Full-process intelligent claim settlement operation method and device

Abstract

The invention provides a full-process intelligent claim settlement operation method and device, and solves the problems of poor material quality, insufficient data standardization and low manual auditing efficiency in health insurance claim settlement in the prior art. The invention discloses a full-process intelligent claim settlement operation method. The method comprises the steps of obtaining perception information; the perception information is perception information of an image uploaded by a client; determining identification information according to the artificial intelligence identification perception information; the identification information comprises medical bill fields, diagnosis codes or medicine details; and determining a claim settlement type according to the combination of the identification information, auditing the perception information according to a preset rule corresponding to the claim settlement type, and outputting auditing information. According to the invention, through the innovative design of intelligent image quality detection, multi-source data fusion and five-layer decision-making architecture, full-process intelligent processing of health insurance claim settlement is realized.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a full-process intelligent claims settlement method and device. Background Art

[0002] As a crucial component of my country's multi-tiered medical security system, commercial health insurance faces numerous technical bottlenecks in its intelligent transformation of claims services. First, online claims processing suffers from image capture quality issues. Claim documents captured by customers themselves often lack clarity, significantly reducing OCR recognition accuracy and leading to significant issues with missed documents. Second, medical data standardization is insufficient, manifested in inconsistent bill formats, varying electronic invoice standards, and significant discrepancies in clinical diagnostic terminology. Furthermore, the intelligent transformation of the initial claims review process is difficult, requiring verification of multiple factors. Manual initial review is inefficient, and existing image classification technology needs improvement in accuracy, lacking cross-document logical verification capabilities. Current insurance claims systems suffer from three major technical gaps: a difficult balance between user experience and quality control, a significant conflict between standardization and flexibility, and insufficient coordination between rule engines and AI models. Existing claims systems still lack intelligent material quality testing, automatic adaptation of non-standard data, and full automation of the initial review process.

[0003] In this context, it is necessary to build a full-process intelligent claims processing method and device that integrates intelligent image acquisition, intelligent recognition and intelligent review. This method and device has become a key technological breakthrough to promote the high-quality development of the industry. Summary of the Invention

[0004] This application proposes a full-process intelligent claims processing method and device to solve the problems of poor material quality, insufficient data standardization and low manual review efficiency in health insurance claims in the existing technology.

[0005] In a first aspect, the present invention provides a full-process intelligent claims settlement method, comprising the following steps:

[0006] Acquiring perception information; the perception information is the perception information of the image uploaded by the customer;

[0007] Determining identification information based on artificial intelligence recognition perception information; the identification information includes medical bill fields, diagnosis codes, or drug details;

[0008] Determine the claim type based on the combination of identification information, review the perceived information based on the preset rules corresponding to the claim type, and output the review information.

[0009] Furthermore, before obtaining the perception information, the following steps are also included:

[0010] Select several areas to sample from the uploaded image;

[0011] Performing clarity assessment on the samples through a deep learning model to generate a regional score;

[0012] Based on the regional score aggregation result, it is determined whether the overall clarity of the uploaded image is qualified.

[0013] Preferably, before acquiring the perception information, a data integrity check is also performed, including the following steps:

[0014] Extract feature signatures from the binary stream of the uploaded image to determine the actual file type;

[0015] Compare the actual document type with the medical standard library to determine the type of medical event;

[0016] Determining the preset required document types based on the type of medical event;

[0017] Compare the actual file type with the required file type to determine the completeness of the identifying information.

[0018] In one embodiment, the perception information further includes the file type of the uploaded image, the preset rule includes a file entry rule, and determining the file entry rule includes the steps of:

[0019] Determine institution-specific document entry rules based on the element layer and the decision layer;

[0020] Review uploaded files according to the described file entry rules.

[0021] In one embodiment, reviewing the perception information further includes the steps of:

[0022] Constructing a standard name directory; the standard name directory includes a standard hospital directory, a diagnosis directory, and a medical details directory;

[0023] Obtain perception information and extract entity names;

[0024] The matching engine performs the matching and outputs the standardized code.

