Claim settlement case risk identification method and device, computer equipment and storage medium

Abstract

The embodiment of the invention belongs to the field of artificial intelligence, and relates to a claim settlement case risk identification method, which comprises the steps of obtaining case information and node information of a claim settlement case; when the node information is an input node, inputting the case information into a first risk identification model to obtain a first risk identification result, the first risk identification result displaying that the claim settlement case is high-risk or medium-risk; when it is determined that the claim settlement case is at a medium risk or the node information is an auditing node according to the first risk identification result, the case information is input into a second risk identification model to obtain a second risk identification result, and the second risk identification result shows that the claim settlement case is at a medium risk or a low risk. The invention further provides a claim settlement case risk identification device, computer equipment and a storage medium. According to the invention, the accuracy of insurance claim settlement case risk identification is improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a method, device, computer equipment and storage medium for identifying risks in claims cases. Background Art

[0002] In the modern insurance industry, claims review is an important part of ensuring the interests of both insurance companies and insured persons. When an insured person files a claim, if a pre-existing condition or false disclosure is found, it will usually lead to the insurance company terminating the contract or refusing to pay. At present, in the risk identification process of claims cases, the judgment of whether the insured has a pre-existing condition or false disclosure risk mainly relies on manual review or risk interception based on established business rules. Cases that have been intercepted usually require further risk point evidence investigation by a third-party supplier to confirm that the case is risk-free before compensation can be made.

[0003] Risk control that relies on manual review and rules often seems to be unable to cope with complex cases. Since claims cases involve long time spans and multiple submissions of records, data consistency at different review stages is difficult to ensure, which affects the accuracy of risk assessment. In addition, the risk judgment of the case by the auditor is often limited by personal experience and differences in rule interpretation. This human factor leads to frequent risk omissions or misjudgments. Summary of the invention

[0004] The purpose of the embodiments of the present application is to propose a method, device, computer equipment and storage medium for identifying risks in claims cases to solve the problem of low accuracy in risk identification of claims cases.

[0005] In order to solve the above technical problems, the present application embodiment provides a method for identifying risk of claims cases, which adopts the following technical solutions:

[0006] Obtain case information and node information of claim cases;

[0007] When the node information is an input node, the case information is input into a first risk identification model to obtain a first risk identification result, and the first risk identification result shows that the claim case is high risk or medium risk;

[0008] When the claim case is determined to be medium risk or the node information is an audit node based on the first risk identification result, the case information is input into the second risk identification model to obtain a second risk identification result, and the second risk identification result shows that the claim case is medium risk or low risk.

[0009] In order to solve the above technical problems, the embodiment of the present application also provides a claim case risk identification device, which adopts the following technical solution:

[0010] Case acquisition module, used to obtain case information and node information of claim cases;

[0011] A first identification module, for inputting the case information into a first risk identification model to obtain a first risk identification result when the node information is an input node, wherein the first risk identification result indicates that the claim case is of high risk or medium risk;

[0012] The second identification module is used to input the case information into a second risk identification model to obtain a second risk identification result when it is determined that the claim case is of medium risk or the node information is an audit node based on the first risk identification result. The second risk identification result shows that the claim case is of medium risk or low risk.

[0013] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0014] Obtain case information and node information of claim cases;

[0015] When the node information is an input node, the case information is input into a first risk identification model to obtain a first risk identification result, and the first risk identification result shows that the claim case is high risk or medium risk;

[0016] When the claim case is determined to be medium risk or the node information is an audit node based on the first risk identification result, the case information is input into the second risk identification model to obtain a second risk identification result, and the second risk identification result shows that the claim case is medium risk or low risk.

[0017] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:

[0018] Obtain case information and node information of claim cases;

[0019] When the node information is an input node, the case information is input into a first risk identification model to obtain a first risk identification result, and the first risk identification result shows that the claim case is high risk or medium risk;

[0020] When the claim case is determined to be medium risk or the node information is an audit node based on the first risk identification result, the case information is input into the second risk identification model to obtain a second risk identification result, and the second risk identification result shows that the claim case is medium risk or low risk.

