Insurance claim settlement method, device, equipment and medium

Abstract

The invention relates to the technical field of insurance claim settlement, and discloses an insurance claim settlement method, device and equipment, and a medium, which are used for accurately classifying claim settlement events based on collected multi-dimensional data of the claim settlement events, and quantifying the loss degree and authenticity score at the same time. Then, analyzing the personal information of the party, and calculating the fraud risk level of the party; and finally, integrating the event assessment result and the personal risk rating, and automatically generating an optimal claim settlement scheme. The whole-process intelligent processing mode can be applied to financial insurance scenes and medical claim settlement scenes, so that all-directional detection of fraud risks is realized, and the fairness and accuracy of the claim settlement process are ensured. Furthermore, the claim settlement period is shortened, the claim settlement processing efficiency is greatly improved, automation and intelligence of insurance claim settlement are realized, and the operation burden of users and insurance companies is effectively relieved.

Description

Technical Field

[0001] The present invention relates to the field of insurance claim settlement technology, and in particular to an insurance claim settlement method, device, equipment and medium. Background Art

[0002] In related technologies, insurance companies typically conduct manual reviews based solely on structured case materials submitted by customers, relying on traditional rule-based engines for decision-making. Due to a lack of intelligent analytical tools, fraudulent activity and the authenticity of case materials rely primarily on the judgment of claims adjusters, resulting in low processing efficiency and a high rate of misjudgment. This manual and inefficient review mechanism not only reduces the accuracy of underwriting conclusions but also forces customers to repeatedly submit paper supporting documents, prolonging the claims settlement cycle and severely undermining the efficiency of financial services and the user experience of healthcare services. Summary of the Invention

[0003] The present invention provides an insurance claims method, apparatus, computer equipment, and medium to address the technical issues that an inefficient manual review mechanism not only reduces the accuracy of underwriting conclusions but also forces customers to repeatedly submit paper supporting documents, prolonging the claims settlement cycle and seriously damaging the efficiency of financial services and the user experience of medical and health services.

[0004] In a first aspect, an insurance claim settlement method is provided, comprising:

[0005] Obtaining event data of claims and personal information of claims adjusters;

[0006] Based on the event data, determine the event type, loss level and authenticity score;

[0007] Generate behavioral profiles of claims adjusters based on personal information, and determine their fraud risk level based on the behavioral profiles;

[0008] Generate insurance claim plans for claim events based on event type, loss level, authenticity score and fraud risk level.

[0009] In a second aspect, an insurance claims settlement device is provided, comprising:

[0010] The acquisition module is used to obtain the event data of the claim event and the personal information of the claim adjuster;

[0011] A determination module, for determining an event type, a loss level, and a authenticity score based on event data;

[0012] A first generation module is used to generate a behavioral profile of the claims adjuster based on the personal information, and to determine the fraud risk level of the claims adjuster based on the behavioral profile;

[0013] The second generation module is used to generate an insurance claim plan for a claim event based on the event type, loss level, authenticity score and fraud risk level.

[0014] In a third aspect, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned insurance claim settlement method when executing the computer program.

[0015] In a fourth aspect, a computer-readable storage medium is provided, which stores a computer program. When the computer program is executed by a processor, the steps of the above-mentioned insurance claim settlement method are implemented.

[0016] In the solution implemented by the above-mentioned insurance claims method, device, computer equipment and storage medium, first, based on the multi-dimensional data collected from the claims event, the claims event is accurately classified, and the degree of loss and authenticity score are quantified. Subsequently, the personal information of the parties is analyzed to calculate their fraud risk level. Finally, the event assessment results (type, loss, authenticity) are integrated with the individual risk rating to automatically generate the optimal claims solution. This full-process intelligent processing model not only achieves comprehensive detection of fraud risks and ensures the fairness and accuracy of the claims process, but also shortens the claims cycle, greatly improves the efficiency of claims processing, realizes the automation and intelligence of insurance claims, and effectively reduces the operational burden on users and insurance companies. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0018] Figure 1 This is a schematic diagram of an application environment of an insurance claim settlement method according to an embodiment of the present invention;

[0019] Figure 2 This is a flow chart of an insurance claim settlement method according to an embodiment of the present invention;

[0020] Figure 3 yes Figure 2 A schematic flow chart of a specific implementation of step S20;

[0021] Figure 4 yes Figure 2 A schematic flow chart of a specific implementation of step S30;

[0022] Figure 5This is a schematic structural diagram of an insurance claim settlement device according to an embodiment of the present invention;

[0023] Figure 6 is a structural diagram of a computer device in one embodiment of the present invention;