[0025] In one embodiment, reviewing the perception information further includes the steps of:

[0026] Determine access protocols, review rules, and basic confidence weights for multiple data sources;

[0027] Receive data and calculate comprehensive confidence;

[0028] Generate normalized claims dataset according to priority rules.

[0029] In one embodiment, reviewing the perception information further includes the steps of:

[0030] Extract features from perceived information and convert them into risk signals;

[0031] Encoding business knowledge into weighted risk rules, and triggering the weighted risk rules through risk signals;

[0032] The weights of the triggered weighted risk rules are aggregated to generate risk scores, and risk levels are divided according to the risk scores.

[0033] In a second aspect, an embodiment of the present application further provides a full-process intelligent claims processing device for implementing the method described in any one of the embodiments of the first aspect, including: an acquisition module for acquiring perception information; the perception information is the perception information of the image uploaded by the customer. A determination module for identifying the perception information based on artificial intelligence to determine the identification information; the identification information includes medical bill fields, diagnosis codes, or drug details. It is also used to determine the claim type based on the combination of identification information. An output module is used to review the perception information according to the preset rules corresponding to the claim type and output the review information.

[0034] In a third aspect, an embodiment of the present application further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in any one of the embodiments of the first aspect.

[0035] In a fourth aspect, an embodiment of the present application further provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method described in any one of the embodiments of the first aspect when executing the computer program.

[0036] At least one of the above technical solutions adopted in the embodiments of the present application can achieve the following beneficial effects:

[0037] This application realizes the intelligent processing of the entire process of health insurance claims through the innovative design of intelligent image quality detection, multi-source data fusion and five-layer decision-making architecture. Significantly improve the efficiency of claims processing, and shorten the material review time from several hours of traditional manual processing to minutes through intelligent photography and OCR recognition technology. Greatly improve the recognition accuracy, and use deep learning algorithms to accurately analyze unstructured data such as medical bills and diagnostic reports. Effectively reduce operating costs and replace manual operations with automated preliminary reviews. Enhance risk prevention and control capabilities, and realize intelligent risk identification based on the five-layer architecture of elements-logical operations-rules-indicators-models, thereby improving the detection rate of fraud cases. Improve user experience, and the intelligent guidance function for completing materials increases the customer's one-time submission pass rate and shortens the claims cycle. The solution described in this application has achieved a comprehensive improvement in the quality and efficiency of claims services while ensuring that risks are controllable. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0039] Figure 1 A flowchart of a full-process intelligent claims settlement method provided in an embodiment of the present application;

[0040] Figure 2 A flow chart of a method for clarity assessment provided in an embodiment of the present application;

[0041] Figure 3 A flowchart of a method for file integrity assessment provided in an embodiment of the present application;

[0042] Figure 4 A flowchart of a method for determining file entry rules provided in an embodiment of the present application;

[0043] Figure 5 A flow chart of a method for outputting standardized codes provided in an embodiment of the present application;

[0044] Figure 6 A flow chart of a method for generating a normalized claims data set provided in an embodiment of the present application;

[0045] Figure 7 A flow chart of the risk classification method provided in an embodiment of the present application;

[0046] Figure 8 A structural diagram of a full-process intelligent claims processing device provided in an embodiment of the present application;

[0047] Figure 9 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0048] To make the purpose, technical solutions, and advantages of this application more clear, the technical solutions of this application will be clearly and completely described below in conjunction with the specific embodiments of this application and the corresponding drawings. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0049] The following describes in detail the technical solutions provided by various embodiments of the present application in conjunction with the accompanying drawings.

[0050] Through the independently developed full-process intelligent claims system, while ensuring that risks are controllable, we have achieved full automation of the entire process of some cases from reporting to settlement and payment, setting a new benchmark for the industry's intelligent transformation.

[0051] Figure 1 A flowchart of a full-process intelligent claims settlement method provided in an embodiment of the present application includes steps 110 to 130.

[0052] Step 110: Acquire perception information; the perception information is the perception information of the image uploaded by the customer;

[0053] The medical imaging files uploaded by the customer are received through the client interface. The medical imaging files include scanned copies or photos of medical invoices, diagnosis reports, and drug lists.