[0021] Compared with the prior art, the embodiments of the present application mainly have the following beneficial effects: the claim case processing is divided into an entry node and a review node, and different risk identification models are used at different nodes. The first risk identification model performs preliminary screening of cases at the entry stage, effectively filters out high-risk cases, and avoids unnecessary manual review investment; the second risk identification model performs further analysis on cases other than high-risk cases at the review stage, and further judges them as medium-risk or low-risk, thereby improving the accuracy of risk judgment and reducing the possibility of misjudgment and missed judgment; the phased processing method improves the efficiency of the claims process, improves the accuracy of risk case identification, reduces the potential losses of insurance companies, and can better manage and control claims risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0023] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0024] Figure 2 is a flow chart of an embodiment of a method for identifying risk of claims cases according to the present application;

[0025] Figure 3 It is a structural schematic diagram of an embodiment of a claim case risk identification device according to the present application;

[0026] Figure 4 It is a structural diagram of an embodiment of a computer device according to the present application. DETAILED DESCRIPTION

[0027] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0028] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0029] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0030] like Figure 1 As shown, the system architecture 100 may include a terminal device 101, a network 102 and a server 103. The terminal device 101 may be a laptop 1011, a tablet computer 1012 or a mobile phone 1013. The network 102 is used to provide a medium for a communication link between the terminal device 101 and the server 103. The network 102 may include various connection types, such as wired, wireless communication links or optical fiber cables.

[0031] The user can use the terminal device 101 to interact with the server 103 through the network 102 to receive or send messages, etc. Various communication client applications can be installed on the terminal device 101, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0032] The terminal device 101 can be various electronic devices with a display screen and supporting web browsing. In addition to the laptop computer 1011, the tablet computer 1012 or the mobile phone 1013, the terminal device 101 can also be an e-book reader, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 (Moving Picture Experts Group Audio Layer IV) player, a laptop computer and a desktop computer, etc.

[0033] The server 103 may be a server that provides various services, such as a background server that provides support for a web page displayed on the terminal device 101 .

[0034] It should be noted that the claim case risk identification method provided in the embodiment of the present application is generally executed by a server, and accordingly, the claim case risk identification device is generally set in the server.

[0035] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0036] Continue to refer Figure 2 , shows a flow chart of an embodiment of a method for identifying risk of claims cases according to the present application. The method for identifying risk of claims cases comprises the following steps:

[0037] Step S201, obtaining case information and node information of the claim case.

[0038] In this embodiment, the claim case risk identification method is executed on an electronic device (eg Figure 1 The server shown in the figure) can communicate with the terminal device through a wired connection or a wireless connection. It should be noted that the above wireless connection method may include but is not limited to 3G / 4G / 5G connection, Wi-Fi connection, Bluetooth connection, Wi MAX connection, Zigbee connection, UWB (Ultra Wi-Fi Band) connection, and other wireless connection methods currently known or to be developed in the future.

[0039] Specifically, the case information and node information of the claim case are obtained. Case information is the relevant information of the claim case. In medical and health-related insurance, case information may include the claim information of the insured person's historical claim cases, claim record information, risk interception information, hospital visit information, bill information, basic information of the insured person, policy-related information, historical and current case report primary diagnosis, secondary diagnosis, disease description information, disease category classification information, OCR (Optical Character Recognition) image analysis information, etc., combined with the previous diseases and previous disease time extracted by OCR, the risk of previous diseases and the risk of false disclosure of the insured person can be identified.

[0040] Node information indicates the stage of case processing, such as the entry stage or the review stage. The node where the claim case first enters the claim system and the case information is entered is recorded as the entry node (or the first review). The subsequent case review process may involve information supplementation, manual negotiation, case closing and other links, which can be collectively referred to as review nodes. At the review node, the risk identification model can be called multiple times.

[0041] In the present application, risk identification models are established for the entry node and the review node respectively, a first risk identification model is established for the entry node, and a second risk identification model is established for the review node.

[0042] Step S202, when the node information is an input node, the case information is input into the first risk identification model to obtain a first risk identification result, and the first risk identification result shows that the claim case is high risk or medium risk.

[0043] Specifically, when the node information is an input node, the case information is input into the first risk identification model. The first risk identification model preliminarily evaluates the risk level of the claim case.

[0044] This application divides the risk level of cases into high risk, medium risk and low risk. Among them, high risk means black, which needs to be intercepted immediately; medium risk means gray, which means there is a certain risk and needs further review or manual review; low risk means white, which does not need to be intercepted.

