Risk detection method and device for insurance claim settlement data, and related product
By combining training decision tree models and using multiple models, we can identify the types of risk injuries and abnormal situations in the insurance claims process, solve the problem of inaccurate risk judgment in the existing technology, and improve the abnormal situations in the insurance claims process Identification ability.
Patent Information
- Application Number
- CN202410988850.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-23
- Publication Date
- 2024-11-22
AI Technical Summary
The existing technology lacks accurate risk judgment in the insurance claims process, which makes abnormal situations difficult to identify and risks of fraud, loss expansion and leakage.
By obtaining the data on judgment and historical insurance claims, the decision tree model is trained to identify risk injury types and abnormal situations, and combined with the claims interval value and condition interval value, multiple models are used for risk detection.
It improves the accuracy of abnormal situations during the insurance claims process, enhances the ability to judge fraud, loss expansion and leakage risks, and improves detection efficiency and accuracy.
Smart Images

Figure CN119006174A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of manual instructions or the financial field, and in particular to a risk detection method, device and related products for insurance claim data. Background Art
[0002] With the continuous development of technology, vehicles have become one of the important means of transportation for the general public. With the appearance of more and more vehicles on the road, traffic accidents are becoming more and more frequent. Traffic accidents may cause losses to people or vehicles. Most of the losses caused by traffic accidents will be compensated by insurance.
[0003] However, in the process of insurance claims, there are phenomena such as unreasonable treatment plans, over-treatment, unreasonable use of drugs, unreasonable disability assessments and unreasonable compensation. It is also common to use traffic accidents to treat one's own original diseases. In the event of an accident, it is also very common to falsify disability assessment materials and seriously exaggerate the disability level. In addition, there are also criminals who fabricate false personal injury accidents or deliberately cause insurance accidents to defraud insurance compensation. Some claims agents falsify claim materials and demand compensation amounts that exceed the actual losses. Some injured people use traffic accidents to reimburse their own medical expenses for diseases, self-paid drugs that far exceed medical insurance standards, days of missed work and nursing that far exceed the requirements of medical orders, and unreasonable compensation and support.
[0004] During the insurance claims process, existing technologies can pre-train personal injury claims risk models and use the personal injury claims risk models to determine abnormal situations during the insurance claims process.
[0005] However, the personal injury claims risk model trained by existing technologies lacks reference values and cannot accurately judge risks in the ever-changing insurance claims process.
[0006] Therefore, how to increase the accuracy of judging abnormal situations in the insurance claims process has become a technical problem that technical personnel in this field urgently need to solve. Summary of the invention
[0007] Based on the above problems, the present application provides a risk detection method for insurance claim data to improve the accuracy of abnormal situations existing in the insurance claim process.
[0008] The risk detection method for insurance claim data provided by this application includes the following steps:
[0009] Obtain pending insurance claims data and historical insurance claims data;
[0010] parsing the injury data in the historical insurance claim data, and determining first training data in the historical insurance claim data;
[0011] Training a first basic model based on the first training data to obtain a first model, wherein the first basic model is a decision tree model;
[0012] Determining a risk injury type based on the injury data in the historical insurance claim data, wherein the proportion of abnormal historical insurance claim data in the historical insurance claim data corresponding to the risk injury type is higher than a preset proportion;
[0013] Determining second training data from the historical insurance claim data according to the risk injury type;
[0014] Training the first basic model based on the second training data to obtain a second model;
[0015] Get the claim interval value and condition interval value corresponding to the insurance claim;
[0016] Using the claim interval value, the condition interval value and the historical insurance claim data to train the second basic model to obtain a third model;
[0017] The insurance claim data to be judged are respectively used in the first model, the second model and the third model, and the risk detection result corresponding to the insurance claim data to be judged is obtained based on the first model, the second model and the third model.
