Claim settlement processing method and device based on artificial intelligence, computer equipment and medium

Abstract

The invention belongs to the technical field of artificial intelligence, and relates to an artificial intelligence-based claim settlement processing method and device, computer equipment and a storage medium, and the method comprises the steps: receiving a medical document image file and insurance policy information inputted by a user; performing character recognition and extraction processing on the document image file to obtain a character information list; performing entity recognition on the text information list based on an entity recognition model to obtain document information; performing time sequence segmentation processing on the document information based on the segmentation model to obtain a document information sequence; processing the document information sequence based on an archive construction model to obtain medical archive data; based on the first prompt text, performing statistics on the medical archive data and the insurance policy information by using a statistical model to obtain medical claim amount data; and performing output processing on the medical claim amount data. The method can be applied to claim settlement processing scenes in the fields of financial science and technology and medical health insurance, and improves the efficiency and accuracy of claim settlement processing by combining the use of a plurality of models.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology and can be applied to fields such as fintech and healthcare insurance, particularly to artificial intelligence-based claims processing methods, devices, computer equipment, and storage media. Background Technology

[0002] In the field of insurance claims, especially in cases involving medical expenses, claims personnel must handle a large volume of complex medical documentation. These documents are diverse, including but not limited to doctor's slips, prescription slips, and hospitalization slips, with varying formats and the potential for duplicate expense records. This necessitates a significant investment of time and effort in meticulous organization and summarization to ensure the accuracy and completeness of the information. Furthermore, claims personnel must manually verify that the medical expenses occurred within the validity period stipulated in the insurance contract. This process involves checking and sorting the timestamps of each document, which is time-consuming and error-prone. Moreover, claims personnel must accurately calculate the amounts recorded on the medical documents, including complex calculations such as addition, deduction of deductibles, and calculation of reimbursement ratios, significantly increasing workload and the risk of errors.

[0003] In summary, the existing claims processing procedures face multiple challenges when handling medical document information, including information compilation, time verification, and amount calculation. This results in low efficiency and a high risk of errors, highlighting the shortcomings of the current claims processing procedures in terms of efficiency and accuracy. Therefore, there is an urgent need to develop an efficient and accurate claims processing method to overcome the deficiencies of existing technologies and improve the overall effectiveness of the claims processing process. Summary of the Invention

[0004] The purpose of this application is to provide an artificial intelligence-based claims processing method, apparatus, computer equipment, and storage medium to solve the technical problems of low processing efficiency and accuracy in existing claims processing procedures.

[0005] Firstly, an artificial intelligence-based claims processing method is provided, including:

[0006] Receives medical document image files and insurance policy information input by the user;

[0007] The document image file is subjected to text recognition and extraction processing to obtain a corresponding list of text information;

[0008] The text information list is subjected to entity recognition based on a preset entity recognition model to obtain the corresponding document information;

[0009] The document information is segmented into a time series based on a preset segmentation model to obtain the corresponding document information sequence.

[0010] Based on a preset archive construction model, the document information sequence is processed to obtain the corresponding medical record data;

[0011] Based on the preset first prompt text, a preset statistical model is used to perform statistical processing on the medical record data and the insurance policy information to obtain the corresponding medical claim amount data.

[0012] The medical claim amount data is then output and processed.

[0013] Secondly, an artificial intelligence-based claims processing device is provided, comprising:

[0014] The first receiving module is used to receive medical document image files and insurance policy information input by the user;

[0015] The first processing module is used to perform text recognition and extraction processing on the document image file to obtain a corresponding list of text information.

[0016] The recognition module is used to perform entity recognition on the text information list based on a preset entity recognition model to obtain the corresponding document information;

[0017] The segmentation module is used to perform time-series segmentation processing on the document information based on a preset segmentation model to obtain the corresponding document information sequence.

[0018] The second processing module is used to process the document information sequence based on a preset archive construction model to obtain the corresponding medical record data.

[0019] The statistics module is used to perform statistical processing on the medical record data and the insurance policy information based on the preset first prompt text and using a preset statistical model to obtain the corresponding medical claim amount data.

[0020] The output module is used to process and output the medical claim amount data.

[0021] Thirdly, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the aforementioned artificial intelligence-based claims processing method.

[0022] Fourthly, a computer-readable storage medium is provided, which stores a computer program that, when executed by a processor, implements the steps of the aforementioned claims processing method based on artificial intelligence.

[0023] In the aforementioned AI-based claims processing method, apparatus, computer equipment, and storage medium, the system first receives medical document image files and insurance policy information input by the user; then, it performs text recognition and extraction processing on the document image files to obtain a corresponding list of text information; next, it performs entity recognition on the list of text information based on a preset entity recognition model to obtain the corresponding document information; then, it performs time-series segmentation processing on the document information based on a preset segmentation model to obtain the corresponding document information sequence; and finally, it performs data processing on the document information sequence based on a preset archive construction model to obtain the corresponding medical archive data; subsequently, based on a preset first prompt text, it uses a preset statistical model to perform statistical processing on the medical archive data and the insurance policy information to obtain the corresponding medical claim amount data; and finally, it outputs the medical claim amount data. Based on the above claims processing process, this application combines entity recognition models, segmentation models, file construction models, and statistical models to process the input medical document image files and insurance policy information. This enables automated organization and summarization of medical document information, and efficient execution of time verification and claims amount calculation. It effectively optimizes human resource allocation, shortens claims processing time, improves the efficiency and accuracy of claims processing, and ensures the accuracy of the generated medical claims amount data. Attached Figure Description

[0024] To more clearly illustrate the solutions in this application, the accompanying drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is an exemplary system architecture diagram to which this application can be applied;

[0026] Figure 2 This is a flowchart of one embodiment of the AI-based claims processing method according to this application;

[0027] Figure 3 This is a schematic diagram of the structure of one embodiment of the AI-based claims processing device according to this application;

[0028] Figure 4 This is a schematic diagram of the structure of one embodiment of the computer device according to this application. Detailed Implementation

[0029] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein in the specification of the application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application; the terms "comprising" and "having," and any variations thereof, in the specification, claims, and foregoing drawings of this application, are intended to cover non-exclusive inclusion. The terms "first," "second," etc., in the specification, claims, or foregoing drawings of this application are used to distinguish different objects, not to describe a particular order.