[0045] At the input node, the first risk identification model outputs only two types of results: high risk and medium risk. High risk needs to be intercepted immediately and is not allowed to enter the subsequent review process; medium risk means that the claim case has a certain risk, but it does not meet the standard for black, and further review is required.

[0046] Step S203, when the claim case is determined to be medium risk or the node information is an audit node according to the first risk identification result, the case information is input into the second risk identification model to obtain a second risk identification result, and the second risk identification result shows that the claim case is medium risk or low risk.

[0047] Specifically, if the result of the first risk identification model shows that the claim case is of medium risk, or the node information shows that the case is at the review node, the case information is input into the second risk identification model. The second risk identification model will output the second risk identification result and classify the claim case as medium risk or low risk. The second risk identification model can conduct a more detailed risk assessment of medium risk cases in the middle and late stages of case processing to ensure the accuracy of risk identification.

[0048] Because the case review spans a long period of time, the same claim case may have multiple submission records, and the feature values ​​of the claim case obtained at different submission record time points will not be completely consistent. If the time points are not distinguished, and the case information is simply input into the model for training, there will be a problem of premature leakage of label data. For example, a claim case calls the risk identification model a total of 3 times at the review node; the case has been intercepted during the second review. At this time, the information obtained by the model, such as the "historical number of intercepted risks", will be included in this interception record, causing the data label to leak in advance, which in turn leads to inaccurate or invalid model predictions. Even if the model is judged as high risk at this time, it is of no help to the auditor.

[0049] To this end, this application establishes risk identification models for the input node and the review node respectively. At the input node, when a claim case enters the system for the first time, it is necessary to identify and intercept claims cases that are obviously high-risk as accurately as possible, reduce the number of high-risk cases entering the review stage, reduce the possibility of data leakage from the source, and prevent these cases from wasting review resources. Therefore, judging high risk is to terminate the further processing of high-risk cases at the preliminary stage.

[0050] At the review node, further evaluation is conducted on those claims that were not judged as high risk (black) at the entry node. These claims may have been initially reviewed or even partially processed manually. During the review stage, case information may have been supplemented or modified. If the model continues to make high-risk judgments, it may make wrong judgments based on incomplete or misleading information (because this information has been evaluated at the entry stage). Therefore, the review stage no longer makes high-risk judgments, but focuses on secondary interception of cases with unclear risks through "gray judgment", and excludes cases that are obviously risk-free through "white judgment".

[0051] The “gray” judgment at the audit node is a prudent measure to ensure that no potential risks are missed after the initial assessment, which helps to reduce missed judgments and ensure that high-risk cases are not mistakenly passed due to omissions in the initial assessment.

[0052] In this embodiment, the claim case processing is divided into an entry node and a review node, and different risk identification models are used at different nodes. The first risk identification model performs a preliminary screening of cases at the entry stage, effectively filters out high-risk cases, and avoids unnecessary manual review investment; the second risk identification model performs further analysis on cases other than high-risk cases at the review stage, and further determines them as medium-risk or low-risk, thereby improving the accuracy of risk judgment and reducing the possibility of misjudgment and omission; the phased processing method improves the efficiency of the claims process, improves the accuracy of risk case identification, reduces the potential losses of insurance companies, and can better manage and control claims risks.

[0053] Furthermore, the above-mentioned step of obtaining case information of the claim case may include: obtaining online log data of the case system; performing stability testing on the online log data; when it is determined that the online log data is stable according to the stability test, extracting information from the online log data to obtain case information of the claim case.

[0054] Specifically, online log data is obtained from the case system. The online log data contains various records generated during the case handling process, such as user operations, system responses, data transmission, etc., covering a variety of information related to the case.

[0055] Perform stability checks on online log data to assess whether the distribution of current data is consistent with the data distribution during model training, and then determine whether the current data is stable. If the feature distribution of the data deviates too much from the data during training, the model prediction results may be inaccurate.

[0056] After the online log data is determined to be stable based on stability testing, useful case information is extracted from the online log data, including identifying and filtering out feature values ​​that are meaningful for risk assessment for subsequent risk identification model evaluation.

