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

Abstract

The invention belongs to the technical field of artificial intelligence, and relates to an artificial intelligence-based claim settlement processing method and device, computer equipment and a storage medium, and the method comprises the steps: receiving a claim settlement application submitted by a user through a front-end interface; performing data preprocessing on the multi-dimensional claim settlement data to obtain first claim settlement data; performing feature derivation processing on the first claim settlement data to construct second claim settlement data; calling a preset claim settlement prediction model; performing prediction processing on the second claim settlement data based on a claim settlement prediction model to obtain a corresponding claim settlement prediction result; checking the claim settlement prediction result; and if the claim settlement prediction result passes the audit, performing corresponding claim settlement processing on the claim settlement application based on the claim settlement prediction result. In addition, the claim settlement prediction result can be stored in the block chain. The method can be applied to claim settlement processing scenes in the fields of financial science and technology and medical health insurance, realizes automatic and intelligent processing of the claim settlement process, and effectively improves the claim settlement efficiency and processing accuracy.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology and can be applied to the fields of financial technology and medical health insurance, and in particular to claims processing methods, devices, computer equipment and storage media based on artificial intelligence. Background Art

[0002] In the traditional insurance claims service model, the claims process primarily relies on prescriptive rules for judgment and processing. This approach, when dealing with complex and ever-changing insurance claims, is inefficient and inaccurate. Specifically, the traditional claims process typically reviews claims applications one by one based on pre-set rules and conditions, lacking a deep understanding and intelligent analysis of the complexity and diversity of claims cases. This mechanical approach not only makes it difficult to quickly respond to claims requests, resulting in extended claims processing cycles and a poor customer experience, but is also prone to misjudgments or omissions in complex claims cases, impacting the accuracy and fairness of claims processing.

[0003] For example, in the financial sector, in auto insurance claims, the traditional claims process may simply make judgments based on the extent of vehicle damage and repair costs, without fully considering key factors such as the specific cause of the accident, the division of liability, and the insured's historical claims record. If a car insurance accident involves multiple parties and complex legal disputes, the traditional claims process may make unreasonable claims decisions, such as overpayments or denials, due to its inability to fully assess the case. This not only harms the interests of the insurance company but also affects the protection of the rights and interests of the insured. In addition, when handling a large number of claims, the traditional claims process often requires a large amount of manpower and material resources for review and processing, resulting in high claims costs and further exacerbating the operating pressure on insurance companies.

[0004] Similar issues exist in the healthcare sector, particularly in the medical claims process. Traditional claims processes often focus solely on the invoiced amount of medical expenses and whether basic treatments are covered by the insurance policy. These processes fail to consider critical information such as the actual severity of the patient's condition, the impact of previous medical history on current treatment, and the rationality and necessity of the treatment plan. For example, for patients with complex chronic conditions, treatment may involve a variety of integrated therapies, including specialized examinations, long-term medication, and rehabilitation. Traditional claims processes, failing to fully understand the complex medical logic behind these treatments, can lead to unreasonable claims decisions based solely on superficial expense items and simple clauses. These decisions can include refusing to cover necessary but expensive treatments or incorrectly denying claims for treatments deemed excessive. This not only makes it difficult for patients to obtain the compensation they deserve, hindering their continued care, but also damages the insurance company's credibility with customers and can potentially lead to disputes between doctors, patients, and insurers.

[0005] Therefore, there is an urgent need to provide an insurance claims optimization system based on intelligent algorithms to realize the automation and intelligence of the claims process, improve the efficiency and accuracy of claims, and enhance customers' claims experience and satisfaction. Summary of the Invention

[0006] The purpose of the embodiments of the present application is to propose an artificial intelligence-based claims processing method, apparatus, computer equipment and storage medium to solve the technical problems of low existence rate and low accuracy of the existing insurance claims process.

[0007] First, an AI-based claims processing method is provided, including:

[0008] Receive a claim application submitted by a user through a front-end interface; wherein the claim application carries multi-dimensional claim data, and the multi-dimensional claim data includes at least policyholder information, agent information, and case information;

[0009] Performing data preprocessing on the multi-dimensional claims data to obtain corresponding first claims data;

[0010] Performing feature derivation processing on the first claim data to construct corresponding second claim data;

[0011] Calling a preset claims prediction model; wherein the claims prediction model is a model obtained by training a tuned target lightweight gradient boosting machine model based on pre-constructed sample data;

[0012] Performing prediction processing on the second claim data based on the claim prediction model to obtain a corresponding claim prediction result;

[0013] Reviewing the claims forecast results;

[0014] If the claim prediction result passes the review, the claim application will be processed accordingly based on the claim prediction result.