[0030] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0031] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0032] like Figure 1 As shown, system architecture 100 may include terminal device 101, network 102, and server 103. Terminal device 101 may be a laptop 1011, tablet 1012, or mobile phone 1013. Network 102 is used as a medium to provide a communication link between terminal device 101 and server 103. Network 102 may include various connection types, such as wired, wireless communication links, or fiber optic cables, etc.

[0033] Users can use terminal device 101 to interact with server 103 via network 102 to receive or send messages, etc. Various communication client applications can be installed on terminal device 101, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social media platform software, etc.

[0034] Terminal device 101 can be various electronic devices with a display screen and support web browsing. In addition to laptops 1011, tablets 1012, or mobile phones 1013, terminal device 101 can also be an e-book reader, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 player (Moving Picture Experts Group Audio Layer IV), a laptop computer, and a desktop computer, etc.

[0035] Server 103 can be a server that provides various services, such as a backend server that provides support for the pages displayed on terminal device 101.

[0036] It should be noted that the AI-based claims processing method provided in this application is generally executed by a server / terminal device, and correspondingly, the AI-based claims processing device is generally installed in the server / terminal device.

[0037] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0038] Continue to refer to Figure 2 This document illustrates a flowchart of an embodiment of the AI-based claims processing method according to this application. The order of steps in the flowchart can be changed, and some steps can be omitted, depending on different requirements. The AI-based claims processing method provided in this application can be applied to any scenario requiring claims processing, and thus can be applied to products in these scenarios, such as claims processing scenarios in the fintech and healthcare insurance fields. The AI-based claims processing method includes the following steps:

[0039] Step S201: Receive the medical document image file and insurance policy information input by the user.

[0040] In this embodiment, the AI-based claims processing method operates on electronic devices (e.g., Figure 1The server / terminal device shown can acquire user-input medical document image files and insurance policy information via wired or wireless connections. It should be noted that the aforementioned wireless connection methods may include, but are not limited to, 3G / 4G / 5G connections, WiFi connections, Bluetooth connections, WiMAX connections, Zigbee connections, UWB (ultra-wideband) connections, and other currently known or future-developed wireless connection methods. The implementing entity of this application is specifically a claims processing system, or a medical claims processing system, which can be simply referred to as the system. The aforementioned medical document image files serve as original evidence in the user's claims application, recording the patient's medical treatment (such as diagnoses, examination reports, expense lists, etc.). The aforementioned insurance policy information defines the terms and scope of coverage of the insurance contract, clarifying which medical treatments are reimbursable and how the amount is calculated.

[0041] Step S202: Perform text recognition and extraction processing on the document image file to obtain the corresponding text information list.

[0042] In this embodiment, the text recognition and extraction processing of the document image file described above yields a corresponding list of text information. The specific implementation process will be further described in detail in subsequent embodiments of this application, and will not be elaborated upon here.

[0043] Step S203: Perform entity recognition on the text information list based on the preset entity recognition model to obtain the corresponding document information.

[0044] In this embodiment, the aforementioned list of text information can be input into the entity recognition model, which then performs entity recognition on the list and outputs the corresponding document information. The specific construction process of the entity recognition model will be described in more detail in subsequent embodiments and will not be elaborated upon here.

[0045] Step S204: Perform time series segmentation processing on the document information based on a preset segmentation model to obtain the corresponding document information sequence.

[0046] In this embodiment, the specific implementation process of performing time-series segmentation processing on the document information based on the preset segmentation model to obtain the corresponding document information sequence will be further described in detail in subsequent specific embodiments of this application, and will not be elaborated on here.

[0047] Step S205: Based on the preset archive construction model, the document information sequence is processed to obtain the corresponding medical record data.

[0048] In this embodiment, the above-mentioned document information sequence can be input into the above-mentioned file construction model, and the file construction model can be used to construct medical records from the document information sequence and output the corresponding medical record data.

[0049] The construction process of the above-mentioned medical record construction model includes medical record template design, data annotation, model training, model optimization and iteration. Specifically, (1) Medical record template design includes: Template goal: Design a standardized medical record template for uniformly recording and storing patient medical information. Template content: Define the key information fields that need to be included in the medical record, such as patient basic information, medical time, department, diagnosis description, cost details, etc. Determine the data type and format requirements of each field to ensure data consistency and comparability. (2) Data annotation includes: Data collection: Collect a certain number (e.g., 5000) of user medical data. These data should contain multiple document information, covering different medical scenarios and situations. Annotation process: Professional annotators annotate the collected medical data according to the designed medical record template. Annotation content includes filling the document information into the corresponding fields of the template to generate complete medical record labels. (3) Model training includes: Model selection: Select a model with a small number of parameters (e.g., qwen7b) for fine-tuning to improve training efficiency and inference speed. Fine-tuning strategy: The LoRA algorithm is used to fine-tune the model and optimize it for the medical record generation task. A certain amount of labeled data (e.g., 4000 records) is used for training, and the remaining data (e.g., 1000 records) is used as a test set to evaluate the model performance. (4) Model optimization and iteration include: evaluating the model performance based on the performance of the test set and identifying existing problems and deficiencies. Iterative optimization: the model is further optimized based on the evaluation results, such as adjusting fine-tuning parameters and increasing training data, to improve the accuracy and efficiency of medical record generation and obtain the final record construction model. In addition, the reasoning process of the record construction model includes: inputting the time series version of the document information into the fine-tuned model. The model integrates the document information into a complete medical record based on the medical record template and training data. During the reasoning process, the model will ensure the integrity and accuracy of the record information and handle any missing or contradictory information.

[0050] Step S206: Based on the preset first prompt text, use a preset statistical model to perform statistical processing on the medical record data and the insurance policy information to obtain the corresponding medical claim amount data.