[0024] Figure 7 FIG. 2 is another structural diagram of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0025] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0026] The insurance claim settlement method provided by the embodiment of the present invention can be applied in the following situations: Figure 1 In an application environment, the client communicates with the server through the network. The server obtains the event data of the claim event and the personal information of the claims adjuster; based on the event data, determines the event type, loss level and authenticity score; based on the personal information, generates a behavioral profile of the claims adjuster, and determines the fraud risk level of the claims adjuster based on the behavioral profile; based on the event type, loss level, authenticity score and fraud risk level, generates an insurance claim plan for the claim event. Through the above method, not only is all-round detection of fraud risks achieved, ensuring the fairness and accuracy of the claims process, but also shortens the claims cycle, greatly improves the efficiency of claims processing, realizes the automation and intelligence of insurance claims, and effectively reduces the operational burden of users and insurance companies. Among them, the client can be but is not limited to various personal computers, laptops, smart phones, tablets and portable wearable devices. The server can be implemented with an independent server or a server cluster consisting of multiple servers. The present invention is described in detail below through specific embodiments.

[0027] See also Figure 2 As shown, Figure 2 A flowchart of an insurance claim settlement method provided by an embodiment of the present invention includes the following steps:

[0028] S10: Obtaining event data of the claim event and personal information of the claim adjuster.

[0029] The insurance claims settlement method provided by this invention can be applied in the financial and medical fields. In financial scenarios, insurance claims settlement can provide risk protection for investors and financial institutions. When a contractually agreed-upon insurance event occurs, a fast and standardized claims settlement process ensures fund repayment and maintains financial market stability. In the healthcare sector, the insurance claims settlement system, through standardized medical expense review and compensation mechanisms, not only protects patients' medical rights and reduces their financial burden, but also optimizes the allocation efficiency of medical resources.

[0030] Specifically, during the insurance claims process, comprehensive data related to the claim must be collected, including objective factual information such as the time, location, and course of the incident. Furthermore, necessary personal information such as the adjuster's identity, contact information, and bank account details must be obtained in accordance with applicable laws and regulations. This provides a reliable basis for subsequent claims review, liability determination, and compensation calculation.

[0031] For example, in financial claims scenarios, it is necessary to systematically collect objective event data including the time and location of the accident, and proof of loss (such as property appraisal reports, transaction records, etc.); medical claims require obtaining detailed medical records, expense lists, and other medical vouchers. At the same time, it is necessary to standardize the collection of necessary information such as the identity information, contact information, and financial accounts of the claims adjusters. In actual application, regardless of the scenario, technical means such as encrypted transmission and hierarchical authority management are required to provide data support for subsequent intelligent underwriting and rapid claims settlement while ensuring data security, thereby improving claims settlement efficiency and optimizing user experience.

[0032] S20: Based on the event data, determine the event type, loss level and authenticity score.

[0033] In this step, during the insurance claims process, collected event data is analyzed and verified from multiple dimensions to accurately determine the event type, assess the loss level, and verify the authenticity of the event. This improves the accuracy of claims processing and ensures fair and reasonable claims outcomes.

[0034] For example, in financial scenarios, claims review must focus on analyzing transaction records, market fluctuation data, or asset valuation reports to determine whether the loss falls within the insurance coverage, and cross-validate to ensure the authenticity of the data and the reasonableness of the loss amount. In medical scenarios, claims review must verify the necessity and rationality of the medical treatment based on medical records, examination reports, and expense lists to ensure that the treatment items comply with the insurance terms and conditions. At the same time, through medical insurance data comparison and medical expert review, fraud risks such as excessive medical treatment or false diagnosis and treatment can be prevented.

[0035] In one embodiment of the present application, Figure 3As shown, a specific event data analysis solution is provided. In S20, based on the event data, the event type, loss level and authenticity score are determined, which specifically includes the following steps S21-S24:

[0036] S21: Acquire historical claim data of various data types of historical insurance claim events, wherein the data types include image data, voice data, text data, and video data.

[0037] S22: Based on the historical claims data of various data types, a multimodal data analysis model is trained, wherein the multimodal data analysis model includes an image analysis model, a speech processing model, a text analysis processing model, and a video understanding processing model.

[0038] For steps S21-S22, full-dimensional data assets of historical claims events are collected, including image data, voice data, text data, and video data. Based on these structured and unstructured multi-source heterogeneous data, deep learning technology is used to train specialized single-modal analysis models. Among them, the image recognition model is used to extract visual features, the voice processing model is used to parse semantic information, the text analysis model is used to mine key elements of documents, and the video understanding model is used to capture temporal behavior characteristics. Finally, through the multimodal fusion algorithm, the outputs of these single-modal models are jointly modeled, providing core technical support for the intelligent analysis of insurance claims.

[0039] For example, in the multimodal data collection of historical insurance claims, the financial insurance scenario primarily includes: image data, centered around accident scene photos, documenting property damage; voice data, focusing on recordings of claim reports for analysis of the reporter's statements; text data, including structured documents such as financial transaction vouchers, insurance contracts, and claim application forms completed by claims adjusters; and video data, primarily derived from accident scene surveillance footage, used to reconstruct the incident process. The healthcare insurance scenario encompasses: image data, including accident scene photos and medical imaging materials (such as X-rays and CT scans); voice data, focusing on collecting records of doctor-patient communication to assist in determining the diagnosis and treatment process; text data, primarily medical records, including diagnostic reports, prescriptions, and other medical documents; and video data, primarily involving surgical procedure records and other medical imaging. This multidimensional data collection provides targeted training material for the subsequent construction of domain-adapted multimodal analysis models, ensuring that the models accurately understand the specialized content in different business scenarios.

