment record extraction, and disease course tracking, and assigns them to the corresponding processing modules. The system needs to be able to parse and understand the newly arrived tasks, including reading task descriptions from the database, identifying task types (such as text classification, information extraction, etc.), and evaluating the complexity of tasks and the required resources; and prioritize tasks according to their importance, urgency, and expected completion time. Priorities can be based on business rules, system policies, or user-defined weights.

[0045] The resource management module monitors the processing status of each agent, reasonably allocates computing resources, ensures the overall efficient operation of the system, and tries to assign tasks with the highest priority to agents with lower current load and matching capabilities. If no suitable agent is found, wait or re-evaluate the task priority;

[0046] The communication module establishes the communication mechanism of the system to promote information sharing and collaborative processing of complex problems.

[0047] Further improvements, such as Figure 4 As shown: the disease diagnosis and identification module includes named entity recognition, timestamp extraction, disease classification and disease map establishment, and the disease map establishment includes symptom map, treatment method map, drug map, examination item map, complication map and cause map;

[0048] The named entity recognition module identifies disease names in text;

[0049] The timestamp extraction module extracts disease-related time information from the text, such as diagnosis time and treatment time;

[0050] The disease classification module classifies the identified diseases according to ICD codes or other classification systems;

[0051] Symptom maps describe the relationship between different diseases and their typical symptoms, which helps to quickly locate possible diseases based on symptoms;

[0052] The treatment method map shows the treatment means and methods corresponding to various diseases, including drug treatment, surgical treatment, physical therapy, etc.

[0053] The drug atlas records which drugs are used to treat which diseases, as well as information such as drug usage, dosage, and side effects;

[0054] The examination item map lists the examination items required for different diseases, such as blood tests, imaging tests, etc., to assist in diagnosis;

[0055] The complication map describes the possible complications of a disease and their probability, helping to assess disease risk;

[0056] The etiology map analyzes which diseases or factors may cause different symptoms, assisting doctors in making differential diagnoses.

[0057] Further improvements, such as Figure 5 As shown: the treatment record extraction module includes medication information extraction, treatment effect evaluation and information integration and verification. The medication information extraction includes drug name identification and dosage and medication cycle. The treatment effect evaluation includes symptom change identification and quantitative index extraction.

[0058] The information integration and verification module integrates the extracted key information such as treatment items, medication information, and treatment effects to form a complete treatment record, and ensures the accuracy and consistency of the extracted information through cross-validation or logical checking methods;

[0059] The drug name recognition module uses a drug-specific NER model to identify drug names in drug treatment records;

[0060] Dosage and medication cycle: Use regular expression matching, rule extraction, or machine learning model prediction to extract specific information such as drug dosage and medication cycle (e.g., once a day, two tablets each time) from the text;

[0061] The symptom change identification module analyzes the symptom changes described in the text, such as "pain relief", "stable condition", etc., to evaluate the treatment effect;

[0062] The quantitative index extraction module extracts quantitative indicators (such as blood pressure values, blood sugar levels, etc.) from the text, and compares them with the values ​​before and after treatment to evaluate the treatment effect.

[0063] Further improvements, such as Figure 6 As shown: the information integration and structuring module includes data association, data verification and result output;

[0064] Data association links data from different sources based on information such as timeline and disease type.

[0065] Data validation ensures the accuracy and consistency of extracted information through cross-validation or logical checking. After data integration, an integrity check is performed to ensure that required fields are not empty and that key data items are not repeated or omitted. Through cross-module or cross-data source cross-validation, it is ensured that the integrated data is logically consistent, such as matching task assignment records with agent execution records.

[0066] Result output: The structured data is output in database, JSON, XML and other formats for use by insurance companies or other systems.

[0067] During the data integration process, when it is found that multiple data sources have inconsistent descriptions of the same entity or attribute, conflict identification is performed.

[0068] Conflict resolution strategy: Priority strategy: Resolve conflicts based on the credibility of the data source or preset priority rules.

[0069] Aggregation strategy: For quantifiable data (such as scores, quantities, etc.), statistical methods such as mean and median can be used to synthesize the values ​​of multiple data sources.

[0070] Negotiation strategy: For complex conflicts, human intervention or negotiation through intelligent algorithms may be required.

[0071] Further improvements, such as Figure 7 As shown: the feedback and optimization module includes user feedback collection, performance evaluation and model optimization;

[0072] The user feedback collection module collects user feedback through questionnaires, user interviews, etc.

[0073] The performance evaluation module regularly evaluates key indicators such as the system's processing speed, accuracy, and recall rate.