[0018] In a possible implementation, the historical insurance claim data includes abnormal historical insurance claim data, and the risk detection method for insurance claim data further includes:
[0019] determining abnormal case characteristics in the abnormal historical insurance claims data;
[0020] Adjusting parameters of the first model based on the abnormal case characteristics to obtain an adjusted first model;
[0021] The step of applying the to-be-determined insurance claim data to the first model, the second model, and the third model respectively, and obtaining the risk detection result corresponding to the to-be-determined insurance claim data based on the first model, the second model, and the third model comprises:
[0022] The insurance claim data to be judged are respectively used with the adjusted first model, the second model and the third model, and the risk detection results corresponding to the insurance claim data to be judged are obtained based on the adjusted first model, the second model and the third model.
[0023] In a possible implementation, the risk detection method for insurance claim data further includes:
[0024] Acquire insurance claim data to be verified, wherein the insurance claim data to be verified includes a verification tag, and the verification tag is used to indicate whether the corresponding insurance claim data to be verified is abnormal;
[0025] Using the first model, the second model, and the third model respectively on the insurance claim data to be verified, and obtaining verification results corresponding to the insurance claim data to be verified based on the first model, the second model, and the third model;
[0026] The verification result is compared with the label corresponding to the insurance claim data to be verified to obtain a comparison result, and the comparison result is used to indicate the adjustment of the parameters of the first model, the second model and the third model.
[0027] In a possible implementation, the risk detection method for insurance claim data further includes:
[0028] Determine a cost threshold based on one or more of factors such as the target region corresponding to the historical insurance claim data, characteristics of the injured, injury data, and treatment methods;
[0029] If the cost corresponding to the insurance claim data to be determined is higher than the cost threshold, an indication result of abnormal cost is added to the risk detection result.
[0030] The present application also provides a risk detection device for insurance claim data, the risk detection device for insurance claim data comprising the following modules:
[0031] The first acquisition module is used to acquire the insurance claim data to be judged and the historical insurance claim data;
[0032] A parsing module, configured to parse the injury data in the historical insurance claim data and determine first training data in the historical insurance claim data;
[0033] A first model training module, used for training a first basic model based on the first training data to obtain a first model, wherein the first basic model is a decision tree model;
[0034] A risk injury determination module, used to determine a risk injury type based on the injury data in the historical insurance claim data, wherein the proportion of abnormal historical insurance claim data in the historical insurance claim data corresponding to the risk injury type is higher than a preset proportion;
[0035] A second training data determination module, configured to determine second training data in the historical insurance claim data according to the risk injury type;
[0036] A second model training module, used for training the first basic model based on the second training data to obtain a second model;
[0037] The second acquisition module is used to obtain the claim interval value and condition interval value corresponding to the insurance claim;
[0038] A third model training module, used to train the second basic model using the claim interval value, the condition interval value and the historical insurance claim data to obtain a third model;
[0039] The risk detection module is used to use the first model, the second model and the third model to respectively detect the insurance claim data to be judged, and obtain the risk detection result corresponding to the insurance claim data to be judged based on the first model, the second model and the third model.
[0040] In a possible implementation, the historical insurance claim data includes abnormal historical insurance claim data, and the risk detection device for insurance claim data further includes:
[0041] An abnormal case feature determination module, used to determine abnormal case features in the abnormal historical insurance claims data;
[0042] A first parameter adjustment module, used to adjust the parameters of the first model based on the abnormal case characteristics to obtain an adjusted first model;
[0043] The risk detection module is specifically used for:
[0044] The insurance claim data to be judged are respectively used with the adjusted first model, the second model and the third model, and the risk detection results corresponding to the insurance claim data to be judged are obtained based on the adjusted first model, the second model and the third model.
[0045] In a possible implementation, the risk detection device for insurance claim data further includes:
[0046] A third acquisition module is used to acquire the insurance claim data to be verified, wherein the insurance claim data to be verified includes a verification tag, and the verification tag is used to indicate whether the corresponding insurance claim data to be verified is abnormal;
[0047] A verification module, using the first model, the second model and the third model respectively to obtain a verification result corresponding to the insurance claim data to be verified based on the first model, the second model and the third model;
[0048] The second parameter adjustment module is used to compare the verification result with the label corresponding to the insurance claim data to be verified to obtain a comparison result, and the comparison result is used to indicate the adjustment of the parameters of the first model, the second model and the third model.