[0051] In this embodiment, the specific implementation process of using a preset statistical model to statistically process the medical record data and the insurance policy information based on the preset first prompt text to obtain the corresponding medical claim amount data will be further described in detail in subsequent specific embodiments of this application, and will not be elaborated on here.

[0052] Step S207: Output the medical claim amount data.

[0053] In this embodiment, the medical claim amount data can be sent to the aforementioned user to complete the output processing of the medical claim amount data. The method of sending the medical claim amount data is not specifically limited; for example, it can be sent via email, SMS, or displayed on a user interface.

[0054] This application first receives medical document image files and insurance policy information input by the user; then, it performs text recognition and extraction processing on the document image files to obtain a corresponding list of text information; next, it performs entity recognition on the list of text information based on a preset entity recognition model to obtain the corresponding document information; then, it performs time series segmentation processing on the document information based on a preset segmentation model to obtain the corresponding document information sequence; and then, it performs data processing on the document information sequence based on a preset archive construction model to obtain the corresponding medical archive data; subsequently, based on a preset first prompt text, it uses a preset statistical model to perform statistical processing on the medical archive data and the insurance policy information to obtain the corresponding medical claim amount data; finally, it outputs the medical claim amount data. Based on the above claims processing process, this application combines entity recognition models, segmentation models, file construction models, and statistical models to process the input medical document image files and insurance policy information. This enables automated organization and summarization of medical document information, and efficient execution of time verification and claims amount calculation. It effectively optimizes human resource allocation, shortens claims processing time, improves the efficiency and accuracy of claims processing, and ensures the accuracy of the generated medical claims amount data.

[0055] In some optional implementations of this embodiment, before step S203, the electronic device may further perform the following steps:

[0056] Obtain pre-labeled medical document sample data and divide the medical document sample data into training dataset and test dataset.

[0057] In this embodiment, labeled medical document sample data is prepared according to actual needs. The data includes key entities that need to be identified (including but not limited to document type, consultation time, patient name, department, diagnosis description, medical institution, admission time, discharge time, examination items, fee name and fee amount, etc.) and their corresponding labeling information.

[0058] Call the preset entity link model.

[0059] In this embodiment, an LLM model with a small number of parameters (such as qwen2-7b) is selected as the base model for the entity recognition model to ensure the model's generalization ability and computational efficiency.

[0060] Based on a preset low-rank adaptation algorithm, the entity link model is fine-tuned using the training dataset to obtain the corresponding first processing model.

[0061] In this embodiment, the fine-tuning training process includes: using the LoRA (Low-Rank Adaptation) algorithm to fine-tune the entity linking model to improve the recognition accuracy of specific entities. The LoRA algorithm adapts to downstream tasks by adding low-rank matrices; during training, only the parameters of these low-rank matrices are updated, while the pre-trained model parameters remain unchanged. Furthermore, by training the LoRA adapter on the training dataset, the model parameters of the entity linking model are adjusted to optimize the recognition ability of specific entities, resulting in a well-trained first processing model.

[0062] The first processing model is evaluated based on the test dataset to obtain the corresponding model evaluation results.

[0063] In this embodiment, the first processing model is evaluated by using a test dataset. This involves verifying the structured output of the first processing model, classifying and analyzing errors found during the verification process, identifying which entity types or scenarios the first processing model performs poorly, and calculating metrics such as accuracy and recall of the first processing model. The overall performance of the model in the current task is evaluated, and the generated verification results are used as the evaluation results of the model.

[0064] The first processing model is optimized based on a model fine-tuning strategy that matches the model evaluation results to obtain the corresponding second processing model.

[0065] In this embodiment, based on the obtained verification results (model evaluation results), the specific aspects that the first processing model needs to be optimized can be identified, such as low recognition accuracy of certain entities, performance degradation in specific scenarios, etc., and targeted optimization schemes (i.e. model fine-tuning strategies) can be formulated, such as adjusting LoRA parameters, increasing training data, improving data annotation quality, etc., and the optimization scheme can be applied to optimize the first processing model to obtain the corresponding second processing model.

[0066] The process includes: 1) LoRA parameter tuning: Parameter analysis: Analyzing the current LoRA adapter parameter settings to identify parameters that may affect model performance. Parameter tuning: Adjusting parameters such as the LoRA rank or learning rate based on the problem location to optimize the model's performance for specific entities or scenarios. Experimental validation: After parameter tuning, conducting experiments using a validation dataset to evaluate the effectiveness of the parameter tuning. 2) Training data augmentation: Data supplementation: Collecting or generating more training data for entities or scenarios where the model performs poorly. Data annotation improvement: Providing high-quality annotations to the new data to ensure consistency and accuracy. Data balancing: Checking the distribution of training data to ensure a balanced amount of data for various entities and scenarios, avoiding model bias towards certain types. 3) Model retraining: Incremental training: Performing incremental training on the model using the adjusted LoRA parameters and augmented training data. Training monitoring: Monitoring changes in the model's loss function and accuracy during training to ensure model convergence and achieves expected performance.

[0067] In addition, the optimized model can be re-validated using a test dataset to evaluate the optimization effect. The performance of the model before and after optimization can be compared to confirm the effectiveness of the optimization measures. Furthermore, after model deployment, its performance in real-world applications should be continuously monitored, and user feedback and new validation data should be collected. Based on the monitoring results and user feedback, the validation and optimization process should be repeated cyclically to continuously improve the model's entity recognition accuracy and completeness.

[0068] The entity recognition model is generated based on the second processing model.

[0069] In this embodiment, the specific implementation process of generating the entity recognition model based on the second processing model will be further described in detail in subsequent specific embodiments of this application, and will not be elaborated on here.

[0070] This application acquires pre-labeled medical document sample data, divides it into training and testing datasets, then calls a preset entity linking model, and fine-tunes the model using the training dataset based on a preset low-rank adaptation algorithm to obtain a first processing model. The first processing model is then evaluated using the testing dataset to obtain a model evaluation result. Subsequently, the first processing model is optimized based on a model fine-tuning strategy matching the evaluation result to obtain a second processing model. Finally, the entity recognition model is generated based on the second processing model. Through this model construction process, this application can efficiently and accurately construct the required entity recognition model and effectively ensure the accuracy and completeness of entity recognition.