[0040] Through the above method, a multimodal intelligent analysis architecture is adopted to integrate five core modules: text understanding, image recognition, video analysis, voice processing, and sensor data analysis, to build a complete intelligent claims evidence processing system.

[0041] S23: Input the event data into the multimodal data analysis model to obtain the feature vector of the claim event.

[0042] In this step, during the insurance claim analysis process, the collected multi-source heterogeneous data is subjected to standardized preprocessing, including data cleaning and format conversion. The preprocessed data will be input into the corresponding professional models in the multimodal data analysis model for processing according to their data types. For example, the image analysis model performs deep feature extraction on visual data to identify injury characteristics or key indicators of medical images; the speech processing model extracts the reporter's characteristics and key factual elements through voiceprint recognition and semantic analysis; the text analysis model uses natural language processing technology to mine entity relationships and clause matching in documents; the video understanding model restores the event process or diagnosis and treatment operation details through spatiotemporal feature analysis. Each modality model synchronously processes the input data through a parallel computing architecture and outputs a feature vector that has undergone in-depth analysis.

[0043] In actual application scenarios, applicants can submit a variety of supporting materials through multiple channels, including but not limited to: accident description reports in structured text, high-definition on-site inspection photos, surveillance videos recording the entire incident, detailed voice-recorded testimony, and real-time sensor data collected by IoT devices. Feature extraction methods based on multimodal data analysis can more comprehensively and accurately characterize the essential characteristics of claims events, significantly improving the accuracy and efficiency of automated claims processing. The system achieves fusion analysis of multi-source heterogeneous data through a deep neural network architecture: natural language understanding models are used to parse text semantic features, key visual elements from images and videos are extracted based on computer vision technology, speech recognition and sentiment analysis algorithms are used to process audio information, and the spatiotemporal features of sensor data are integrated for cross-validation. This multimodal collaborative analysis mechanism not only enables the intelligent processing of evidence materials, but also can identify potential contradictions through cross-modal feature comparison, significantly improving the accuracy and efficiency of claims review, and providing comprehensive intelligent support for insurance claims decision-making.

[0044] S24: Based on the feature vector, determine the event type, loss level and authenticity score of the claim event.

[0045] In this step, based on the results of multimodal feature extraction, a three-dimensional assessment of the claim event is performed through intelligent analysis. Specifically, the integrated classification model is first used to perform multi-level pattern recognition on the feature vector, accurately outputting major categories such as vehicle damage, medical care, and property, as well as detailed scenario labels such as collision type, disease classification, and loss form. Secondly, the feature vector is quantitatively analyzed to generate a five-level loss level (L1-L5) with a confidence interval. Finally, through cross-modal consistency verification and temporal logic analysis, an authenticity score of 0-100 points and the corresponding risk level warning are output.

[0046] Through the above methods, key information can be extracted from various types of data to determine the authenticity and severity of the incident, thereby improving the accuracy of claims determination.

[0047] In one embodiment of the present application, a specific claim event analysis solution is provided. In S24, based on the feature vector, the event type, loss level, and authenticity score of the claim event are determined. The steps S241-S242 are as follows:

[0048] S241: Based on historical claims data of historical claims events, a claims event analysis model is trained, wherein the claims event analysis model includes an event classification model, a loss assessment model, and an authenticity verification model.

[0049] S242: Input the feature vector into the claim event analysis model to obtain the event type, loss level and authenticity score of the claim event.

[0050] For steps S241-S242, a complete claims event analysis model system is constructed based on the multimodal data of historical claims events. The system includes three core model components: an event classification model, which uses deep learning to distinguish between types of property losses and credit defaults in financial insurance, as well as medical scenarios such as outpatient and inpatient care in health insurance; a loss assessment model, which develops a market-related loss algorithm for financial assets and establishes a standardized assessment system for medical expenses based on diagnosis and treatment projects; and an authenticity verification model, which identifies financial transaction anomalies and medical fraud patterns through cross-modal feature comparison. During the insurance claims processing process, feature vectors are input into these three models in parallel, and structured assessment results are output: precise event type labels (such as "vehicle damage-collision", "medical-Class A cancer", etc.), five levels of loss levels (L1-L5), and an authenticity score of 0-100 points.

[0051] In practical applications, adaptive machine learning models are employed to optimize fraud detection, claims calculations, and other functions through continuous online learning. These models can adapt to the latest data and fraud patterns to keep pace with the changing insurance business. Furthermore, based on technologies such as reinforcement learning and federated learning, these models share anonymous empirical data across multiple nodes within the insurance company, preserving individual privacy while enabling continuous model upgrades and optimizations to enhance judgment accuracy.