[0049] In a possible implementation, the risk detection device for insurance claim data further includes:
[0050] A cost threshold determination module, used to determine a cost threshold based on one or more of factors such as a target region corresponding to the historical insurance claim data, characteristics of the injured, injury data, and treatment methods;
[0051] The expense detection module is used to add an expense abnormality risk detection result to the risk detection result if the expense corresponding to the insurance claim data to be judged is higher than the expense threshold.
[0052] The present application also provides an electronic device, the electronic device comprising a processor and a memory:
[0053] The memory is used to store a computer program and transmit the computer program to the processor;
[0054] The processor is used to execute the steps of the above-mentioned risk detection method for insurance claim data according to the instructions in the computer program.
[0055] The present application also provides a computer-readable storage medium, characterized in that the computer-readable storage medium is used to store a computer program, and when the computer program is executed by an electronic device, the steps of the risk detection method for insurance claim data are implemented.
[0056] Compared with the prior art, this application has the following beneficial effects:
[0057] The method provided in the present application first obtains the insurance claim data to be judged and the historical insurance claim data, and determines the first training data in the historical insurance claim data by parsing the injury data in the historical insurance claim data. The first model is obtained by training the first basic model based on the first training data, and the first basic model is a decision tree model. Since the first basic model is a decision tree model, the decision tree model is more suitable for the complex scenario of insurance claims, and the efficiency of insurance claim risk detection can be increased by selecting the decision tree model. Since the training of the first model utilizes the injury data in the historical insurance claim data, and combines the basic model in the form of a decision tree, the first basis has a more accurate judgment ability for fraudulent insurance claim data. Then, the risk injury type is determined according to the injury data in the historical insurance claim data, and the second training data is determined in the historical insurance claim data according to the risk injury type. The second model is obtained by training the first basic model based on the second training data. Since the second model is also a decision tree model, and the second model refers to the risk injury type in the historical insurance claim data, the second model can make a more accurate judgment on the loss expansion situation in the insurance claim process. Finally, the second basic model is trained using the claim interval value, condition interval value and historical insurance claim data to obtain the third model. When training the third model, since the claim interval value and condition interval value are used, the third model can more accurately predict the occurrence of leakage by fully learning the interval values in the insurance claim process. This application uses the first model, the second model and the third model in combination, so that the insurance claim risk can be accurately and quickly judged in complex insurance claim scenarios. BRIEF DESCRIPTION OF THE DRAWINGS
[0058] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0059] Figure 1 A flow chart of a risk detection method for insurance claims data provided in an embodiment of the present application;
[0060] Figure 2 A schematic diagram of the structure of a risk detection device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0061] As described above, the existing technology can pre-train the personal injury claim risk model and use the personal injury claim risk model to judge abnormal situations in the insurance claim process. However, the personal injury claim risk model in the existing technology cannot accurately assess the insurance claim risk due to the lack of reference values in various situations. In addition, since the personal injury claim risk model in the relevant technology uses the Logistic regression algorithm and the BP neural network algorithm to assess the risk of insurance claim data, this leads to low training efficiency and low decision-making efficiency of the personal injury claim risk model.
[0062] After research, it was found that in the scenario of insurance claims, especially auto insurance claims, the main risks can be summarized into three categories. The first category is fraud risks, which include but are not limited to situations where the personal injury is inconsistent with the auto insurance judgment or the disability appraisal materials are falsified. The second category is the risk of loss expansion, which means the expansion of losses. The risk of loss expansion includes but is not limited to situations where the compensation corresponding to the personal injury far exceeds the average value of the injury or the injury of the injured person does not fall within the scope of compensation but is compensated. The third category is the risk of leakage. In the scenario of auto insurance claims, leakage means generating greater compensation within the interval, or generating compensation that exceeds the maximum value of the interval. The risk of leakage is not limited to situations where the maximum compensation amount appears in the interval of the compensation amount corresponding to the insurance claim.