[0057] In this embodiment, the stability test of online log data ensures the consistency of online log data at different time points, avoids the negative impact of data distribution anomalies on model prediction results, and enables the risk identification model to perform analysis under the premise of ensuring data quality, thereby improving the accuracy of model predictions; when it is determined that the data is stable, information extraction is used to ensure that the case information is currently the most relevant and useful, further improving the risk identification effect of the model.

[0058] In this application, offline data is no longer used but online log data is used for modeling. Due to the complexity of the risk control system itself, there are often inconsistencies between online and offline data, and because there is a certain review cycle for cases, offline data will lag behind. Recent online log data is selected for modeling. The model learns the online and accumulated data distribution during the current actual call process, the rules of risk judgment, and stable fitting; in addition, the online log data field format has been basically standardized, which can realize rapid iteration of the model, and enhance the robustness and stability of the model at the data level.

[0059] Furthermore, the above-mentioned step of performing stability testing on online log data may include: obtaining historical modeling data of the risk identification model as basic data; obtaining data within a preset time period from the online log data as comparison data; calculating the PS I value of each feature in the basic data and the comparison data; and performing stability testing on the online log data based on the calculated PS I value.

[0060] Specifically, data from the modeling period is obtained from historical data related to the risk identification model (including the first risk identification model and the second risk identification model); the historical modeling data reflects the data distribution during model training and is the reference standard for model risk assessment. It will serve as basic data and contains all feature values ​​and their distribution used by the model during the development and training process.

[0061] Extract data within a preset time range from the latest online log data, such as data within the last two weeks or a month; this data is called comparative data and is used to compare with the basic data. The comparative data reflects the characteristic distribution of case data in the current actual business operation.

[0062] In order to compare the difference between the basic data and the comparison data, the system uses the Population Stability Index (PSI) value as a measurement indicator. The PSI value can quantify the distribution difference of two data sets on a certain feature. The PSI value is calculated for each feature, and the degree of change in the data distribution is judged by these PSI values.

[0063] Based on the calculated PS I value, it is possible to determine whether the online log data is stable. Generally, if the PS I value is lower than a certain threshold (such as 0.1), it means that the data distribution is relatively stable and has not changed significantly; if the PS I value exceeds the threshold, it means that the current data distribution may have changed significantly, which may affect the prediction performance of the model. Therefore, the stability test results can be generated based on the test results of the PS I value.

[0064] In this embodiment, the stability of the data is judged by the difference in feature distribution between the historical modeling data and the current comparison data, so as to find out whether there is data distribution anomaly; the calculation of the PSI value provides a quantitative data stability evaluation indicator for the model, which can respond quickly when the feature distribution changes significantly, reducing the risk of misjudgment and missed judgment caused by changes in data distribution, and enhancing the adaptability and robustness of the model in a dynamic business environment.

[0065] Furthermore, after the above-mentioned step of performing stability testing on online log data, it may also include: when it is determined based on the stability test that there is data distribution anomaly, generating model update evaluation information based on data distribution change information; identifying change factors that cause changes in feature data, and generating rule adjustment information based on the change factors; and sending the model update evaluation information and rule adjustment information to the terminal logged in by the target account.

[0066] Specifically, when the stability test results show that the data distribution is abnormal, the system will analyze these changes and generate corresponding model update evaluation information, which includes the risks that may occur in the model under the current data environment, indicating that the model may no longer adapt to the new data distribution and needs to be retrained or updated. Model update evaluation information provides a basis for subsequent model maintenance and upgrades.

[0067] In order to adjust the model and rules more accurately, it is also possible to identify specific factors that cause changes in feature data. For example, if the number of reports of a certain type of disease suddenly increases over a period of time, the system can identify this change and analyze its impact. By identifying these change factors, corresponding rule adjustment information can be generated. This adjustment information may include the update of risk judgment criteria, the addition of new risk rules, or the optimization of existing rules. Insurance companies can formulate rules in a timely manner based on such changes and deal with cases involving change factors within a certain period of time accordingly.

[0068] After the model update assessment information and rule adjustment information are generated, they are sent to the terminal logged in by the target account, usually the terminal logged in by the account used by the risk control expert or system maintenance personnel. With this information, relevant personnel can quickly understand the risk status of the current system and make corresponding adjustment decisions to ensure the accuracy of the system.