[0071] In some optional implementations, generating the entity recognition model based on the second processing model includes the following steps:

[0072] Obtain the preset model quantization strategy.

[0073] In this embodiment, the selection of the above-mentioned model quantization strategy is not specifically limited, and any one of the strategies such as dynamic quantization, static quantization, and mixed-precision quantization can be used. Dynamic quantization strategies include quantizing model weights during inference, suitable for scenarios where high accuracy is not required. Static quantization strategies include quantizing the model after training, which typically provides better performance improvement. Mixed-precision quantization strategies include combining FP16 and FP32 precision to balance speed and accuracy.

[0074] Based on the model quantization strategy, a preset quantization tool is invoked to quantize the second processing model, resulting in a quantized third processing model.

[0075] In this embodiment, the quantization tool described above can be a quantization tool provided by a deep learning framework. Based on this quantization tool, a selected model quantization strategy can be used to quantize the second processing model, specifically quantizing the model's weights and activation values, thereby reducing its storage and computational requirements and obtaining the quantized third processing model.

[0076] The effectiveness of the third processing model is verified based on the test dataset.

[0077] In this embodiment, after the quantization process of the third processing model is completed, the accuracy of the third processing model is evaluated using the aforementioned test dataset to ensure that the quantization process does not significantly reduce the model performance. If the accuracy of the third processing model is within a preset range, the third processing model is determined to have passed the model effect verification; otherwise, the third processing model is determined to have passed the model effect verification.

[0078] If the third processing model passes the model performance verification, then the third processing model will be used as the entity recognition model.

[0079] In this embodiment, if the accuracy of the third processing model decreases too much, further adjustments to the quantization parameters or attempts to try other quantization methods can be considered.

[0080] This application obtains a preset model quantization strategy, and then, based on the model quantization strategy, calls a preset quantization tool to quantize the second processing model to obtain a quantized third processing model. Subsequently, the third processing model is validated based on the test dataset. If the third processing model passes the validation, it is used as the entity recognition model. Based on the above processing flow, this application, by compressing the second processing model using a model quantization strategy, can significantly reduce the size and computational resource requirements of the generated entity recognition model, improve inference speed, and simultaneously maintain the accuracy of the entity recognition model as much as possible. This will help reduce the operating costs of the claims processing system, improve processing efficiency, and provide users with faster and more efficient services.

[0081] In some alternative implementations, step S204 includes the following steps:

[0082] Retrieve the second prompt text related to the document classification.

[0083] In this embodiment, based on the actual document information classification requirements, a clear and specific prompt (second prompt text) is pre-designed to guide the segmentation model in classifying and organizing documents according to the chronological order of medical visits. For example, the content of the second prompt text includes: explicitly requiring the segmentation model to divide documents into categories such as first medical visit, second medical visit, etc., and organize them according to the time interval of each visit; emphasizing that each visit may contain multiple documents (such as outpatient, inpatient, and expense details), and ensuring that these documents are categorized according to the corresponding time of visit; and providing examples or format instructions to help the segmentation model understand the output format and expected results.

[0084] The document information and the second prompt text are input into the segmentation model.

[0085] In this embodiment, the segmentation model described above can be a large parameter model (qwen2-72b), which has powerful reasoning and contextual understanding capabilities.

[0086] Based on the segmentation model, the document information is segmented into a time series according to the second prompt text to obtain the corresponding initial document information sequence.

[0087] In this embodiment, the aforementioned document information can be a structured information dictionary. The reasoning process based on the segmentation model includes: the segmentation model, guided by the prompt (second prompt text), analyzes the medical treatment time information in the document information, divides the documents into different medical treatment time periods according to their chronological order, and, during the reasoning process, considers the correlation between documents to ensure that documents within the same medical treatment time period are correctly classified, thereby generating a document information dictionary divided by medical treatment time – a time-series version, i.e., the aforementioned initial document information sequence. This initial document information sequence is a structured dictionary, where each key represents a medical treatment time period (e.g., "first medical treatment"), and the corresponding value is a list of all relevant document information within that time period.

[0088] The initial document information sequence is verified based on a preset verification strategy.

[0089] In this embodiment, the accuracy and completeness of time division can be detected by verifying the initial document information sequence output by the segmentation model. If the accuracy and completeness of time division both pass the detection, the initial document information sequence is determined to have passed the verification; otherwise, the initial document information sequence is determined to have failed the verification.

[0090] If the initial document information sequence passes verification, then the initial document information sequence is used as the document information sequence.

[0091] In this embodiment, the model parameters of the second prompt text or the segmentation model can be fine-tuned based on the verification results to improve the accuracy of time series segmentation.

[0092] This application obtains a second prompt text related to document classification, then inputs the document information and the second prompt text into a segmentation model. Based on the segmentation model, the document information is segmented into a time series according to the second prompt text to obtain a corresponding initial document information sequence. Subsequently, the initial document information sequence is verified based on a preset verification strategy. If the initial document information sequence passes verification, it is used as the final document information sequence. Based on the above processing flow, this application can efficiently and accurately complete the time series segmentation of document information by using a segmentation model, effectively improving the generation efficiency of document information sequences and ensuring the accuracy of the obtained document information sequences.

[0093] In some alternative implementations, step S206 includes the following steps:

[0094] Retrieve the first prompt text related to the calculation of medical claim amounts.

[0095] In this embodiment, based on the statistical needs of medical claim amounts, a clear and specific prompt (i.e., the first prompt text) is pre-set to guide the statistical model on how to calculate various claim amounts based on structured medical records and insurance policy information. Specifically, the content of the first prompt text includes: explicitly requiring the statistical model to calculate medical expenses, hospitalization allowances, lost wages, and disability benefits separately, and to calculate the total amount; providing detailed calculation rules and examples to ensure that the statistical model can accurately understand the calculation methods and basis for each expense; and specifying the output format, requiring the statistical model to return results according to the specified template, including details of each expense and the total amount.