[0063] 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 drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0064] It is understandable that the method provided in the present application can be applied to a processing device, which is a processing device that can perform risk detection on insurance claim data, for example, a terminal device or server that can perform risk detection on insurance claim data. The method provided in the present application can be executed independently by a terminal device or a server, or it can be applied to a network scenario in which a terminal device and a server communicate, and is executed by the cooperation of a terminal device and a server. Among them, the terminal device can be a computer, a mobile phone and other devices. The server can be understood as an application server or a Web server. In actual deployment, the server can be a stand-alone server or a cluster server.
[0065] Figure 1 A flow chart of a risk detection method for insurance claim data provided in this application, the method comprising the following steps:
[0066] S101: Obtaining insurance claim data to be determined and historical insurance claim data.
[0067] The processing device obtains insurance claim data to be determined and historical insurance claim data. The insurance claim data to be determined is insurance claim data waiting to be determined whether there is a risk, and the historical insurance claim data is insurance claim data that has been processed.
[0068] Historical insurance claims data can be obtained in a database. In order to have sufficient data support when training the model, in a possible implementation, the processing device can pre-build the database required for model training.
[0069] The database may include injury data, disability standard evaluation data, injury relationship data, medical insurance information data, prosthetic information data, income data, legal document data, hospital information data and appraisal information data, etc.
[0070] Injury data may include injury types of various levels and corresponding injury codes, diagnostic basis and treatment principles corresponding to different injury types, reference medical expenses corresponding to each injury type, reference value of disability compensation coefficient, surgery rate, disability rate, litigation rate, time point of disability assessment, time point of litigation and other data.
[0071] The disability standard assessment data may include data such as applicable injury type, corresponding clauses, disability level and corresponding disability description.
[0072] Injury relationship data may include data such as the relationship between injury type and medicine, diagnosis and treatment, and disability.
[0073] Medical insurance information data may include data such as drug information and medical treatment information that falls within the scope of medical insurance.
[0074] The prosthetic information data may include data such as reference prices corresponding to prosthetics of different parts and models.
[0075] Income data can include data such as disposable income and consumer spending in various regions.
[0076] Legal document data may include data such as laws, regulations and judicial interpretations related to road traffic and auto insurance in various regions.
[0077] Hospital information data includes social credit codes, detailed addresses, longitude and latitude, and hospital grades corresponding to multiple hospitals.
[0078] The appraisal information data includes the appraisal agency and appraiser’s social credit code, detailed address, appraisal qualifications and other data that conduct the personal injury appraisal.
[0079] S102: Analyze injury data in historical insurance claim data, and determine first training data in the historical insurance claim data.
[0080] The processing device analyzes the injury data in the historical insurance claim data and determines the first training data in the historical insurance claim data. Since the injury is an important reference data for insurance claims, especially for auto insurance claims, the injury data of the injured directly determines the amount of insurance claims. The processing device can determine whether there is an anomaly in the historical insurance claim data through analysis.
[0081] For example, the injury data in a historical insurance claim data indicates that the injury type of the injured person is frostbite, but in the scenario of auto insurance claims, frostbite will not occur. The processing device can determine through analysis that the historical insurance claim data is an abnormal historical insurance claim number, and the data can be used as the first training data.
[0082] For another example, the injury data in a historical insurance claim data indicates that the injury type of the injured person is a burn. The processing device can preliminarily analyze and determine that the historical insurance claim data is a normal historical insurance claim number, and the data can also be used as the first training data.
[0083] For another example, the injury data in a historical insurance claim data indicates that the injury type of the injured person is inconsistent with the injury type used to determine the compensation. The injured person's injury type is injury A, but the injury type used to determine the compensation amount is injury B. The processing device can determine that there is an anomaly in the historical insurance claim data through analysis, and the data can be used as the first training data.