[0069] In this embodiment, by generating model update evaluation information, it is possible to timely evaluate whether the model needs to be updated to cope with changes in data distribution and reduce the risk of model failure; identifying the change factors that cause changes in feature data and generating rule adjustment information ensures that the system can quickly adapt to new risk environments; model update evaluation information and rule adjustment information provide detailed basis for risk control experts or system maintenance personnel, enabling them to adjust risk identification models and related rules in a timely manner, thereby improving the flexibility and adaptability of the claims case risk identification system and enabling the system to maintain a high level of risk identification accuracy in a dynamic business environment.

[0070] Furthermore, after the above-mentioned step of inputting the case information into the second risk identification model to obtain the second risk identification result, it can also include: when the second risk identification result shows that the claim case is medium risk, generating case review information based on the case information and the second risk identification result; and sending the case review information to the terminal logged in by the target account.

[0071] Specifically, when the second risk identification model determines that a claim case is of medium risk, case review information is generated based on this determination result and case information. The case review information provides auditors with a comprehensive case background, enabling them to conduct more accurate risk assessment and judgment.

[0072] The case review information is sent to the terminal where the target account is logged in, which is usually a device operated by an auditor or risk control expert. After the terminal receives the information, the relevant personnel can quickly understand the risk status of the case and conduct necessary reviews.

[0073] In this embodiment, by generating case review information, detailed case background and risk warnings are provided to auditors. When the model cannot fully determine the risk level, it can be supplemented with manual review to reduce misjudgments or omissions due to model limitations. The automatic generation and sending of case review information ensures that auditors can obtain and handle medium-risk cases as soon as possible, reducing the losses that insurance companies may suffer due to incomplete risk identification and improving the accuracy of claims review.

[0074] Furthermore, after the above-mentioned step of sending the case review information to the terminal logged in by the target account, it can also include: extracting each feature value in the case information; matching each feature value with each risk rule in the rule base to obtain the triggered target rule; generating risk explanation information according to the triggered target rule; and displaying the risk explanation information through the terminal.

[0075] Specifically, after the case review information is sent to the auditor's terminal, the terminal extracts various risk-related feature values ​​from the case's composite information, including the insured's age, health status, claims history, insurance amount, etc.

[0076] The extracted feature values ​​are matched with the risk rules stored in the rule base, where each risk rule can be a combination of multiple specific rules. The rule base stores a large number of rules generated based on historical data and risk assessment models to identify high-risk or medium-risk claims. When the feature value of a case matches a risk rule, the rule is considered "triggered" and may generate high or medium risk.

[0077] When it is detected that a feature value triggers certain risk rules, corresponding risk explanation information is generated based on these triggered rules. Risk explanation information usually includes which feature values ​​trigger which rules, and what specific risks these rules may bring. This information can help auditors understand why the case is marked as medium risk and provide detailed explanations about potential risks.

[0078] The generated risk explanation information is displayed to the auditor through the terminal. The auditor can not only understand the reasons for the generation of the risk identification results, but also clearly see the specific factors that trigger the risk. This transparent risk prompt can help auditors make more accurate judgments during the review process and avoid misjudgments or omissions due to insufficient information or misunderstandings.

[0079] In this embodiment, by extracting feature values ​​and matching them with risk rules in the rule base, specific risk points can be accurately identified, thereby improving the accuracy of risk identification. By generating and displaying risk explanation information, a clear basis for risk assessment is provided to auditors, reducing uncertainty and interference from human factors in the audit process, speeding up audit decision-making, and improving the accuracy and fairness of audit results. Auditors can conduct more accurate risk assessments based on this detailed information to ensure the transparency and compliance of the claims process.

[0080] Furthermore, the above-mentioned claims case risk identification method may also include: obtaining each historical claims case that has been processed; binning the characteristic values ​​associated with each historical claims case, and determining the target binning method through binning evaluation indicators; based on the target binning method, generating multiple candidate risk rules through cross-combination of characteristic values; based on screening indicators and business rules, screening and combining candidate risk rules to obtain multiple risk rules, and storing each risk rule in a rule library.