[0054] For example, the customer used a mobile phone to take photos of the hospital fee invoice (JPEG format) and discharge summary (PDF format).

[0055] Through the intelligent image quality assessment and integrity verification mechanism, the system realizes the automatic pre-processing of materials uploaded by customers. The system adopts an image quality detection model based on deep learning to perform multi-dimensional intelligent analysis of uploaded medical imaging files: first, the clarity features of key parts of the image are extracted through regional sampling technology, and the focus, brightness uniformity and text readability of each area are evaluated using a convolutional neural network; secondly, the file binary signature analysis and metadata verification technology are used to accurately identify the true file format of the image; finally, combined with the medical business rule library, the integrity and compliance of the uploaded materials are dynamically verified. For materials that do not meet the quality standards, the system automatically generates friendly prompts and guides customers to re-upload; for missing materials, it intelligently recommends the file types that need to be supplemented. This step effectively improves the accuracy of subsequent identification and processing through upfront quality control, while optimizing the customer's self-service reporting experience, laying a high-quality original data foundation for the entire process of intelligent claims.

[0056] Step 120: Determine identification information based on the artificial intelligence recognition perception information; the identification information includes medical bill fields, diagnosis codes, or drug details;

[0057] Use deep convolutional neural networks (such as ResNet50) for image classification, combined with OCR technology (such as Tesseract) to extract text information from images uploaded by customers.

[0058] For example, identify key fields such as amount and hospital from invoice images.

[0059] Through intelligent information recognition and structured processing mechanisms, in-depth analysis of medical imaging data can be achieved.

[0060] The system uses multimodal artificial intelligence technology, including computer vision OCR recognition and natural language processing (NLP) algorithms, to accurately analyze uploaded medical receipts, medical records and other images.

[0061] Through pre-trained deep learning models, the system can extract key field information from complex medical documents, including structured data such as medical institution name, diagnosis and treatment date, drug details, and fee amount.

[0062] Especially for complex scenarios unique to the medical industry, such as handwriting, seal coverage, and multi-copy invoices, the system uses adaptive image enhancement technology and contextual semantic understanding algorithms to ensure the accurate extraction of key information.

[0063] The extracted field information is standardized to form a structured data set, providing a reliable data basis for subsequent claim type determination and automated review.

[0064] The embodiments of the present application use AI recognition technology to effectively solve the recognition problems caused by inconsistent formats and diverse content of medical information, and provide accurate data support for the entire process of automated claims.

[0065] Step 130: Determine the claim type based on the combination of identification information, review the perception information according to the preset rules corresponding to the claim type, and output review information.

[0066] Determine the claim type based on a combination of multiple identification information.

[0067] Through intelligent claim type determination and rule review mechanism, automated processing of claim applications is achieved.

[0068] The system first matches the preset claim type matrix based on the extracted key field combination (such as diagnosis code, drug details, fee amount, etc.), and classifies the cases into different types such as routine outpatient treatment, inpatient surgery or special oncology drugs.

[0069] Each claim type is associated with a specific audit rule library, and the system automatically loads the corresponding rules for multi-dimensional verification, including core links such as medical rationality analysis and compliance review.

[0070] The audit process uses rule engine technology to execute automated decision-making, and ultimately outputs structured results that include audit conclusions and risk scores.

[0071] Furthermore, it supports three handling methods: direct approval, conditional approval, and manual review. Through the intelligent classification and review mechanism, it significantly improves the efficiency of claims processing while ensuring the rigor of risk control.

[0072] Furthermore, before acquiring the perception information in step 110, Figure 2 As shown, it also includes the steps:

[0073] Step 100-11, select several areas of the uploaded image for sampling;

[0074] The system uses an adaptive grid division algorithm to dynamically generate a sampling area distribution plan based on image size and content features. Furthermore, it gives priority to covering areas containing key information (such as the bill amount column, diagnosis conclusion, etc.).