[0052] In one embodiment of the present application, a specific claim event analysis model training scheme is provided. In S241, the claim event analysis model is trained based on historical claim data of historical claim events, specifically including the following steps S2411-S2416:

[0053] S2411: Label historical claims data with claim events;

[0054] S2412: Train a deep learning model based on the labeled historical claims data to obtain an event classification model.

[0055] For steps S2411-S2412, structured and unstructured historical claims data are collected from channels such as the insurance company's database and claims record system. These data cover claims events of various insurance businesses. The data include but are not limited to basic information of claims adjusters, insurance policy information, description of claims events, claims amounts, relevant supporting documents, etc. Missing values, outliers and other issues are handled through the data cleaning process to ensure data quality. Historical claims data are uniformly labeled according to the type of claims event. A sample set is constructed from the cleaned and labeled historical claims data, where the sample set includes at least a training set. An adaptive deep learning architecture is selected based on the characteristics of the business scenario, and the selected model is trained using the sample set. By continuously adjusting the model's hyperparameters (such as learning rate, tree depth, number of hidden layer neurons, etc.), combined with the evaluation results of the validation set, the model performance is optimized to ultimately obtain a trained event classification model.

[0056] For example, event type tags for financial and insurance scenarios include: Property Damage: Auto Accidents (collision / scratches / overturnings / spontaneous combustion); Property Damage (fire / water damage / theft / explosion); Natural Disasters (typhoons / floods / earthquakes); Financial Credit: Loan Defaults (personal / corporate credit defaults); Guarantee Insurance (performance bond / bid bond defaults); Trade Credit (buyer bankruptcy / delinquency); Liability Insurance: Third-Party Liability (traffic accidents / premises liability); Professional Liability (medical malpractice / attorney negligence); Product Liability (damage caused by defective products). Event type tags for healthcare scenarios include: Medical Services: Outpatient Treatment (general outpatient / emergency room treatment); Inpatient Treatment (general hospitalization / ICU treatment); Surgical Treatment (major / medium / minor surgery); Disease Type (serious illness / chronic disease); Accidental Injury; Special Medical Treatment: Maternity Protection (childbirth / pregnancy complications); Dental Treatment (implant / orthodontic treatment), etc.

[0057] S2413: Extract loss assessment feature vectors from historical claims data and label the loss assessment feature vectors.

[0058] S2414: Train the ensemble learning model based on the labeled loss assessment feature vector to obtain a loss assessment model.

[0059] For S2413-2414, multi-dimensional loss assessment features are extracted from historical claims data, and relevant personnel annotate the feature vectors with loss level labels based on industry standards. The Stacking ensemble learning framework is used to build a loss assessment model based on the annotated loss assessment feature vectors, where the sample set includes at least a training set. The ensemble learning model is trained using the sample set, and by continuously adjusting the model's hyperparameters (such as learning rate, tree depth, number of hidden layer neurons, etc.), combined with the evaluation results of the validation set, the model performance is optimized, and ultimately a trained loss assessment model is obtained.

[0060] For example, financial scenario features include: quantitative features: loss amount, asset depreciation rate, repair cost, etc.; qualitative features: loss location, damage rating (1-5 levels); temporal features: loss duration, repair cycle, etc. Medical scenario features include: cost features: treatment item costs, drug costs, material costs; clinical features: treatment method, length of stay, surgical level; and compliance features: medical insurance catalog match and treatment necessity score.

[0061] S2415: Extract claim event authenticity feature vectors from historical claim data and label the claim event authenticity feature vectors.

[0062] S2416: Train a logistic regression model based on the labeled claim event authenticity feature vector to obtain an authenticity verification model.

[0063] For steps S2415-S2416, authenticity discrimination features are extracted from multi-dimensional historical claims data, and a team of anti-fraud experts triple-labels the authenticity label (authentic, suspicious, or fraudulent), risk level, and fraud type (first-party, third-party, or group fraud). An enhanced logistic regression framework is used to construct an authenticity verification model based on the labeled authenticity discrimination feature vector, where the sample set includes at least a training set. The enhanced logistic regression model is trained using the sample set, and the model performance is optimized by continuously adjusting the model's hyperparameters (such as learning rate, tree depth, number of hidden layer neurons, etc.) combined with the evaluation results of the validation set, ultimately obtaining a trained authenticity verification model.

[0064] For example, authenticity discrimination features include: cross-modal consistency features: image-text consistency (matching between accident photos and text descriptions); audio-text consistency (matching between the recorded report and the written statement); spatiotemporal consistency (logic of the time and location of the incident); behavioral pattern features (frequency and patterns of historical claims by claims adjusters); correlation analysis with third-party institutions (such as 4S dealerships and hospitals); and abnormal fund flow features. Scenario-specific features: financial scenarios (abnormal transaction flow patterns and asset valuation deviations); and medical scenarios (reasonableness of treatment programs and compliance with drug use).