[0081] Specifically, obtain all the historical claims cases that have been processed and have accurate risk assessment results. Group or bin the characteristic values ​​(such as time difference, amount, etc.) associated with each historical claims case to find the characteristic interval that can effectively distinguish risks. You can use a variety of binning methods (such as equal frequency binning, equal interval binning, chi-square binning, etc.), and select the target binning method that can best distinguish risks through binning evaluation indicators (such as KS value).

[0082] Then, the grid method is used to cross-combine the feature values, which may involve a combination of two or three features, and generate a large number of candidate risk rules (such as rules with an interception rate of 80%). They are then initially screened based on the hit rate of the candidate risk rules (such as rules with a hit case number greater than 100), and a group of candidate risk rules with higher confidence are retained.

[0083] Next, we can perform single-factor risk analysis on each feature value to screen out the element values ​​of high-risk factors. Using the element values ​​of these high-risk factors, we can screen out the relevant candidate risk rules, calculate the interception rate of these rules through the validation set (such as the data of the last two months), retain the rules whose interception rate reaches the preset threshold, and sort them in reverse order by interception rate. After screening, the size of the candidate risk rules is greatly reduced, and these rules have a high confidence level (only dozens to 100 high-confidence rules remain).

[0084] Finally, audit specialists or business personnel manually screen and combine the rules, eliminate rules that do not conform to business logic, form the final high-risk rules, and store these rules in the rule library.

[0085] In this embodiment, feature binning and feature value cross-combination make it possible to extract the most valuable feature combination for risk identification from complex data, thereby improving the predictive ability of the rules; by screening and combining candidate rules, risk rules that meet actual business needs can be automatically generated, avoiding the subjectivity and limitations of manual rule-making; after storing these rules in the rule base, when faced with new cases, high-risk cases can be identified with the help of these verified rules, and risk explanations can be provided, thereby improving the efficiency and accuracy of the risk identification process.

[0086] It should be emphasized that in order to further ensure the privacy and security of the above case information, the above case information can also be stored in a blockchain node.

[0087] The blockchain referred to in this application is a new application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, encryption algorithm, etc. Blockchain is essentially a decentralized database, a string of data blocks generated by cryptographic methods. Each data block contains a batch of network transaction information, which is used to verify the validity of its information (anti-counterfeiting) and generate the next block. Blockchain can include the underlying blockchain platform, platform product service layer, and application service layer.

[0088] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0089] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0090] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through computer-readable instructions, and the computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0091] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0092] Further references Figure 3 , as a response to the above Figure 2 In order to realize the method shown in the figure, the present application provides an embodiment of a risk identification device for claim cases, and the device embodiment is Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0093] like Figure 3 As shown, the claim case risk identification device 300 described in this embodiment includes: a case acquisition module 301, a first identification module 302 and a second identification module 303, wherein:

[0094] The case acquisition module 301 is used to obtain case information and node information of the claim case.

[0095] The first identification module 302 is used to input the case information into the first risk identification model to obtain a first risk identification result when the node information is an input node. The first risk identification result shows that the claim case is high risk or medium risk.

[0096] The second identification module 303 is used to input the case information into the second risk identification model to obtain a second risk identification result when the claim case is determined to be medium risk or the node information is an audit node according to the first risk identification result. The second risk identification result shows that the claim case is medium risk or low risk.

[0097] In this embodiment, the claim case processing is divided into an entry node and a review node, and different risk identification models are used at different nodes. The first risk identification model performs a preliminary screening of cases at the entry stage, effectively filters out high-risk cases, and avoids unnecessary manual review investment; the second risk identification model performs further analysis on cases other than high-risk cases at the review stage, and further determines them as medium-risk or low-risk, thereby improving the accuracy of risk judgment and reducing the possibility of misjudgment and omission; the phased processing method improves the efficiency of the claims process, improves the accuracy of risk case identification, reduces the potential losses of insurance companies, and can better manage and control claims risks.

[0098] In some optional implementations of this embodiment, the case acquisition module 301 may include:

[0099] The log acquisition submodule is used to obtain the online log data of the case system.

[0100] The stability detection submodule is used to perform stability detection on online log data.

[0101] The information extraction submodule is used to extract information from the online log data to obtain case information of the claim case when the online log data is determined to be stable according to the stability test.