[0096] The statistical model construction process includes data annotation and model training. Data annotation includes: Data preparation: Using a certain number (e.g., 5000) of already annotated user medical records and corresponding insurance policy information as annotation data, ensuring the data covers various medical scenarios and insurance types so the model can learn the cost calculation rules under different circumstances. Annotation process: Professional annotators annotate the medical records and insurance policy information according to the designed prompt and calculation rules. Annotation content includes: annotating the calculation results of various costs, including medical expenses, hospitalization allowances, lost wages, and disability benefits; annotating the total amount to ensure the sum of all costs matches the total amount; and recording any special situations and handling methods during the annotation process for subsequent model training and optimization.

[0097] Model training includes: Model selection: Selecting a model with fewer parameters (such as qwen7b) for fine-tuning to improve training efficiency and inference speed while maintaining sufficient generalization ability. Fine-tuning strategies include: LoRA algorithm application: Using the LoRA algorithm to fine-tune the model, optimizing it for the medical claims calculation task. Training data allocation: Using a first specified amount (e.g., 4000 cases) of labeled data for training, and the remaining data (e.g., 1000 cases) as a test set to evaluate model performance. Training process includes: Training objective: Training the model to accurately calculate various claims amounts and the total amount based on structured medical records and insurance policy information. Training monitoring: Monitoring the loss function and accuracy during training to ensure model convergence and achieves the expected performance.

[0098] Obtain the preset calculation rules.

[0099] In this embodiment, the above calculation rules are statistical rules constructed based on the actual medical claim amount calculation needs, including the processing logic for calculating various claim amounts.

[0100] Based on the statistical model, the medical record data and the insurance policy information are statistically processed according to the calculation rules and the first prompt text to obtain the corresponding claim amounts.

[0101] In this embodiment, the statistical processing based on the statistical model includes: inputting structured medical record data and insurance policy information into the statistical model; the statistical model parses the input information and calculates various claim amounts based on the prompt (first prompt text) and training data; during the inference process, the statistical model strictly follows the calculation rules and insurance terms to calculate costs, ensuring the accuracy and compliance of the output claim amount statistics.

[0102] Specifically, the calculation logic for the claim amount includes: Medical expenses: calculated based on the expense details in the medical documents, combined with the reimbursement ratio and deductible in the insurance policy. Hospitalization allowance: determined based on the daily allowance amount stipulated in the insurance policy and the number of days of hospitalization in the medical documents. Lost wages: calculated based on the policy terms (such as based on a percentage of wages or a fixed amount) and the number of days of lost work in the medical documents. Disability compensation: determined based on the disability rating table in the insurance policy and the disability assessment results in the medical documents.

[0103] Sum all the claimed amounts to obtain the total claim amount.

[0104] In this embodiment, the aforementioned compensation amounts include medical expenses, hospitalization allowances, lost wages, and disability compensation. The total compensation amount is obtained by summing up the various compensation amounts.

[0105] The medical claim amount data is obtained by integrating all the claim amounts and the total claim amount.

[0106] In this embodiment, the aforementioned medical claim amount data includes medical expenses, hospitalization allowance, lost wages, disability compensation, and the total claim amount.

[0107] This application obtains a first prompt text related to the statistical calculation of medical claim amounts; and a preset calculation rule; then, based on the statistical model, it performs statistical processing on the medical record data and the insurance policy information according to the calculation rule and the first prompt text to obtain the corresponding claim amounts; subsequently, it sums all the claim amounts to obtain the corresponding total claim amount; and finally, it integrates all the claim amounts and the total claim amount to obtain the medical claim amount data. Based on the above processing flow, this application, by combining the first prompt text and the statistical model to perform statistical processing on medical record data and insurance policy information, can automate and accurately complete the calculation of claim amounts, improve the efficiency of claim amount calculation, reduce human error, and ensure the accuracy of the generated medical claim amount data.

[0108] In some optional implementations of this embodiment, step S202 includes the following steps:

[0109] The medical document image file is preprocessed to obtain the corresponding target image file.

[0110] In this embodiment, the aforementioned medical document image file is an image file uploaded by the user. This medical document image file can be various types of medical document images, such as consultation slips, prescription slips, and hospitalization slips, and there must be at least one such file. The aforementioned preprocessing may include format conversion (ensuring the image is in a format supported by the OCR model) and image enhancement (improving image clarity for easier subsequent recognition).

[0111] The target image file is processed using a preset text recognition tool to obtain the corresponding text recognition result.

[0112] In this embodiment, the PaddleOCR model can be selected as the text recognition tool. This model supports text recognition in multiple languages ​​and scenarios, and has high accuracy and robustness. Furthermore, by batch inputting preprocessed target image files into the PaddleOCR model, automated text recognition and data extraction are performed to obtain the corresponding text recognition results. Specifically, the recognition process includes: detecting text regions in each image and locating the text regions within the image; performing character recognition on the detected text regions to convert the text in the image into editable text information; and extracting the text information, including the text content and its position information within the image.

[0113] The text recognition results are integrated to obtain a corresponding result list.

[0114] In this embodiment, the text recognition results of each target image file can be integrated into a list, where each element in the list corresponds to the recognition result of one image, and the results are output in dictionary format to obtain the corresponding result list. Specifically, the dictionary format output includes: the text recognition result of each image is stored in dictionary format, containing the following information: image_id: a unique identifier used to distinguish different image files; text_info: a list of extracted text information, which may contain multiple text segments (such as recognition results of different text regions); position_info: the position information of the text in the image, used for subsequent layout analysis or verification.

[0115] The result list is validated based on a preset text validation strategy.

[0116] In this embodiment, the result list can be verified based on the above text verification strategy to check whether there are obvious recognition errors (such as garbled characters, typos, etc.). If there are, the result list is determined to pass the verification; otherwise, the result list is determined to fail the verification.

[0117] If the result list passes the verification, then the result list will be used as the text information list.

[0118] In this embodiment, when the result list fails verification, the user can also modify or supplement the result list, especially in cases of inaccurate or missing identification, the user can manually enter or correct relevant information.