[0065] S30: Generate a behavioral profile of the claims adjuster based on the personal information, and determine the fraud risk level of the claims adjuster based on the behavioral profile.

[0066] In this step, assessing the fraud risk level of claims adjusters during the insurance claims process is a core component of preventing claims fraud. Based on the adjuster's basic identity information (i.e., personal information), multi-dimensional data sources, including but not limited to the adjuster's social network, historical claims records, and associated financial accounts, are integrated to construct dynamic behavioral profiles using graph computing technology. Based on association analysis based on knowledge graphs, the system can identify risk characteristics such as unusual social circles, high-frequency claims associations, and abnormal capital flows, thereby quantifying the fraud risk level.

[0067] Through the above method, the relationship between users and high-risk groups or suspected fraud accounts can be automatically analyzed, which not only realizes the accurate identification of individual fraud behaviors, but also effectively discovers potential group fraud chains, thereby significantly improving the anti-fraud capabilities of insurance institutions, further reducing the risk of claims fraud, and ensuring the stable operation of claims business.

[0068] For example, in the insurance business, claims adjusters typically include policyholders, insureds, and beneficiaries. However, to effectively prevent insurance fraud risks, the scope of risk control needs to be expanded to include all related parties associated with the claims event. Specifically, in the financial insurance sector, claims adjusters also include: co-owners and mortgagees in property insurance; debtors and guarantors in credit insurance; third-party claimants in liability insurance; and fund managers and counterparties in investment-type insurance. In the health insurance context, relevant claims adjusters also include: medical service providers (attending physicians and medical institutions); parties involved in expense payments (medical insurance agencies, third-party medical management companies); and legal guardians in special circumstances (for minors / incapacitated persons). Furthermore, auxiliary personnel such as insurance intermediaries (brokers, agents) and legal representatives (lawyers, authorized agents) also fall within the scope of claims adjusters that require risk control considerations. This comprehensive, multi-layered approach to defining claims adjusters lays a solid foundation for building a comprehensive insurance anti-fraud system.

[0069] In one embodiment of the present application, Figure 4 As shown, a specific fraud risk level determination scheme is provided. In S30, a behavioral profile of the claims adjuster is generated based on personal information, and the fraud risk level of the claims adjuster is determined based on the behavioral profile. The scheme specifically includes the following steps S31-S33:

[0070] S31: Based on the claims adjuster's personal information, determine the claims adjuster's historical claims information, associated group information, and associated account information.

[0071] S32: Generate a behavioral profile of the claims adjuster based on personal information, historical claims information, associated group information, and associated account information.

[0072] For steps S31-S32, the historical claims records of the claims adjusters are integrated, including time series features such as the number of claims, amount, and type; related group information covers entities such as co-insured persons, service providers (such as 4S stores, medical institutions) and their related attributes; related account information includes transaction frequency, amount pattern and other features of fund current accounts. A social relationship graph analysis model based on graph neural network (GNN) is constructed. The model generates a comprehensive behavioral profile of the claims adjusters through multi-dimensional feature fusion technology. The portrait system includes the following core dimensions: First, from the level of individual behavioral characteristics, the historical claims patterns of the claims adjusters are analyzed (including time series features such as accident frequency, claim amount distribution, accident type preference, etc.); second, in the group association dimension, the social characteristics such as the connection strength with high-risk groups and the density of the capital network are quantified; finally, the behavioral deviation is identified through anomaly detection algorithms (such as recent sudden changes in claim frequency, abnormal fund flow in related accounts, and other risk signals). These features are extracted and aggregated through graph neural networks, ultimately forming a structured behavioral portrait containing 128 risk characteristic factors. Each factor is standardized and weighted, which can not only comprehensively characterize the risk characteristics of claims adjusters, but also support quantitative comparison and dynamic updates, providing accurate data support for subsequent fraud probability calculations.

[0073] S33: Determine the fraud risk level of claims adjusters based on behavioral profiling.

[0074] In this step, based on the constructed refined behavioral portrait, a multimodal risk assessment model is used to conduct a quantitative analysis of the fraud probability of the claimant. The model first uses an integrated learning method (including gradient boosting decision trees and deep neural networks) to perform importance weighting and feature interaction analysis on 128 risk feature factors. The core features are: the abnormality of personal historical behavior (such as the claim frequency exceeding the 95% percentile of the same group), the social network risk transmission index (such as the strength of association with known fraud nodes), and the matching degree of suspicious fund flow patterns (such as the similarity of typical fraud fund paths). Subsequently, the model integrates the evaluation results of each dimension through a Bayesian probability framework, outputs a standardized fraud risk score of 0-1000 points, and automatically divides it into five fraud risk levels (R1-R5).

[0075] S40: Generate an insurance claim plan for the claim event based on the event type, loss level, authenticity score and fraud risk level.