[0015] Secondly, an artificial intelligence-based claims processing device is provided, comprising:

[0016] A receiving module, configured to receive a claim application submitted by a user through a front-end interface; wherein the claim application carries multi-dimensional claim data, and the multi-dimensional claim data includes at least policyholder information, agent information, and case information;

[0017] A preprocessing module, configured to perform data preprocessing on the multi-dimensional claims data to obtain corresponding first claims data;

[0018] a derivation module, configured to perform feature derivation processing on the first claim data to construct corresponding second claim data;

[0019] A first calling module is used to call a preset claim prediction model; wherein the claim prediction model is a model obtained by training a tuned target lightweight gradient boosting machine model based on pre-constructed sample data;

[0020] A prediction module, configured to perform prediction processing on the second claim data based on the claim prediction model to obtain a corresponding claim prediction result;

[0021] An audit module, used for auditing the claim prediction result;

[0022] A processing module is used to perform corresponding claim processing on the claim application based on the claim prediction result if the claim prediction result passes the review.

[0023] In a third aspect, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above-mentioned artificial intelligence-based claims processing method are implemented.

[0024] In a fourth aspect, a computer-readable storage medium is provided, which stores a computer program. When the computer program is executed by a processor, the steps of the above-mentioned artificial intelligence-based claims processing method are implemented.

[0025] In the solution implemented by the above-mentioned artificial intelligence-based claims processing method, device, computer equipment and storage medium, a claim application submitted by a user through a front-end interface is first received; wherein, the claim application carries multi-dimensional claims data, and the multi-dimensional claims data includes at least insurance information, agent information and case information; then the multi-dimensional claims data is preprocessed to obtain corresponding first claims data; and feature derivative processing is performed on the first claims data to construct corresponding second claims data; then a preset claims prediction model is called; wherein, the claims prediction model is a model obtained by training a tuned target lightweight gradient boosting machine model based on pre-constructed sample data; subsequently, the second claims data is predicted based on the claims prediction model to obtain a corresponding claims prediction result; the claims prediction result is further reviewed; if the claims prediction result passes the review, the claims application is processed accordingly based on the claims prediction result. After receiving a claim application submitted by a user through the front-end interface, this application obtains second claim data by performing data preprocessing and feature derivation processing on the multi-dimensional claim data. It then performs prediction processing on the second claim data based on the invoked claim prediction model to obtain a claim prediction result. When the claim prediction result passes review, the application is processed accordingly based on the claim prediction result. In this way, by processing claims based on the use of the claim prediction model, the claims process is automated and intelligent, effectively improving claims efficiency and processing accuracy, and contributing to increased customer satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the solutions in this application, a brief introduction will be given below to the drawings required for use in the description of the embodiments of this application. Obviously, the drawings described below are some embodiments of this application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0027] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0028] Figure 2 is a flow chart of an embodiment of an artificial intelligence-based claims processing method according to the present application;

[0029] Figure 3 is a structural diagram of an embodiment of an artificial intelligence-based claims processing device according to the present application;

[0030] Figure 4 It is a structural diagram of an embodiment of a computer device according to the present application. DETAILED DESCRIPTION

[0031] Unless otherwise defined, all technical and scientific terms used herein have the same meanings as commonly understood by those skilled in the art to which this application belongs. The terms used in the specification of the application are for the purpose of describing specific embodiments only and are not intended to limit this application. The terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of this application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0032] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0033] 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.

[0034] like Figure 1 As shown, system architecture 100 may include a terminal device 101, a network 102, and a server 103. Terminal device 101 may be a laptop computer 1011, a tablet computer 1012, or a mobile phone 1013. Network 102 is a medium for providing a communication link between terminal device 101 and server 103. Network 102 may include various connection types, such as wired or wireless communication links or fiber optic cables.

[0035] The user can use the terminal device 101 to interact with the server 103 via 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.

[0036] The terminal device 101 can be various electronic devices with a display screen and supporting web browsing. In addition to the laptop computer 1011, tablet computer 1012 or 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.

[0037] The server 103 may be a server that provides various services, such as a background server that provides support for web pages displayed on the terminal device 101 .

[0038] It should be noted that the AI-based claims processing method provided in the embodiments of the present application is generally executed by a server / terminal device, and accordingly, the AI-based claims processing device is generally set in the server / terminal device.

[0039] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is merely illustrative. Any number of terminal devices, networks and servers may be provided as required.

[0040] Continue to refer Figure 2 , shows a flowchart of an embodiment of the claims processing method based on artificial intelligence according to the present application. According to different needs, the order of the steps in the flowchart can be changed, and some steps can be omitted. The claims processing method based on artificial intelligence provided by the embodiment of the present application can be applied to any scenario that requires claims processing, and the claims processing method based on artificial intelligence can be applied to products in these scenarios, for example, claims processing in the field of finance and insurance. The claims processing method based on artificial intelligence includes the following steps:

[0041] Step S201, receiving a claim application submitted by a user through a front-end interface; wherein the claim application carries multi-dimensional claim data, and the multi-dimensional claim data includes at least insurance information, agent information, and case information.