[0119] This application preprocesses the medical document image file to obtain a corresponding target image file. Then, it uses a preset text recognition tool to recognize and process the target image file, obtaining corresponding text recognition results. These results are then integrated to obtain a result list. Subsequently, the result list is verified using a preset text verification strategy. If the result list passes verification, it is used as the text information list. Through this process, after preprocessing the medical document image file to obtain the target image file, this application, by using a text recognition tool and text verification strategy, can efficiently and accurately complete the text recognition and extraction processing of the document image file, ensuring the accuracy of the generated text information list. This improves the accuracy of subsequent claims processing based on the text information list.

[0120] In some optional implementations of this embodiment, before step S206, the electronic device may further perform the following steps:

[0121] The system uses pre-set data collection tools to receive feedback data related to claims processing submitted by target users through feedback channels.

[0122] In this embodiment, the feedback channels may include online forms, emails, customer service hotlines, and social media. A standardized feedback template is pre-set, which includes: Basic information: user name, contact information, feedback time, etc. Problem description: A detailed description of the problem encountered, including the error type and the scenario in which it occurred. Suggestions: The user's suggestions or expectations for improving the claims processing. Attachment support: Allows users to upload screenshots, documents, and other attachments to more intuitively illustrate the problem.

[0123] Specifically, feedback data from various channels can be collected using automated data acquisition tools and stored in a secure database to ensure data integrity and traceability. Furthermore, the collected feedback data can be categorized and tagged for easier subsequent analysis and processing. Regular backups of the feedback data are also essential to prevent data loss.

[0124] The feedback data is filtered to obtain the corresponding target feedback data.

[0125] In this embodiment, the collected feedback data can be preliminarily screened to exclude irrelevant or duplicate data, thereby obtaining the corresponding target feedback data.

[0126] The target feedback data is analyzed using a preset data analysis tool to obtain the corresponding feedback analysis results.

[0127] In this embodiment, data analysis tools can be used to conduct in-depth analysis of the target feedback data, identify common problems and user needs, and obtain corresponding feedback analysis results. Furthermore, the feedback data can be prioritized based on its severity, scope of impact, and user needs.

[0128] Based on the feedback analysis results, the statistical model is optimized accordingly.

[0129] In this embodiment, the statistical model can be optimized based on the feedback analysis results, such as adjusting fine-tuning parameters and increasing training data. After optimization, the statistical model is retested and validated to ensure the problem is resolved. The effectiveness of the feedback mechanism can be evaluated periodically, including the number of feedback responses, processing efficiency, and user satisfaction. Based on the evaluation results, the feedback mechanism can be continuously improved, such as optimizing the feedback process and adding feedback channels. Furthermore, regular user satisfaction surveys are conducted to understand users' opinions and suggestions on the feedback mechanism. Based on the survey results, the design and implementation strategies of the feedback mechanism are adjusted. Therefore, the continuous iteration and improvement of the feedback mechanism will help build a more robust and user-friendly claims processing system.

[0130] This application receives feedback data related to claims processing submitted by target users through feedback channels using a pre-set data collection tool. The feedback data is then filtered to obtain corresponding target feedback data. This target feedback data is then analyzed using a pre-set data analysis tool to obtain corresponding feedback analysis results. Subsequently, the statistical model is optimized based on these feedback analysis results. Through this process, this application can promptly understand user needs and problems, quickly and intelligently analyze user feedback data using data analysis tools to obtain feedback analysis results, and optimize the statistical model accordingly, thereby improving user satisfaction and the overall performance of the statistical model.

[0131] In some alternative implementations, the user information obtained is subject to user consent and complies with relevant laws and policies.

[0132] Furthermore, any software tools or components not belonging to our company that appear in the embodiments of this application are merely illustrative examples and do not represent actual use.

[0133] Furthermore, this application proposes a medical claims intelligent processing system based on a large model, which relies on large language model (LLM) technology and aims to automate key tasks in the medical claims process:

[0134] 1. For the first time, the system has achieved automated organization and summarization of medical document information, and can efficiently perform time verification and amount calculation. This process not only optimizes human resource allocation but also improves the accuracy and completeness of information summarization. Furthermore, the system significantly shortens processing time and provides real-time results for medical record organization and claim amount calculation, thereby significantly enhancing the user experience.

[0135] 2. To address the specific needs of medical claims, this system utilizes professionally labeled data, fine-tunes three LLMs using the LoRa algorithm, and designs two specific prompts to fully leverage the inference capabilities of the LLMs. These measures aim to improve the accuracy of document information mapping, the completeness and correctness of medical record generation, and the precision of claim amount identification and calculation. Through these innovations, the system can provide more efficient and accurate claims processing services.

[0136] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0137] It should be emphasized that, in order to further ensure the privacy and security of the aforementioned medical claim amount data, the aforementioned medical claim amount data can also be stored in a blockchain node.

[0138] The blockchain referred to in this application is a novel application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms. Essentially, a blockchain is a decentralized database, a chain of data blocks linked together using cryptographic methods. Each data block contains information about a batch of network transactions, used to verify the validity of the information (anti-counterfeiting) and generate the next block. A blockchain can include an underlying blockchain platform, a platform product service layer, and an application service layer.

[0139] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0140] Foundational artificial intelligence technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0141] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by instructing related hardware through computer-readable instructions. These computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the methods described above. The aforementioned storage medium can be a non-volatile storage medium such as a magnetic disk, optical disk, or read-only memory (ROM), or random access memory (RAM).

[0142] It should be understood that although the steps in the flowcharts of the accompanying figures are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the accompanying figures may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0143] Further reference Figure 3 As a response to the above Figure 2 The implementation of the method shown in this application provides an embodiment of an artificial intelligence-based claims processing device, which is similar to... Figure 2 Corresponding to the method embodiments shown, this device can be specifically applied to various electronic devices.

[0144] like Figure 3 As shown, the AI-based claims processing device 300 described in this embodiment includes: a first receiving module 301, a first processing module 302, an identification module 303, a segmentation module 304, a second processing module 305, a statistics module 306, and an output module 307. Wherein:

[0145] The first receiving module 301 is used to receive medical document image files and insurance policy information input by the user;

[0146] The first processing module 302 is used to perform text recognition and extraction processing on the document image file to obtain a corresponding list of text information.