[0084] S103: Training a first basic model based on first training data to obtain a first model.
[0085] The processing device trains the first basic model based on the first training data to obtain the first model. The first basic model in the embodiment of the present application is a decision tree model. Since the structure of the tree in the decision tree model is visualized, the decision tree model can make it easier for business personnel to understand and explain, thereby increasing the efficiency of training to obtain the first model.
[0086] In one possible implementation, the processing device may determine abnormal case features in abnormal historical insurance claim data. Abnormal case features include, but are not limited to, delayed reporting, multiple appearances of the accident driver, reporting at a location other than the accident scene, and the relationship between the persons in danger. In the process of insurance claims, if there are abnormal case features, the risk of the insurance claim will be greater than the risk without abnormal case features. The processing device may adjust the parameters of the first model based on the abnormal case features to obtain the adjusted first model. Adjusting the parameters of the first model may be adjusting the decision tree of the first model.
[0087] S104: Determine the risk injury type based on the injury data in the historical insurance claim data.
[0088] The processing device determines the risk injury type according to the injury data in the historical insurance claim data. The proportion of abnormal historical insurance claim data in the historical insurance claim data corresponding to the risk injury type is higher than a preset proportion.
[0089] S105: Determine second training data from historical insurance claim data according to the risk injury type.
[0090] Since the insurance claim risk of the risk injury is relatively high, the processing device can use the risk injury to determine the second training data in the historical insurance claim data.
[0091] S106: Train the first basic model based on the second training data to obtain a second model.
[0092] The processing device trains the first basic model based on the second training data to obtain a second model. Since the insurance claim risk of risk injuries is relatively high, the second model trained using risk injuries can more accurately judge risk injuries.
[0093] S107: Obtain the claim interval value and condition interval value corresponding to the insurance claim.
[0094] In the process of insurance claims, there are claim interval values and condition interval values. For example, a claim interval value corresponding to a historical insurance claim data is 1000-5000. When making an insurance claim, the final compensation amount needs to be determined within 1000-5000 according to the specific situation.
[0095] For example, the conditional interval value corresponding to a historical insurance claim data is the age of the injured person, and the conditional interval value is 20-80 years old. The older the injured person is, the greater the amount of compensation.
[0096] S108: Using the claim interval value, the condition interval value and the historical insurance claim data to train the second basic model to obtain a third model.
[0097] The processing device uses the claim interval value, the condition interval value and the historical insurance claim data to train the second basic model to obtain the third model. Since the claim interval value and the condition interval value are used when training the third model, the third model can accurately evaluate the interval value in the insurance claim process and effectively prevent the occurrence of leakage.
[0098] S109: Using the first model, the second model and the third model respectively to detect the insurance claim data to be determined, and obtaining risk detection results corresponding to the insurance claim data to be determined based on the first model, the second model and the third model.
[0099] The risk detection result can indicate the insurance claim risk corresponding to the insurance claim data to be determined.
[0100] The method provided in the present application first obtains the insurance claim data to be judged and the historical insurance claim data, and determines the first training data in the historical insurance claim data by parsing the injury data in the historical insurance claim data. The first model is obtained by training the first basic model based on the first training data, and the first basic model is a decision tree model. Since the first basic model is a decision tree model, the decision tree model is more suitable for the complex scenario of insurance claims, and the efficiency of insurance claim risk detection can be increased by selecting the decision tree model. Since the training of the first model utilizes the injury data in the historical insurance claim data, and combines the basic model in the form of a decision tree, the first basis has a more accurate judgment ability for fraudulent insurance claim data. Then, the risk injury type is determined according to the injury data in the historical insurance claim data, and the second training data is determined in the historical insurance claim data according to the risk injury type. The second model is obtained by training the first basic model based on the second training data. Since the second model is also a decision tree model, and the second model refers to the risk injury type in the historical insurance claim data, the second model can make a more accurate judgment on the loss expansion situation in the insurance claim process. Finally, the second basic model is trained using the claim interval value, condition interval value and historical insurance claim data to obtain the third model. When training the third model, since the claim interval value and condition interval value are used, the third model can more accurately predict the occurrence of leakage by fully learning the interval values in the insurance claim process. This application uses the first model, the second model and the third model in combination, so that the insurance claim risk can be accurately and quickly judged in complex insurance claim scenarios.