[0075] For example, the image is divided into an M×N grid, and K regions (eg, K=9) are randomly selected for sampling.

[0076] Each sampled area is cropped and resized to ensure consistency in subsequent evaluations. This technology effectively addresses the issue of traditional full-image inspection being insensitive to local quality issues.

[0077] Step 100-12: performing clarity evaluation on the sample using a deep learning model to generate a regional score;

[0078] Accurate clarity assessment is achieved through a deep convolutional neural network. The system uses a pre-trained model as its infrastructure, optimized specifically for the characteristics of medical imaging, and outputs a quality score based on the input of the sampled area image.

[0079] For example, through edge detection algorithms, a sharpness score (0-100 points) is calculated for each area.

[0080] For example, evaluation dimensions include core metrics such as text edge sharpness, background noise level, and detail preservation. The model was trained on a dataset containing historical medical bills and achieved high accuracy in recognizing common problems such as blur, reflections, and occlusions.

[0081] Step 100-13: Determine whether the overall clarity of the uploaded image is qualified based on the regional score aggregation result.

[0082] Overall quality judgment is achieved through a multi-level aggregation algorithm.

[0083] The system uses a weighted average strategy to comprehensively score each area, with key information areas given higher weights.

[0084] At the same time, a dynamic threshold mechanism is set up to directly determine unqualified when serious quality defect areas appear (score <30 points).

[0085] For borderline cases (total score within ±5 points of the passing score), the review process is automatically triggered. This judgment method ensures strictness while reducing false rejections due to local problems.

[0086] For example, when the area score exceeding a threshold ratio (eg, 70%) is ≥ 60 points, the overall clarity is determined to be qualified.

[0087] Preferably, before acquiring the perception information in step 110, a data integrity check is also included, such as Figure 3 As shown, it includes the following steps:

[0088] Step 100-21: extract the feature signature from the binary stream of the uploaded image to determine the actual file type;

[0089] For example, this application can achieve accurate identification of file types through binary feature analysis technology. The system uses a file header signature detection algorithm to read the binary feature code of the uploaded file and match it with a preset medical file feature library.

[0090] It can also achieve accurate recognition of document text through image OCR recognition technology. The system performs semantic analysis based on the recognized content and matches it with the preset medical sorting feature library.

[0091] The embodiment of the present application can accurately identify 10 types of standard medical documents, such as medical bills, outpatient medical records, admission and discharge records, and expense lists, effectively avoiding the security risks of relying solely on file name judgment.

[0092] Step 100-22: Compare the actual file type with the medical standard library to determine the medical event type;

[0093] Medical event type classification is achieved through the medical standard intelligent matching engine.

[0094] The system has a built-in database of diagnosis and treatment standards for a large number of medical institutions. Based on the identified file type combination, for example: hospitalization invoice + discharge summary, it automatically associates the corresponding medical event type, for example: hospitalization.

[0095] A knowledge graph-based reasoning algorithm is used to perform multi-dimensional matching based on file content and type features.

[0096] Step 100-23: Determine the preset required file type according to the medical event type;

[0097] A dynamic rules engine determines the required document list. Based on the type of medical event (e.g., outpatient surgery, inpatient treatment), the system retrieves the corresponding material requirements from a pre-set rule library. This includes basic required documents, such as invoices and diagnosis certificates, and conditional required documents, such as surgical records corresponding to surgical expenses. The rule library supports differentiated configuration by region and medical institution level, covering most diagnosis and treatment scenarios nationwide.

[0098] Step 100-24: Compare the actual file type with the required file type to determine the integrity of the identification information.

[0099] Integrity verification is achieved through an intelligent comparison algorithm. The system compares the actual file type with the list of required file types.

[0100] Furthermore, a fuzzy matching technique based on edit distance is used to handle the expression differences.

[0101] The verification results include two types of output: missing file reminders and suspicious file warnings, and automatically generate customer-friendly supplementary material guidance.