[0147] The recognition module 303 is used to perform entity recognition on the text information list based on a preset entity recognition model to obtain the corresponding document information;

[0148] The segmentation module 304 is used to perform time series segmentation processing on the document information based on a preset segmentation model to obtain the corresponding document information sequence.

[0149] The second processing module 305 is used to process the document information sequence based on a preset archive construction model to obtain the corresponding medical record data.

[0150] The statistics module 306 is used to perform statistical processing on the medical record data and the insurance policy information based on the preset first prompt text and using a preset statistical model to obtain the corresponding medical claim amount data.

[0151] The output module 307 is used to output and process the medical claim amount data.

[0152] In this embodiment, the operations performed by the above modules or units correspond one-to-one with the steps of the AI-based claims processing method in the aforementioned implementation method, and will not be repeated here.

[0153] In some optional implementations of this embodiment, the AI-based claims processing device further includes:

[0154] The third processing module is used to acquire pre-labeled medical document sample data and divide the medical document sample data into a training dataset and a test dataset.

[0155] The calling module is used to invoke the preset entity link model;

[0156] The training module is used to fine-tune the entity link model using the training dataset based on a preset low-rank adaptation algorithm to obtain the corresponding first processing model.

[0157] The evaluation module is used to evaluate the first processing model based on the test dataset and obtain the corresponding model evaluation result.

[0158] The first optimization module is used to optimize the first processing model based on a model fine-tuning strategy that matches the model evaluation result, so as to obtain the corresponding second processing model.

[0159] The generation module is used to generate the entity recognition model based on the second processing model.

[0160] In some optional implementations of this embodiment, the generation module includes:

[0161] The first acquisition submodule is used to acquire the preset model quantization strategy;

[0162] The quantization submodule is used to call a preset quantization tool to quantize the second processing model based on the model quantization strategy, so as to obtain the quantized third processing model.

[0163] The first verification submodule is used to verify the model performance of the third processing model based on the test dataset.

[0164] The first determining submodule is used to use the third processing model as the entity recognition model if the third processing model passes the model effect verification.

[0165] In some optional implementations of this embodiment, the segmentation module 304 includes:

[0166] The second acquisition submodule is used to acquire the second prompt text related to the document classification;

[0167] The input submodule is used to input the document information and the second prompt text into the segmentation model;

[0168] The segmentation submodule is used to perform time-series segmentation processing on the document information based on the segmentation model and the second prompt text to obtain the corresponding initial document information sequence.

[0169] The second verification submodule is used to verify the initial document information sequence based on a preset verification strategy.

[0170] The second determining submodule is used to use the initial document information sequence as the document information sequence if the initial document information sequence passes verification.

[0171] In some optional implementations of this embodiment, the statistics module 306 includes:

[0172] The third acquisition submodule is used to acquire the first prompt text related to the statistical medical claim amount;

[0173] The fourth submodule is used to obtain preset calculation rules;

[0174] The statistics submodule is used to perform statistical processing on the medical record data and the insurance policy information based on the statistical model, according to the calculation rules and the first prompt text, to obtain the corresponding claim amounts.

[0175] The summation submodule is used to sum all the claimed amounts to obtain the corresponding total claim amount;

[0176] The first integration submodule is used to integrate all the claimed amounts and the total claimed amount to obtain the medical claim amount data.

[0177] In some optional implementations of this embodiment, the first processing module 302 includes:

[0178] The preprocessing submodule is used to preprocess the medical document image file to obtain the corresponding target image file;

[0179] The recognition submodule is used to perform recognition processing on the target image file based on a preset text recognition tool to obtain the corresponding text recognition result;

[0180] The second integration submodule is used to integrate the text recognition results to obtain a corresponding result list.

[0181] The third verification submodule is used to verify the result list based on a preset text verification strategy;

[0182] The third determining submodule is used to use the result list as the text information list if the result list passes the verification.

[0183] In some optional implementations of this embodiment, the AI-based claims processing device further includes:

[0184] The second receiving module is used to receive feedback data related to claims processing submitted by target users through feedback channels based on preset collection tools;

[0185] The filtering module is used to filter the feedback data to obtain the corresponding target feedback data;

[0186] The analysis module is used to analyze the target feedback data based on preset data analysis tools to obtain corresponding feedback analysis results;

[0187] The second optimization module is used to perform corresponding model optimization processing on the statistical model based on the feedback analysis results.

[0188] To address the aforementioned technical problems, embodiments of this application also provide a computer device. Please refer to [link / reference needed]. Figure 4 , Figure 4 This is a basic structural block diagram of the computer device in this embodiment.

[0189] The computer device 4 includes a memory 41, a processor 42, and a network interface 43 that are interconnected via a system bus. It should be noted that only the computer device 4 with components 41-43 is shown in the figure; however, it should be understood that it is not required to implement all the shown components, and more or fewer components can be implemented alternatively. Those skilled in the art will understand that the computer device described here is a device capable of automatically performing numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes, but is not limited to, microprocessors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0190] The computer device can be a desktop computer, laptop, handheld computer, or cloud server, etc. The computer device can interact with the user via a keyboard, mouse, remote control, touchpad, or voice control.

[0191] The memory 41 includes at least one type of readable storage medium, including flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 41 may be an internal storage unit of the computer device 4, such as the hard disk or memory of the computer device 4. In other embodiments, the memory 41 may also be an external storage device of the computer device 4, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the computer device 4. Of course, the memory 41 may also include both the internal storage unit and its external storage device of the computer device 4. In this embodiment, the memory 41 is typically used to store the operating system and various application software installed on the computer device 4, such as computer-readable instructions for claims processing methods based on artificial intelligence. In addition, the memory 41 can also be used to temporarily store various types of data that have been output or will be output.