[0102] In some optional implementations of this embodiment, the stability detection submodule may include:

[0103] The basic determination unit is used to obtain the historical modeling data of the risk identification model as basic data.

[0104] The comparison determination unit is used to obtain data within a preset time period from the online log data as comparison data.

[0105] The PS I calculation unit is used to calculate the PS I value of each feature in the basic data and the comparative data.

[0106] The stability detection unit is used to perform stability detection on the online log data according to the calculated PSI value.

[0107] In some optional implementations of this embodiment, the case acquisition module 301 may further include:

[0108] The evaluation generation submodule is used to generate model update evaluation information based on data distribution change information when it is determined that there is data distribution anomaly based on stability detection.

[0109] The adjustment generation submodule is used to identify the change factors that cause the characteristic data to change, and to generate rule adjustment information based on the change factors.

[0110] The information sending submodule is used to send the model update evaluation information and rule adjustment information to the terminal logged in by the target account.

[0111] In some optional implementations of this embodiment, the claim case risk identification device 300 may further include:

[0112] The review generation module is used to generate case review information based on the case information and the second risk identification result when the second risk identification result shows that the claim case is of medium risk.

[0113] The review sending module is used to send the case review information to the terminal logged in by the target account.

[0114] In some optional implementations of this embodiment, the claim case risk identification device 300 may further include:

[0115] The feature value extraction module is used to extract the feature values ​​in the case information.

[0116] The rule matching module is used to match each feature value with each risk rule in the rule base to obtain the triggered target rule.

[0117] The explanation generation module is used to generate risk explanation information according to the triggered target rules.

[0118] The information display module is used to display risk explanation information through the terminal.

[0119] In some optional implementations of this embodiment, the claim case risk identification device 300 may further include:

[0120] The historical acquisition module is used to obtain all historical claims cases that have been processed.

[0121] The feature binning module is used to bin the feature values ​​associated with each historical claim case and determine the target binning method through binning evaluation indicators.

[0122] The candidate generation module is used to generate multiple candidate risk rules based on the target binning method and through cross-combination of feature values.

[0123] The screening and combination module is used to screen and combine candidate risk rules based on screening indicators and business rules to obtain multiple risk rules, and store each risk rule in a rule library.

[0124] To solve the above technical problems, the present application also provides a computer device. Figure 4 , Figure 4 This is a basic structural block diagram of the computer device in this embodiment.

[0125] The computer device 4 includes a memory 41, a processor 42, and a network interface 43 that are interconnected and communicated through a system bus. It should be noted that the figure only shows a computer device 4 having a memory 41, a processor 42, and a network interface 43, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to a microprocessor, an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a digital processor (DSP), an embedded device, etc.

[0126] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.

[0127] The memory 41 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 41 can be an internal storage unit of the computer device 4, such as a hard disk or memory of the computer device 4. In other embodiments, the memory 41 can also be an external storage device of the computer device 4, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (FlashCard), etc. equipped on the computer device 4. Of course, the memory 41 can also include both the internal storage unit of the computer device 4 and its external storage device. In this embodiment, the memory 41 is generally used to store the operating system and various application software installed on the computer device 4, such as computer-readable instructions of the risk identification method of the claim case. In addition, the memory 41 can also be used to temporarily store various types of data that have been output or are to be output.

[0128] The processor 42 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip in some embodiments. The processor 42 is generally used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is used to run computer-readable instructions stored in the memory 41 or process data, such as computer-readable instructions for running the claim case risk identification method.

[0129] The network interface 43 may include a wireless network interface or a wired network interface. The network interface 43 is generally used to establish a communication connection between the computer device 4 and other electronic devices.

[0130] The computer device provided in this embodiment can execute the above-mentioned claim case risk identification method. This claim case risk identification method can be the claim case risk identification method of each of the above-mentioned embodiments.

[0131] In this embodiment, the claim case processing is divided into an entry node and a review node, and different risk identification models are used at different nodes. The first risk identification model performs a preliminary screening of cases at the entry stage, effectively filters out high-risk cases, and avoids unnecessary manual review investment; the second risk identification model performs further analysis on cases other than high-risk cases at the review stage, and further determines them as medium-risk or low-risk, thereby improving the accuracy of risk judgment and reducing the possibility of misjudgment and omission; the phased processing method improves the efficiency of the claims process, improves the accuracy of risk case identification, reduces the potential losses of insurance companies, and can better manage and control claims risks.