[0076] In this step, the evaluation results based on the four dimensions of event type, loss level, authenticity score and fraud risk level together form an intelligent claims decision matrix, and ultimately generate differentiated processing solutions including quick claims, supplementary investigation, partial payment or rejection, effectively controlling insurance fraud risks while protecting customer rights and interests.

[0077] In one embodiment of the present application, a specific insurance claim settlement plan generation scheme is provided. In S40, an insurance claim settlement plan is generated based on the event type, loss level, authenticity score, and fraud risk level, specifically including the following steps S41-S42:

[0078] S41: Determine whether the claim is fraudulent based on the authenticity score and fraud risk level, and determine whether the claim is within the coverage based on the event type;

[0079] S42: When it is determined that the claim event does not involve suspicion of fraud and is within the scope of insurance coverage, an insurance claim plan is generated based on the event data and loss level.

[0080] For steps S41-S42, during the insurance claim process, dual verification is performed based on the authenticity score and fraud risk level. When the authenticity score is above the preset threshold and the fraud risk level is below the preset level, the claim application is determined to be free of fraudulent suspicion. Subsequently, the nature of the event is confirmed by the event type and intelligently matched with the insurance coverage specified in the policy terms to ensure that the current claim event falls within the coverage. After both the non-fraudulent and coverage conditions are met, the corresponding compensation ratio, the policy's stipulated deductible and compensation limit, and the handling standards for similar historical cases are determined based on the loss level and event data. This generates an insurance claim plan that includes accurate compensation amount calculation, necessary claim condition descriptions, and tiered approval recommendations.

[0081] Through the above method, insurance claim plans are automatically generated based on each claim request, which not only ensures the accuracy and consistency of claim decisions, but also meets the differentiated needs of different business scenarios. Without the need for manual user operation, the speed and accuracy of claims are greatly improved.

[0082] In one embodiment of the present application, a specific risk assessment solution is provided. In S41, based on the authenticity score and the fraud risk level, it is determined whether the claim event is suspected of fraud, and based on the event type, it is determined whether the claim event is within the insurance coverage. The solution specifically includes the following steps S411-S42:

[0083] S411: Compare the authenticity score with a preset score threshold to determine whether the authenticity score is higher than the preset score threshold;

[0084] S412: Compare the fraud risk level with the preset risk level to determine whether the fraud risk level is lower than the preset risk level.

[0085] For steps S411-S412, the calculated authenticity score (using a standardized scale of 0-100) is first compared with the preset score threshold (currently set to 75 points) adjusted dynamically by the industry. When the case score is continuously higher than the threshold, it is determined to have passed the authenticity verification; at the same time, the fraud risk level (five-level classification of R1-R5) is intelligently matched with the acceptable risk level threshold set by the insurance company's risk preference (usually set to R3), requiring that the case risk level must be strictly lower than the critical value. Only when the strict conditions of "authenticity score>threshold" and "risk level<threshold" are met at the same time, will the case enter the subsequent claims process. Through the dynamic threshold adjustment algorithm, this mechanism can perform adaptive optimization based on the risk characteristics of different product lines (such as auto insurance / health insurance) and different regional markets, ensuring the uniformity of risk control standards while taking into account the differences in business scenarios, effectively balancing the dual needs of risk prevention and control and claims efficiency.

[0086] In actual application scenarios, blockchain technology is introduced to achieve full-process data storage. Specifically, from the moment a claim application is submitted, all key data, including multimodal evidence materials submitted by the customer, the assessment report generated by the system analysis, the claims adjuster's handling opinions, and the final compensation decision, are stored on the chain in real time in the form of hash values, forming an unalterable chain of evidence with a timestamp. At the same time, the system deploys a smart contract engine to convert insurance terms into executable digital contracts: when the multimodal analysis results meet the preset compensation conditions (including authenticity score compliance, loss assessment confirmation, insurance liability matching, etc.), the smart contract automatically triggers the compensation instruction, completing the fully automated processing from claim review to fund transfer. Through the above methods, the transparency and credibility of the claims process are ensured, and efficient and accurate automated compensation is achieved, which significantly reduces operating costs and operational risks while protecting customer rights.

[0087] As can be seen, in the above solution, based on the multi-dimensional data collected from claims, claims are first accurately classified, and the extent of loss and authenticity scores are quantified. Subsequently, the personal information of the parties is analyzed to calculate their fraud risk level. Finally, the event assessment results (type, loss, authenticity) are integrated with the individual risk rating to automatically generate the optimal claims solution. This full-process intelligent processing model not only achieves comprehensive detection of fraud risks and ensures the fairness and accuracy of the claims process, but also shortens the claims cycle, significantly improves the efficiency of claims processing, realizes the automation and intelligence of insurance claims, and effectively reduces the operational burden on users and insurance companies.

[0088] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean 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.