[0102] In one embodiment, in step 110, Figure 4 As shown, the perception information further includes the file type of the uploaded image, and the preset rules include file entry rules. Determining the file entry rules includes the following steps:

[0103] Step 110-1: Determine an organization-specific document entry rule based on the element layer and the decision layer;

[0104] Based on the basic fields defined at the element level and the logical conditions set at the decision level, a standardized document entry template suitable for branches in different regions is automatically generated. This template supports differentiated configuration at the field level.

[0105] Step 110 - 2 : Review the uploaded file according to the file entry rules.

[0106] Based on configured file entry rules, uploaded materials undergo a triple check: first, verifying the existence of key fields (e.g., inpatient medical records must include admission records), second, checking the rationality of data values (e.g., drug dosages do not exceed the upper limit of the pharmacopoeia), and finally, verifying cross-field logical relationships (e.g., time consistency between surgical records and anesthesia records). Cases that do not conform to the rules are automatically flagged as exceptions and correction instructions are generated.

[0107] In one embodiment, in step 130, Figure 5 As shown, the review of perception information also includes the following steps:

[0108] Step 130-11: Construct a standard name directory; the standard name directory includes a standard hospital directory, a diagnosis directory, and a medical detail directory;

[0109] Build a standard medical terminology system. For example, a standard hospital directory contains unified coding and attribute information for medical institutions across the country.

[0110] The diagnostic catalog is based on the ICD-10 standards and expands commonly used clinical expressions.

[0111] The medical detailed catalog integrates medical insurance drugs, medical treatment items and consumables standards.

[0112] The catalog data is updated regularly through synchronization with the health department’s data platform.

[0113] Step 130-12: Acquire perception information and extract entity names;

[0114] Implement intelligent extraction of medical entities. For example, identify three types of entities from unstructured text: medical institution entities, diagnosis entities, and medical behavior entities.

[0115] Step 130-13: Perform code matching through the matching engine and output standardized code.

[0116] Complete standardized encoding mapping. For example, the matching engine adopts a three-level mapping strategy:

[0117] Exact matching prioritizes standard encoding, fuzzy matching handles expression differences, and semantic matching resolves terminology variations through knowledge graphs.

[0118] Furthermore, manual review is automatically triggered for unmatched items.

[0119] In one embodiment, in step 130, Figure 6 As shown, the review of perception information also includes the following steps:

[0120] Step 130-21: Determine access protocols, review rules, and basic confidence weights for multiple data sources;

[0121] For example, the system presets access standards for four types of data sources:

[0122] 1. Electronic invoices are obtained through the direct connection interface of the financial platform and transmitted using TLS1.3 encryption, or obtained through the interface of a third-party bill platform and transmitted using encryption.

[0123] 2. HIS data is converted through HL7 / FHIR standards or obtained through the national medical insurance platform.

[0124] 3. The OCR recognition results retain the original confidence score.

[0125] 4. TPA data must pass MD5 integrity verification. All source data must be marked with the collection time and version number.

[0126] Step 130-22, receiving data and calculating comprehensive confidence;

[0127] Perform dynamic confidence assessment. The calculation model considers: data source baseline weight, time decay factor, and historical accuracy correction.

[0128] Automatically downgrade data priority when the comprehensive score is lower than the threshold.

[0129] Step 130-23: Generate a normalized claims data set according to the priority rule.

[0130] Generate normalized claims data. Conflict resolution strategies include prioritizing high-confidence data, adopting consistent data from multiple sources, and manually reviewing key fields.

[0131] The output data contains complete traceability tags, recording the data source and confidence level of each field.

[0132] In one embodiment, in step 130, Figure 7 As shown, the review of perception information also includes the following steps:

[0133] Step 130-31: extract features from the perception information and convert them into risk signals;

[0134] Achieve risk feature conversion. The system extracts risk signals from raw data, including numerical anomalies, logical contradictions, and time sequence anomalies.

[0135] For example, each signal is labeled with a strength value (0-100 points).

[0136] Step 130-32: Encode the business knowledge into weighted risk rules, and trigger the weighted risk rules through risk signals;

[0137] Build a weighted rule engine. Business rules include basic weights, trigger conditions, and disposal measures.