[0192] In some embodiments, the processor 42 may be a central processing unit (CPU), controller, microcontroller, microprocessor, or other data processing chip. The processor 42 is typically used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is used to execute computer-readable instructions stored in the memory 41 or to process data, for example, to execute computer-readable instructions for the AI-based claims processing method.

[0193] The network interface 43 may include a wireless network interface or a wired network interface, which is typically used to establish communication connections between the computer device 4 and other electronic devices.

[0194] This application also provides another embodiment, namely, providing a computer-readable storage medium storing computer-readable instructions that can be executed by at least one processor to cause the at least one processor to perform the steps of the artificial intelligence-based claims processing method described above.

[0195] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0196] Obviously, the embodiments described above are only some embodiments of this application, not all embodiments. The accompanying drawings show preferred embodiments of this application, but do not limit the patent scope of this application. This application can be implemented in many different forms; rather, the purpose of providing these embodiments is to provide a more thorough and comprehensive understanding of the disclosure of this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing specific embodiments, or make equivalent substitutions for some of the technical features. Any equivalent structures made using the content of this application's specification and drawings, directly or indirectly applied to other related technical fields, are similarly within the scope of patent protection of this application.

Claims

1. A claims processing method based on artificial intelligence, characterized in that, Includes the following steps: Receives medical document image files and insurance policy information input by the user; The document image file is subjected to text recognition and extraction processing to obtain a corresponding list of text information; The text information list is subjected to entity recognition based on a preset entity recognition model to obtain the corresponding document information; The document information is segmented into a time series based on a preset segmentation model to obtain the corresponding document information sequence. Based on a preset archive construction model, the document information sequence is processed to obtain the corresponding medical record data; Based on the preset first prompt text, a preset statistical model is used to perform statistical processing on the medical record data and the insurance policy information to obtain the corresponding medical claim amount data. The medical claim amount data is then output and processed.

2. The claims processing method based on artificial intelligence according to claim 1, characterized in that, Before the step of performing entity recognition on the text information list based on a preset entity recognition model to obtain the corresponding document information, the method further includes: Obtain pre-labeled medical document sample data, and divide the medical document sample data into training dataset and test dataset; Invoke the preset entity link model; Based on the preset low-rank adaptation algorithm, the entity link model is fine-tuned using the training dataset to obtain the corresponding first processing model. The first processing model is evaluated based on the test dataset to obtain the corresponding model evaluation result. The first processing model is optimized based on a model fine-tuning strategy that matches the model evaluation results to obtain the corresponding second processing model. The entity recognition model is generated based on the second processing model.

3. The claims processing method based on artificial intelligence according to claim 2, characterized in that, The step of generating the entity recognition model based on the second processing model specifically includes: Obtain the preset model quantization strategy; Based on the model quantization strategy, a preset quantization tool is invoked to quantize the second processing model, resulting in a quantized third processing model. The third processing model was validated based on the test dataset. If the third processing model passes the model performance verification, then the third processing model will be used as the entity recognition model.

4. The claims processing method based on artificial intelligence according to claim 1, characterized in that, The step of performing time-series segmentation processing on the document information based on a preset segmentation model to obtain the corresponding document information sequence specifically includes: Retrieve the second prompt text related to the document classification; Input the document information and the second prompt text into the segmentation model; Based on the segmentation model, the document information is segmented into a time series according to the second prompt text to obtain the corresponding initial document information sequence. The initial document information sequence is verified based on a preset verification strategy; If the initial document information sequence passes verification, then the initial document information sequence is used as the document information sequence.

5. The claims processing method based on artificial intelligence according to claim 1, characterized in that, The step of statistically processing the medical record data and the insurance policy information based on a preset first prompt text using a preset statistical model to obtain the corresponding medical claim amount data specifically includes: Retrieve the first notification text related to the calculation of medical claim amounts; Obtain the preset calculation rules; Based on the statistical model, the medical record data and the insurance policy information are statistically processed according to the calculation rules and the first prompt text to obtain the corresponding claim amounts; Sum all the claimed amounts to obtain the total claimed amount. The medical claim amount data is obtained by integrating all the claim amounts and the total claim amount.

6. The claims processing method based on artificial intelligence according to claim 1, characterized in that, The step of performing text recognition and extraction processing on the document image file to obtain the corresponding text information list specifically includes: The medical document image file is preprocessed to obtain the corresponding target image file; The target image file is processed using a preset text recognition tool to obtain the corresponding text recognition result; The text recognition results are integrated to obtain a corresponding result list; The result list is verified based on a preset text verification strategy; If the result list passes the verification, then the result list will be used as the text information list.

7. The claims processing method based on artificial intelligence according to claim 1, characterized in that, Before the step of statistically processing the medical record data and the insurance policy information using a preset statistical model based on a preset first prompt text to obtain the corresponding medical claim amount data, the method further includes: Based on the preset data collection tools, the system receives feedback data related to claims processing submitted by target users through feedback channels; The feedback data is filtered to obtain the corresponding target feedback data; The target feedback data is analyzed using a preset data analysis tool to obtain the corresponding feedback analysis results; Based on the feedback analysis results, the statistical model is optimized accordingly.

8. A claims processing device based on artificial intelligence, characterized in that, include: The first receiving module is used to receive medical document image files and insurance policy information input by the user; The first processing module is used to perform text recognition and extraction processing on the document image file to obtain a corresponding list of text information. The recognition module is used to perform entity recognition on the text information list based on a preset entity recognition model to obtain the corresponding document information; The segmentation module is used to perform time-series segmentation processing on the document information based on a preset segmentation model to obtain the corresponding document information sequence. The second processing module is used to process the document information sequence based on a preset archive construction model to obtain the corresponding medical record data. The statistics module is used to perform statistical processing on the medical record data and the insurance policy information based on the preset first prompt text and using a preset statistical model to obtain the corresponding medical claim amount data. The output module is used to process and output the medical claim amount data.

9. A computer device, characterized in that, The method includes a memory and a processor, wherein the memory stores computer-readable instructions, and the processor executes the computer-readable instructions to implement the steps of the claims processing method based on artificial intelligence as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the claims processing method based on artificial intelligence as described in any one of claims 1 to 7.

Images