[0089] In one embodiment, an insurance claim settlement device is provided, which corresponds one-to-one to the insurance claim settlement method in the above embodiment. Figure 5 As shown, the insurance claim settlement device includes: an acquisition module, a determination module, a first generation module, and a second generation module. The functional modules are described in detail as follows:

[0090] The acquisition module is used to obtain the event data of the claim event and the personal information of the claim adjuster;

[0091] A determination module, for determining an event type, a loss level, and a authenticity score based on event data;

[0092] A first generation module is used to generate a behavioral profile of the claims adjuster based on the personal information, and to determine the fraud risk level of the claims adjuster based on the behavioral profile;

[0093] The second generation module is used to generate an insurance claim plan for a claim event based on the event type, loss level, authenticity score and fraud risk level.

[0094] In one embodiment, the determination module is specifically configured to:

[0095] Acquire historical claims data of various data types of historical insurance claims events, wherein the data types include image data, voice data, text data, and video data;

[0096] Based on historical claims data of various data types, multimodal data analysis models are trained. These models include image analysis models, speech processing models, text analysis processing models, and video understanding processing models.

[0097] Input the event data into the multimodal data analysis model to obtain the feature vector of the claim event;

[0098] Based on the feature vector, the event type, loss level and authenticity score of the claim event are determined.

[0099] In one embodiment, the determination module is further configured to:

[0100] Based on historical claims data, a claims event analysis model is trained. The claims event analysis model includes an event classification model, a loss assessment model, and an authenticity verification model.

[0101] The feature vector is input into the claim event analysis model to obtain the event type, loss level and authenticity score of the claim event.

[0102] In one embodiment, the determination module is further configured to:

[0103] Label historical claims data with claims events;

[0104] Train the deep learning model based on the annotated historical claims data to obtain an event classification model;

[0105] Extract loss assessment feature vectors from historical claims data and label the loss assessment feature vectors with loss level labels;

[0106] The ensemble learning model is trained based on the labeled loss assessment feature vector to obtain a loss assessment model;

[0107] Extract the authenticity feature vector of the claim event from the historical claim data and label the authenticity feature vector of the claim event;

[0108] The logistic regression model is trained based on the labeled claim event authenticity feature vector to obtain the authenticity verification model.

[0109] In one embodiment, the first generating module is specifically configured to:

[0110] Based on the claims adjuster's personal information, determine the claims adjuster's historical claims information, related group information, and related account information;

[0111] Generate behavioral profiles of claims adjusters based on personal information, historical claims information, associated group information, and associated account information;

[0112] Determine the fraud risk level of claims adjusters based on behavioral profiling.

[0113] In one embodiment, the second generating module is specifically configured to:

[0114] Determine whether a claim is fraudulent based on the authenticity score and fraud risk level, and determine whether the claim is within the coverage based on the event type;

[0115] When it is determined that the claim event is not suspected of fraud and is within the coverage, an insurance claim plan is generated based on the event data and loss level.

[0116] In one embodiment, the second generating module is further configured to:

[0117] Comparing the authenticity score with a preset score threshold to determine whether the authenticity score is higher than the preset score threshold;

[0118] Compare the fraud risk level with the preset risk level to determine whether the fraud risk level is lower than the preset risk level.

[0119] The present invention provides an insurance claims settlement device. First, based on the multi-dimensional data of the collected claims events, the claims events are accurately classified, and the degree of loss and authenticity score are quantified. Afterwards, the personal information of the parties is analyzed to calculate their fraud risk level. Finally, the event assessment results (type, loss, authenticity) are integrated with the personal risk rating to automatically generate the optimal claims settlement plan. This full-process intelligent processing mode not only realizes the all-round detection of fraud risks and ensures the fairness and accuracy of the claims process. It also shortens the claims cycle, greatly improves the efficiency of claims processing, realizes the automation and intelligence of insurance claims, and effectively reduces the operational burden of users and insurance companies.

[0120] The specific definitions of the insurance claims settlement device can be found in the definitions of the insurance claims settlement method above and will not be repeated here. Each module in the aforementioned insurance claims settlement device may be implemented in whole or in part through software, hardware, or a combination thereof. Each of the aforementioned modules may be embedded in or independent of a processor in a computer device in hardware form, or may be stored in a computer device memory in software form, so that the processor can call and execute the corresponding operations of each module.

[0121] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 6 As shown. The computer device includes a processor, memory, network interface and database connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes non-volatile and / or volatile storage media and internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external client via a network connection. When the computer program is executed by the processor, it implements the functions or steps on the service side of an insurance claims method.

[0122] In one embodiment, a computer device is provided. The computer device may be a client, and its internal structure diagram may be as follows: Figure 7As shown. The computer device includes a processor, memory, a network interface, a display screen, and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external server via a network connection. When executed by the processor, the computer program implements the functions or steps on the client side of an insurance claims settlement method.