[0074] Working principle: The insurance history key information extraction system includes data preprocessing module, task allocation and management module, disease diagnosis and recognition module, treatment record extraction module, information integration and structuring module and feedback optimization module. The data preprocessing module is responsible for receiving original medical data such as electronic medical records, diagnosis books, examination reports, etc., and performing preprocessing work such as cleaning, formatting, and denoising to unify the data format and prepare for subsequent intelligent agent processing. The task allocation and management module dynamically assigns tasks to different intelligent agents according to task requirements and data characteristics, and is responsible for coordination and communication between intelligent agents. The disease diagnosis and recognition module uses natural language processing and machine learning technology to identify key information such as specific disease name, type, and diagnosis time from the text. The treatment record extraction module extracts information related to disease treatment from medical records. The information integration and structuring module integrates the information extracted by each module to form a structured record of medical history data, and outputs the structured data in database, JSON, XML and other formats for use by insurance companies or other systems. The feedback optimization module collects user feedback, evaluates system performance, and continuously optimizes the processing capabilities and accuracy of each module and intelligent agent. Each module of this system is independently responsible for specific functions, ensuring the flexibility and scalability of the system. The use of natural language processing and machine learning can deeply understand the semantic information of medical texts, accurately identify and extract key information, and realize information sharing and collaborative work through efficient communication mechanisms to jointly cope with complex processing tasks. The collaborative mode not only improves processing efficiency, but also enhances the overall performance and robustness of the system.

[0075] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic features of the present invention. Therefore, no matter from which point of view, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the attached claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any figure mark in the claims should not be regarded as limiting the claims involved.

[0076] Finally, a few points should be explained: First, in the description of this application, it should be noted that, unless otherwise specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, and can be mechanical or electrical connections, or internal connectivity between two components, or direct connections. "Up", "down", "left", "right", etc. are only used to indicate relative position relationships. When the absolute position of the object being described changes, the relative position relationship may change.

[0077] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A multi-agent architecture insurance history key information structured extraction system, characterized by: include: A system for extracting key information about insurance history, wherein the system includes a data preprocessing module, a task allocation and management module, a disease diagnosis and identification module, a treatment record extraction module, an information integration and structuring module, and a feedback optimization module; The data preprocessing module is responsible for receiving raw medical data such as electronic medical records, diagnosis books, examination reports, etc., and performing preprocessing such as cleaning, formatting, and denoising to unify the data format in preparation for subsequent intelligent agent processing; The task allocation and management module dynamically allocates tasks to different agents according to task requirements and data characteristics, and is responsible for coordination and communication between agents; The disease diagnosis and identification module uses natural language processing and machine learning technology to identify key information such as the specific disease name, type, and diagnosis time from the text; The treatment record extraction module extracts information related to disease treatment from medical records, such as treatment methods, medication records, and surgical conditions; The information integration and structuring module integrates the information extracted by each module to form a structured record of medical history data, and outputs the structured data in a database, JSON, XML or other format for use by insurance companies or other systems; The feedback optimization module collects user feedback, evaluates system performance, and continuously optimizes the processing capabilities and accuracy of each module and agent.

2. According to the multi-agent architecture insurance pre-existing condition key information structured extraction system of claim 1, it is characterized by: The data preprocessing module includes a data acquisition module and a preprocessing module. The data acquisition module collects data including medical institution systems, public databases and resources, and data directly provided by patients. The preprocessing module includes data cleaning, text segmentation, and text standards.

3. According to the multi-agent architecture insurance pre-existing condition key information structured extraction system of claim 1, it is characterized by: The task allocation and management module includes task allocation, resource management and communication modules.

4. The multi-agent architecture insurance pre-existing condition key information structured extraction system according to claim 1, characterized in that: The disease diagnosis and identification module includes named entity recognition, timestamp extraction, disease classification and disease map establishment, and the disease map establishment includes symptom map, treatment method map, drug map, examination item map, complication map and cause map.

5. The multi-agent architecture insurance history key information structured extraction system according to claim 1, characterized in that: The treatment record extraction module includes medication information extraction, treatment effect evaluation, and information integration and verification. The medication information extraction includes drug name identification, dosage, and medication cycle. The treatment effect evaluation includes symptom change identification and quantitative indicator extraction.

6. The multi-agent architecture insurance history key information structured extraction system according to claim 1, characterized in that: The information integration and structuring module includes data association, data verification and result output.

7. The multi-agent architecture insurance history key information structured extraction system according to claim 1, characterized in that: The feedback and optimization module includes user feedback collection, performance evaluation and model optimization.

Images