[0101] After obtaining the first model, the second model and the third model, the processing device may also verify the performance of the three models. Specifically, the processing device may obtain the insurance claim data to be verified, the insurance claim data to be verified includes a verification tag, and the verification tag is used to indicate whether the corresponding insurance claim data to be verified is abnormal.
[0102] The processing device uses the first model, the second model and the third model respectively to process the insurance claim data to be verified, and obtains verification results corresponding to the insurance claim data to be verified based on the first model, the second model and the third model.
[0103] If the verification result meets the preset verification conditions, it proves that the first model, the second model and the third model achieve the expected performance, and the first model, the second model and the third model can be used for risk detection of insurance claim data.
[0104] If the verification result does not meet the preset verification condition, the comparison result can be used to indicate that the parameters of the first model, the second model and the third model should be adjusted.
[0105] After adjusting the parameters, the above verification can be repeated again until the preset verification conditions are met.
[0106] The embodiments of the present application can continuously optimize the performance of the model and improve decision-making efficiency through the verification process.
[0107] Insurance claims, especially auto insurance claims, are affected by many different factors, such as age and region. In order to fully consider different factors, the processing device can use the target region corresponding to the historical insurance claims data, the characteristics of the injured, the injury data and the treatment method to determine the cost threshold. The injury characteristics can include the injured person's age, gender, occupation, income, location and relationship with the car owner / driver.
[0108] In a possible implementation, the cost threshold may be determined by additional information such as the injured person's length of hospital stay, severity of injury, treatment method, treatment hospital, and appraisal agency.
[0109] If the cost corresponding to the insurance claim data to be determined is higher than the cost threshold, an indication result of the cost anomaly is added to the risk detection result.
[0110] The present application also provides a Figure 2 The structure diagram of the risk detection device for insurance claim data shown in FIG. 200 includes:
[0111] The first acquisition module 201 is used to acquire the insurance claim data to be determined and the historical insurance claim data;
[0112] A parsing module 202, configured to parse the injury data in the historical insurance claim data and determine first training data in the historical insurance claim data;
[0113] A first model training module 203, configured to train a first basic model based on the first training data to obtain a first model, wherein the first basic model is a decision tree model;
[0114] The risk injury determination module 204 is used to determine the risk injury type according to the injury data in the historical insurance claim data, and the proportion of abnormal historical insurance claim data in the historical insurance claim data corresponding to the risk injury type is higher than a preset proportion;
[0115] A second training data determination module 205, configured to determine second training data in the historical insurance claim data according to the risk injury type;
[0116] A second model training module 206, configured to train the first basic model based on the second training data to obtain a second model;
[0117] The second acquisition module 207 is used to obtain the claim interval value and condition interval value corresponding to the insurance claim;
[0118] A third model training module 208, configured to train the second basic model using the claim interval value, the condition interval value and the historical insurance claim data to obtain a third model;
[0119] The risk detection module 209 is used to use the first model, the second model and the third model to respectively detect the insurance claim data to be determined, and obtain the risk detection result corresponding to the insurance claim data to be determined based on the first model, the second model and the third model.
[0120] In a possible implementation, the historical insurance claims data includes abnormal historical insurance claims data, and the apparatus further includes:
[0121] An abnormal case feature determination module, used to determine abnormal case features in the abnormal historical insurance claims data;
[0122] A first parameter adjustment module, used to adjust the parameters of the first model based on the abnormal case characteristics to obtain an adjusted first model;
[0123] The risk detection module is specifically used for:
[0124] The insurance claim data to be judged are respectively used with the adjusted first model, the second model and the third model, and the risk detection results corresponding to the insurance claim data to be judged are obtained based on the adjusted first model, the second model and the third model.