[0123] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the following steps are performed:

[0124] Obtaining event data of claims and personal information of claims adjusters;

[0125] Based on the event data, determine the event type, loss level and authenticity score;

[0126] Generate behavioral profiles of claims adjusters based on personal information, and determine their fraud risk level based on the behavioral profiles;

[0127] Generate insurance claim plans for claim events based on event type, loss level, authenticity score and fraud risk level.

[0128] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0129] Obtaining event data of claims and personal information of claims adjusters;

[0130] Based on the event data, determine the event type, loss level and authenticity score;

[0131] Generate behavioral profiles of claims adjusters based on personal information, and determine their fraud risk level based on the behavioral profiles;

[0132] Generate insurance claim plans for claim events based on event type, loss level, authenticity score and fraud risk level.

[0133] It should be noted that the above functions or steps that can be implemented by the computer-readable storage medium or computer device can be found in the relevant descriptions of the server side and the client side in the aforementioned method embodiment. To avoid repetition, they will not be described one by one here.

[0134] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0135] Those skilled in the art will clearly understand that for the sake of convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0136] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. An insurance claim settlement method, characterized in that: include: Obtaining event data of claims and personal information of claims adjusters; Determining an event type, a loss level, and a authenticity score based on the event data; generating a behavioral profile of the claims adjuster based on the personal information, and determining a fraud risk level of the claims adjuster based on the behavioral profile; An insurance claim settlement plan for the claim event is generated based on the event type, the loss level, the authenticity score, and the fraud risk level.

2. The method according to claim 1, characterized in that The step of determining the event type, loss level, and authenticity score based on the event data specifically includes: Acquire historical claims data of various data types of historical insurance claims events, wherein the data types include image data, voice data, text data, and video data; Training a multimodal data analysis model based on historical claims data of various data types, wherein the multimodal data analysis model includes an image analysis model, a speech processing model, a text analysis processing model, and a video understanding processing model; Inputting the event data into the multimodal data analysis model to obtain a feature vector of the claim event; Based on the feature vector, the event type, the loss level, and the authenticity score of the claim event are determined.

3. The method according to claim 2, characterized in that The step of determining the event type, the loss level, and the authenticity score of the claim event based on the feature vector specifically includes: Training a claims event analysis model based on historical claims data of historical claims events, wherein the claims event analysis model includes an event classification model, a loss assessment model, and an authenticity verification model; The feature vector is input into the claim event analysis model to obtain the event type, the loss level and the authenticity score of the claim event.

4. The method according to claim 3, characterized in that The step of training the claim event analysis model based on the historical claim data of historical claim events specifically includes: Labeling the historical claims data with claims events; Training a deep learning model based on the annotated historical claims data to obtain the event classification model; Extracting loss assessment feature vectors from the historical claims data, and labeling the loss assessment feature vectors with loss level labels; Training the ensemble learning model based on the labeled loss assessment feature vector to obtain the loss assessment model; Extracting claim event authenticity feature vectors from the historical claim data, and labeling the claim event authenticity feature vectors; The logistic regression model is trained based on the labeled claim event authenticity feature vector to obtain the authenticity verification model.

5. The method according to claim 1, wherein The step of generating a behavioral profile of the claims adjuster based on the personal information and determining the fraud risk level of the claims adjuster based on the behavioral profile specifically includes: Based on the personal information of the claims adjuster, determining the claims adjuster's historical claims information, associated group information, and associated account information; generating a behavioral profile of the claims adjuster based on the personal information, the historical claims information, the associated group information, and the associated account information; Based on the behavioral profile, the fraud risk level of the claims adjuster is determined.

6. The method according to any one of claims 1 to 5, characterized in that The step of generating an insurance claim settlement plan for the claim event based on the event type, the loss level, the authenticity score, and the fraud risk level specifically includes: Determine whether a claim is fraudulent based on the authenticity score and fraud risk level, and determine whether the claim is within the coverage based on the event type; When it is determined that the claim event does not involve suspicion of fraud and is within the coverage, an insurance claim plan is generated based on the event data and loss level.

7. The method according to claim 6, characterized in that The steps of determining whether a claim event is suspected of fraud based on the authenticity score and fraud risk level, and determining whether the claim event is within the coverage based on the event type, specifically include: Comparing the authenticity score with a preset score threshold to determine whether the authenticity score is higher than the preset score threshold; The fraud risk level is compared with a preset risk level to determine whether the fraud risk level is lower than the preset risk level.

8. An insurance claim settlement device, characterized in that: include: The acquisition module is used to obtain the event data of the claim event and the personal information of the claim adjuster; a determination module, configured to determine an event type, a loss level, and an authenticity score based on the event data; a first generating module, configured to generate a behavioral profile of the claims adjuster based on the personal information, and determine a fraud risk level of the claims adjuster based on the behavioral profile; The second generating module is used to generate an insurance claim settlement plan for the claim event based on the event type, the loss level, the authenticity score and the fraud risk level.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the insurance claim settlement method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the insurance claim settlement method according to any one of claims 1 to 7 are implemented.

Images