[0138] Step 130-33: The weights of the triggered weighted risk rules are aggregated to generate risk scores, and risk levels are divided according to the risk scores.

[0139] The risk level classification is completed. The aggregation calculation uses a nonlinear function: risk score = Σ(rule weight × signal strength) / attenuation factor.

[0140] For example, there are three risk levels:

[0141] Low-risk cases (<30 points) will be automatically closed.

[0142] Medium risk (30-70 points) conditional payment.

[0143] High-risk (≥70 points) manual investigation.

[0144] The scoring results are fed back to the rule base in real time for model optimization.

[0145] Furthermore, a five-layer risk rating system was constructed. The features extracted at the element layer were converted into risk signals through the logic operation layer. The rule layer encoded business knowledge into weighted risk rules. The model layer performed rating classification based on the output values of the indicator layer.

[0146] The element layer serves as the architectural foundation and is responsible for the structured extraction of raw data.

[0147] Using multimodal feature extraction technology, we accurately extract key fields from unstructured data such as medical images and text. This technology also supports dynamic field expansion, allowing for the addition of new extraction elements as insurance product regulations change.

[0148] The logical operation layer builds the foundational capabilities for risk assessment. By configuring Boolean expressions, the output of the element layer is converted into risk signals. This layer introduces the medical knowledge graph to implement semantic-level operations.

[0149] The rule layer converts insurance business knowledge into executable code. Using a domain-specific language called DSL, it encapsulates three core rules: 1) integrity rules, 2) rationality rules, and 3) compliance rules.

[0150] Each rule is associated with a dynamic weight (, and provincial institutions are supported to adjust according to local medical characteristics. This layer uses a rule engine (such as Drools) to achieve millisecond-level rule matching.

[0151] The indicator layer implements a numerical risk assessment. A multi-dimensional aggregation algorithm integrates parameters such as rule trigger results, data source confidence, and timeliness factors. A nonlinear weighting strategy sets a synergy coefficient of 1.2-1.5 for association rule groups. The typical calculation formula is: Risk Index = Σ(Rule Weight × Time Decay) × Synergy Coefficient. This layer outputs a standardized risk value ranging from 0 to 100 and automatically tags high-risk cases with characteristics (such as "large amount + cross-provincial + no medical history") for use in model-level decision-making.

[0152] The model layer integrates machine learning and business rules to form the final decision.

[0153] For example, a dual-mode architecture is used: for routine cases (risk value <30), the rule engine is directly called to output conclusions. For complex cases, the XGBoost model is used, inputting 28-dimensional features (including indicator layer output and historical claims data) to output three-level handling recommendations.

[0154] A feedback loop is specially set up in this layer to reversely optimize the parameters of each layer based on the manual review results, so as to achieve continuous self-learning of the model.

[0155] Figure 8 A structural diagram of a full-process intelligent claims processing device provided in an embodiment of the present application, used to implement the method described in any embodiment of the first aspect, including:

[0156] The acquisition module 810 is used to acquire perception information; the perception information is the perception information of the image uploaded by the customer.

[0157] Determination module 820 is configured to determine identification information based on artificial intelligence recognition and perception information; the identification information may include medical bill fields, diagnosis codes, or drug details. It is also configured to determine a claim type based on a combination of the identification information.

[0158] The output module 830 is used to review the perception information according to the preset rules corresponding to the claim type and output the review information.

[0159] Furthermore, the acquisition module includes a first acquisition unit for acquiring perception information; the perception information is the perception information of the image uploaded by the customer.

[0160] The determination module includes a first determination unit, which is used to determine identification information based on artificial intelligence recognition perception information; the identification information includes medical bill fields, diagnosis codes or drug details.

[0161] It also includes a second determining unit for determining the claim type according to a combination of identification information.

[0162] The output module includes a first output unit, which is used to review the perception information according to preset rules corresponding to the claim type and output review information.

[0163] In one embodiment, the determination module further includes a third determination unit configured to select a plurality of regional samples from the uploaded image, perform a clarity assessment on the samples using a deep learning model, and generate a regional score. The third determination unit further determines whether the overall clarity of the uploaded image is acceptable based on the aggregated regional scores.