[0125] In a possible implementation, the device further includes:
[0126] A third acquisition module is used to acquire the insurance claim data to be verified, wherein the insurance claim data to be verified includes a verification tag, and the verification tag is used to indicate whether the corresponding insurance claim data to be verified is abnormal;
[0127] A verification module, using the first model, the second model and the third model respectively to obtain a verification result corresponding to the insurance claim data to be verified based on the first model, the second model and the third model;
[0128] The second parameter adjustment module is used to compare the verification result with the label corresponding to the insurance claim data to be verified to obtain a comparison result, and the comparison result is used to indicate the adjustment of the parameters of the first model, the second model and the third model.
[0129] In a possible implementation, the device further includes:
[0130] A cost threshold determination module, used to determine a cost threshold based on one or more of factors such as a target region corresponding to the historical insurance claim data, characteristics of the injured, injury data, and treatment methods;
[0131] The expense detection module is used to add an expense abnormality risk detection result to the risk detection result if the expense corresponding to the insurance claim data to be judged is higher than the expense threshold.
[0132] An embodiment of the present application also provides a risk detection device for insurance claim data, wherein the device includes a memory and a processor, the memory is used to store instructions or codes, and the processor is used to execute the instructions or codes so that the device performs the steps of the risk detection method for insurance claim data described in any embodiment of the present application.
[0133] In practical applications, the computer-readable storage medium may be any combination of one or more computer-readable media, which may be a computer-readable signal medium or a computer-readable storage medium.
[0134] The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, device, or device.
[0135] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.
[0136] The program code embodied on the computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
[0137] Computer program code for performing the operations of the present invention may be written in one or more programming languages or a combination thereof, including object-oriented programming languages such as Java, Smalltalk, C++, and conventional procedural programming languages such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).
[0138] It should be noted that each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. The device embodiment described above is merely schematic, in which the unit described as a separate component may or may not be physically separated, and the component prompted as a unit may or may not be a physical unit, that is, it may be located in one place, or it may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative work.
[0139] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.
Claims
1. A risk detection method for insurance claim data, characterized in that: include: Obtain pending insurance claims data and historical insurance claims data; parsing the injury data in the historical insurance claim data, and determining first training data in the historical insurance claim data; Training a first basic model based on the first training data to obtain a first model, wherein the first basic model is a decision tree model; Determining a risk injury type based on the injury data in the historical insurance claim data, wherein the proportion of abnormal historical insurance claim data in the historical insurance claim data corresponding to the risk injury type is higher than a preset proportion; Determining second training data from the historical insurance claim data according to the risk injury type; Training the first basic model based on the second training data to obtain a second model; Get the claim interval value and condition interval value corresponding to the insurance claim; Using the claim interval value, the condition interval value and the historical insurance claim data to train the second basic model to obtain a third model; The insurance claim data to be judged are respectively used in the first model, the second model and the third model, and the risk detection result corresponding to the insurance claim data to be judged is obtained based on the first model, the second model and the third model.
2. The method according to claim 1, characterized in that The historical insurance claims data includes abnormal historical insurance claims data, and the method further includes: determining abnormal case characteristics in the abnormal historical insurance claims data; Adjusting parameters of the first model based on the abnormal case characteristics to obtain an adjusted first model; The method of using the first model, the second model and the third model respectively to obtain the risk detection result corresponding to the insurance claim data to be determined based on the first model, the second model and the third model comprises: The insurance claim data to be judged are respectively used with the adjusted first model, the second model and the third model, and the risk detection result corresponding to the insurance claim data to be judged is obtained based on the adjusted first model, the second model and the third model.
3. The method according to claim 1, characterized in that The method further comprises: Acquire insurance claim data to be verified, wherein the insurance claim data to be verified includes a verification tag, and the verification tag is used to indicate whether the corresponding insurance claim data to be verified is abnormal; Using the first model, the second model, and the third model respectively on the insurance claim data to be verified, and obtaining verification results corresponding to the insurance claim data to be verified based on the first model, the second model, and the third model; The verification result is compared with the label corresponding to the insurance claim data to be verified to obtain a comparison result, and the comparison result is used to indicate the adjustment of the parameters of the first model, the second model and the third model.