[0132] The present application also provides another embodiment, namely, providing a computer-readable storage medium, wherein the computer-readable storage medium stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor to enable the at least one processor to perform the steps of the claim case risk identification method as described above.

[0133] In this embodiment, the claim case processing is divided into an entry node and a review node, and different risk identification models are used at different nodes. The first risk identification model performs a preliminary screening of cases at the entry stage, effectively filters out high-risk cases, and avoids unnecessary manual review investment; the second risk identification model performs further analysis on cases other than high-risk cases at the review stage, and further determines them as medium-risk or low-risk, thereby improving the accuracy of risk judgment and reducing the possibility of misjudgment and omission; the phased processing method improves the efficiency of the claims process, improves the accuracy of risk case identification, reduces the potential losses of insurance companies, and can better manage and control claims risks.

[0134] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0135] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A method for identifying risk in claims cases, characterized in that: The steps include: Obtain case information and node information of claim cases; When the node information is an input node, the case information is input into a first risk identification model to obtain a first risk identification result, and the first risk identification result shows that the claim case is high risk or medium risk; When the claim case is determined to be medium risk or the node information is an audit node based on the first risk identification result, the case information is input into the second risk identification model to obtain a second risk identification result, and the second risk identification result shows that the claim case is medium risk or low risk.

2. The method for identifying risk of claims cases according to claim 1, characterized in that: The steps of obtaining case information of a claim case include: Obtain online log data from the case system; Performing stability testing on the online log data; When it is determined that the online log data is stable according to the stability detection, information is extracted from the online log data to obtain case information of the claim case.

3. The method for identifying risk of claims cases according to claim 2, characterized in that: The step of performing stability detection on the online log data comprises: Obtain historical modeling data of risk identification models as basic data; Acquire data within a preset time period from the online log data as comparison data; Calculating the PSI value of each feature in the basic data and the comparative data; The online log data is subjected to stability detection according to the calculated PSI value.

4. The method for identifying risk of claims cases according to claim 2, characterized in that: After the step of performing stability detection on the online log data, the method further includes: When it is determined according to the stability detection that there is a data distribution anomaly, generating model update evaluation information according to the data distribution change information; Identifying a change factor that causes a change in the characteristic data, and generating rule adjustment information based on the change factor; The model update evaluation information and the rule adjustment information are sent to the terminal logged in by the target account.

5. The method for identifying risk of claims cases according to claim 1, characterized in that: After the step of inputting the case information into the second risk identification model to obtain the second risk identification result, the method further includes: When the second risk identification result shows that the claim case is of medium risk, generating case review information according to the case information and the second risk identification result; The case review information is sent to the terminal logged in by the target account.

6. The method for identifying risk of claims cases according to claim 5, characterized in that: After the step of sending the case review information to the terminal logged in by the target account, the method further includes: Extracting each feature value in the case information; Matching each characteristic value with each risk rule in the rule base to obtain a triggered target rule; Generate risk explanation information based on the triggered target rules; The risk explanation information is displayed through the terminal.

7. The method for identifying risk of claims cases according to claim 6, characterized in that: The method further comprises: Obtain all historical claims cases that have been processed; The characteristic values ​​associated with each historical claim case are binned, and the target binning method is determined through binning evaluation indicators; Based on the target binning method, multiple candidate risk rules are generated by cross-combining feature values; Based on the screening index and the business rules, the candidate risk rules are screened and combined to obtain a plurality of risk rules, and each risk rule is stored in a rule library.

8. A risk identification device for claims cases, characterized in that: include: Case acquisition module, used to obtain case information and node information of claim cases; A first identification module, for inputting the case information into a first risk identification model to obtain a first risk identification result when the node information is an input node, wherein the first risk identification result indicates that the claim case is of high risk or medium risk; The second identification module is used to input the case information into a second risk identification model to obtain a second risk identification result when it is determined that the claim case is of medium risk or the node information is an audit node based on the first risk identification result. The second risk identification result shows that the claim case is of medium risk or low risk.

9. A computer device, characterized in that: It comprises a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the claim case risk identification method as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the claim case risk identification method according to any one of claims 1 to 7.

Images