[0164] The above embodiment is used to implement the method described in any one of steps 100-11 to 100-13 of the specification.

[0165] In one embodiment, the determination module further includes a fourth determination unit configured to extract a characteristic signature from the binary stream of the uploaded image to determine the actual file type. The fourth determination unit is further configured to compare the actual file type with a medical standard library to determine the type of medical event. The fourth determination unit is further configured to determine a preset required file type based on the medical event type. The fourth determination unit is further configured to compare the actual file type with the required file type to determine the integrity of the identification information.

[0166] The above embodiment is used to implement the method described in any one of steps 100-21 to 100-24 of the specification.

[0167] In one embodiment, the determination module further includes a fifth determination unit configured to determine an institution-specific file entry rule based on the element layer and the decision layer, and to review the uploaded file based on the file entry rule.

[0168] The above embodiment is used to implement the method described in any one of steps 110 - 1 to 110 - 2 of the specification.

[0169] In one embodiment, the determination module further comprises a sixth determination unit for constructing a standard name directory; the standard name directory comprises a standard hospital directory, a diagnosis directory and a medical detail directory;

[0170] The acquisition module further comprises a second acquisition unit for acquiring perception information and extracting entity names;

[0171] The output module further includes a second output unit, which is used to perform code matching through a matching engine and output a standardized code.

[0172] The above embodiment is used to implement the method described in any one of steps 130-11 to 130-13 of the specification.

[0173] In one embodiment, the determination module further includes a seventh determination unit for determining access protocols, review rules, and basic confidence weights for multiple data sources, and for generating a normalized claims data set according to priority rules.

[0174] The acquisition module further includes a third acquisition unit, which is used to receive data and calculate the comprehensive confidence.

[0175] The above embodiment is used to implement the method described in any one of steps 130-21 to 130-23 of the specification.

[0176] In one embodiment, the acquisition module further includes a fourth acquisition unit for extracting features from the perception information and converting them into risk signals;

[0177] The determination module further includes an eighth determination unit configured to encode business knowledge into weighted risk rules, trigger the weighted risk rules via risk signals, generate risk scores through aggregate calculations based on the weights of the triggered weighted risk rules, and classify risk levels based on the risk scores.

[0178] The above embodiment is used to implement the method described in any one of steps 130-31 to 130-33 of the specification.

[0179] It should be noted that the execution entity of each step of the method provided in Example 1 can be the same device, or the method can be executed by different devices. For example, the execution entity of steps 110 and 120 can be device 1, and the execution entity of step 130 can be device 2; for another example, the execution entity of step 110 can be device 1, and the execution entity of steps 120 and 130 can be device 2; and so on.

[0180] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0181] Therefore, the present application also proposes a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in any embodiment of the present application.

[0182] Furthermore, the present application also proposes an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method described in any embodiment of the present application when executing the computer program.

[0183] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0184] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0185] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0186] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory. Memory may include non-permanent storage in a computer-readable medium, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0187] Figure 9 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device 900 shown is merely an example and should not limit the functionality and scope of use of the embodiments of the present application. It includes: one or more processors 920; a storage device 910 for storing one or more programs. When the one or more programs are executed by the one or more processors 920, the one or more processors 920 implement the full-process intelligent claims processing method provided in an embodiment of the present application, which includes:

[0188] Acquiring perception information; the perception information is the perception information of the image uploaded by the customer;

[0189] Determining identification information based on artificial intelligence recognition perception information; the identification information includes medical bill fields, diagnosis codes, or drug details;

[0190] Determine the claim type based on the combination of identification information, review the perceived information based on the preset rules corresponding to the claim type, and output the review information.

[0191] The electronic device 900 further includes an input device 930 and an output device 940 ; the processor 920 , storage device 910 , input device 930 and output device 940 in the electronic device can be connected via a bus or other means, with the figure taking the connection via bus 950 as an example.