4. The method according to claim 1, characterized in that: The method further comprises: Determine a cost threshold based on one or more of factors such as the target region corresponding to the historical insurance claim data, characteristics of the injured, injury data, and treatment methods; If the cost corresponding to the insurance claim data to be determined is higher than the cost threshold, an indication result of abnormal cost is added to the risk detection result.
5. A risk detection device for insurance claim data, characterized in that: include: The first acquisition module is used to acquire the insurance claim data to be judged and the historical insurance claim data; A parsing module, configured to parse the injury data in the historical insurance claim data and determine first training data in the historical insurance claim data; A first model training module, used for training a first basic model based on the first training data to obtain a first model, wherein the first basic model is a decision tree model; A risk injury determination module, used to determine a risk injury type based on the injury data in the historical insurance claim data, wherein the proportion of abnormal historical insurance claim data in the historical insurance claim data corresponding to the risk injury type is higher than a preset proportion; A second training data determination module, configured to determine second training data in the historical insurance claim data according to the risk injury type; A second model training module, used for training the first basic model based on the second training data to obtain a second model; The second acquisition module is used to obtain the claim interval value and condition interval value corresponding to the insurance claim; A third model training module, used to train the second basic model using the claim interval value, the condition interval value and the historical insurance claim data to obtain a third model; The risk detection module is used to use the first model, the second model and the third model to respectively detect the insurance claim data to be judged, and obtain the risk detection result corresponding to the insurance claim data to be judged based on the first model, the second model and the third model.
6. The device according to claim 5, characterized in that The historical insurance claims data includes abnormal historical insurance claims data, and the device further includes: An abnormal case feature determination module, used to determine abnormal case features in the abnormal historical insurance claims data; A first parameter adjustment module, used to adjust the parameters of the first model based on the abnormal case characteristics to obtain an adjusted first model; The risk detection module is specifically used for: The insurance claim data to be judged are respectively used with the adjusted first model, the second model and the third model, and the risk detection result corresponding to the insurance claim data to be judged is obtained based on the adjusted first model, the second model and the third model.
7. The device according to claim 5, characterized in that The device further comprises: A third acquisition module is used to acquire the insurance claim data to be verified, wherein the insurance claim data to be verified includes a verification tag, and the verification tag is used to indicate whether the corresponding insurance claim data to be verified is abnormal; A verification module, using the first model, the second model and the third model respectively to obtain a verification result corresponding to the insurance claim data to be verified based on the first model, the second model and the third model; The second parameter adjustment module is used to compare the verification result with the label corresponding to the insurance claim data to be verified to obtain a comparison result, and the comparison result is used to indicate the adjustment of the parameters of the first model, the second model and the third model.
8. The device according to claim 5, characterized in that The device further comprises: A cost threshold determination module, used to determine a cost threshold based on one or more of factors such as a target region corresponding to the historical insurance claim data, characteristics of the injured, injury data, and treatment methods; The expense detection module is used to add an expense abnormality risk detection result to the risk detection result if the expense corresponding to the insurance claim data to be judged is higher than the expense threshold.
9. An electronic device, characterized in that: comprising a memory and a processor, wherein: The memory is used to store the computer program; The processor is used to execute the computer program to implement the risk detection method for insurance claim data as described in any one of claims 1 to 4.
10. A computer-readable storage medium, characterized in that: Used to store a computer program, wherein when the computer program is executed by a processor, the risk detection method for insurance claim data according to any one of claims 1 to 4 is implemented.
Citation Information
Cited By
Injury assistance model
US12555167B1
Machine-learning method of loss prediction using incident information
US12694454B2
Guided content capture
US12731350B2
Injury assistance model
US12737819B2
Machine-learning method of generating service provider rankings using incident information
US12793024B2