[0192] The storage device 910, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and module units, such as the program instructions corresponding to the full-process intelligent claims processing method in the embodiment of the present application. The storage device 910 may mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system and application programs required for at least one function; the data storage area can store data created based on the use of the terminal, etc. In addition, the storage device 910 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some instances, the storage device 910 may further include a memory remotely located relative to the processor 920, and these remote memories may be connected via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0193] The input device 930 may be used to receive input numbers, character information or voice information, and generate key signal input related to user settings and function control of the electronic device. The output device 940 may include electronic devices such as a display screen and a speaker.

[0194] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0195] Those skilled in the art will appreciate that, unless otherwise specified, the singular forms "a," "an," "the," and "the" may also include the plural forms. It should be understood that when a device or component is "connected" to another device or component, it may be directly connected to the other device or component, or intervening devices or components may be involved. Furthermore, "connected" as used herein may include both partially wireless and partially wired connections.

[0196] In the description of this application, it should be understood that the terms "first", "second", etc. are used for descriptive purposes only and should not be understood as indicating or implying relative importance. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to specific circumstances. In addition, in the description of this application, unless otherwise specified, "multiple" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the previous and subsequent associated objects are in an "or" relationship.

[0197] The foregoing is merely an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A full-process intelligent claims settlement method, characterized by: Contains steps: Acquiring perception information; the perception information is the perception information of the image uploaded by the customer; Determining identification information based on artificial intelligence recognition perception information; the identification information includes medical bill fields, diagnosis codes, or drug details; Determine the claim type based on the combination of identification information, review the perceived information based on the preset rules corresponding to the claim type, and output the review information.

2. The full-process intelligent claims settlement method according to claim 1 is characterized in that: Before obtaining perception information, the following steps are also included: Select several areas to sample from the uploaded image; Performing clarity assessment on the samples through a deep learning model to generate a regional score; Based on the regional score aggregation result, it is determined whether the overall clarity of the uploaded image is qualified.

3. The full-process intelligent claims settlement method according to claim 1 is characterized in that: Before obtaining the perception information, a data integrity check is also performed, including the following steps: Extract feature signatures from the binary stream of the uploaded image to determine the actual file type; Compare the actual document type with the medical standard library to determine the type of medical event; Determining the preset required document types based on the type of medical event; Compare the actual file type with the required file type to determine the completeness of the identifying information.

4. The full-process intelligent claims settlement method according to claim 1 is characterized in that: The perception information further includes the file type of the uploaded image, and the preset rule includes a file entry rule. Determining the file entry rule includes the following steps: Determine institution-specific document entry rules based on the element layer and the decision layer; Review uploaded files according to the described file entry rules.

5. The full-process intelligent claims settlement method according to claim 1 is characterized in that: Reviewing the perception information also includes the following steps: Constructing a standard name directory; the standard name directory includes a standard hospital directory, a diagnosis directory, and a medical details directory; Obtain perception information and extract entity names; The matching engine performs the matching and outputs the standardized code.

6. The full-process intelligent claims settlement method according to claim 1 is characterized in that: Reviewing the perception information also includes the following steps: Determine access protocols, review rules, and basic confidence weights for multiple data sources; Receive data and calculate comprehensive confidence; Generate normalized claims dataset according to priority rules.

7. The full-process intelligent claims settlement method according to claim 1 is characterized in that: Reviewing the perception information also includes the following steps: Extract features from perceived information and convert them into risk signals; Encoding business knowledge into weighted risk rules, and triggering the weighted risk rules through risk signals; The weights of the triggered weighted risk rules are aggregated to generate risk scores, and risk levels are divided according to the risk scores.

8. A full-process intelligent claims processing device for implementing the method described in any one of claims 1 to 7, characterized in that: include: An acquisition module is used to acquire perception information; the perception information is the perception information of the image uploaded by the customer; a determination module for determining identification information based on artificial intelligence recognition of the perceived information; the identification information including medical bill fields, diagnosis codes, or drug details; and for determining a claim type based on a combination of the identification information; The output module is used to review the perception information according to the preset rules corresponding to the claim type and output the review information.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein: When the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

Images