[0042] In this embodiment, the claim processing method based on artificial intelligence is run on the electronic device (e.g. Figure 1The server / terminal device shown in the figure) can obtain claim applications submitted by users through the front-end interface through a wired connection or a wireless connection. It should be noted that the above-mentioned wireless connection methods may include but are not limited to 3G / 4G / 5G connections, WiFi connections, Bluetooth connections, WiMAX connections, Zigbee connections, UWB (Ultra Wide Band) connections, and other wireless connection methods currently known or developed in the future. The execution entity of this application is specifically a claims processing system, which can be simply referred to as the system. This application can be applied to insurance claims processes in the financial field. The above-mentioned front-end interface refers to the interface used by the system to interact with users. Users can use the front-end interface to assist in entering claim applications. The entered information includes multi-dimensional claims data, which includes at least insurance information, agent information, and case information. Specifically, insurance information refers to data related to the policyholder / insured dimension, typically including the policyholder / insured's basic information (age, gender, occupation, etc.), insurance history, health status, etc. These factors can be used to predict the claims risk of the policyholder / insured group. Agent information refers to data related to the agent dimension. Agent information can reveal the agent's integrity and risk control capabilities. Characteristics include the agent's sales record, claims settlement status, etc. Case information refers to data related to the case dimension, including the case processing process (such as submission time, processing time, application amount, etc.).

[0043] In the financial sector, for example, the above-mentioned claim application could be a car insurance claim received by an insurance company. The claim application involves a collision between a car and another car while the car was in motion, with a claim amount of 30,000 yuan. The car owner submitted the claim application through an online claims platform. The multi-dimensional claim data included: basic information (the car owner is 40 years old, has 15 years of driving experience, and is a self-employed business owner); the details of the accident (the car owner was driving straight at an intersection when he was struck by a vehicle on the right that ran a red light); vehicle damage (the front bumper fell off, the left door was dented and deformed, and the headlight was damaged); a detailed repair cost breakdown (8,000 yuan for front bumper replacement, 5,000 yuan for door repair, 12,000 yuan for headlight replacement, and 5,000 yuan for other parts and labor); the accident liability determination (the traffic police determined that the other vehicle was fully responsible, but the other vehicle owner disputed the determination and has not yet signed a liability determination); and the investigator in charge of the case.

[0044] In the field of medical and health insurance, for example, the above-mentioned claim application may be a medical and health insurance claim application received by an insurance company, in which a customer was hospitalized for a sudden heart attack and applied for a claim amount of 80,000 yuan. The customer submitted a claim application through the insurance company's APP. The multi-dimensional claim data contained in the claim application include: basic information (customer age 55, history of hypertension, retired), disease condition (sudden acute myocardial infarction, stable condition after emergency coronary intervention), medical records (10 days of hospitalization, total medical expenses of 80,000 yuan, including 30,000 yuan for surgery, 25,000 yuan for medicines, 15,000 yuan for examinations, and 10,000 yuan for bed fees and other expenses), past medical history (hospitalized for hypertension 5 years ago, long-term use of antihypertensive drugs), and the claims specialist in charge of the case.

[0045] Step S202: pre-process the multi-dimensional claim data to obtain corresponding first claim data.

[0046] In this embodiment, the above-mentioned specific implementation process of preprocessing the multi-dimensional claims data to obtain the corresponding first claims data will be further described in detail in subsequent specific embodiments of this application and will not be elaborated on here.

[0047] Step S203: Perform feature derivation processing on the first claim data to construct corresponding second claim data.

[0048] In this embodiment, the above-mentioned feature derivation processing of the first claim data to construct the specific implementation process of obtaining the corresponding second claim data will be further described in detail in subsequent specific embodiments of this application and will not be elaborated on here.

[0049] Step S204, calling a preset claim prediction model; wherein the claim prediction model is a model obtained by training a tuned target lightweight gradient boosting machine model based on pre-constructed sample data.

[0050] In this embodiment, the above-mentioned claim prediction model is a model obtained by training a tuned target lightweight gradient boosting machine model based on pre-constructed sample data. The specific construction process of the claim prediction model will be further described in detail in subsequent specific embodiments of this application and will not be elaborated on here.

[0051] Step S205: performing prediction processing on the second claim data based on the claim prediction model to obtain a corresponding claim prediction result.

[0052] In this embodiment, the generated second claim data is input into a trained claim prediction model. The model then performs a claim prediction on the second claim data and outputs corresponding claim prediction results, such as data including the claim approval probability and the predicted claim amount. For example, the outputted claim approval probability is 98%, and the predicted claim amount is 50,000 yuan.

[0053] Step S206: review the claim prediction result.

[0054] In this embodiment, the specific implementation process of reviewing the claim prediction results will be described in further detail in subsequent specific embodiments of this application and will not be elaborated on here.

[0055] Step S207: If the claim prediction result passes the review, the claim application is processed accordingly based on the claim prediction result.

[0056] In this embodiment, if it is detected that the above-mentioned claim prediction result has passed the review, it is determined that the claim application does not have the risk of fraud and meets the conditions for quick claim settlement. Then, the system will automatically complete the claim settlement process based on the predicted claim amount contained in the above-mentioned claim prediction result. For example, the customer receives a compensation amount of 50,000 yuan within 24 hours after submitting the claim application.

[0057] This application first receives a claim application submitted by a user through a front-end interface; wherein, the claim application carries multi-dimensional claim data, and the multi-dimensional claim data includes at least insurance information, agent information and case information; then the multi-dimensional claim data is pre-processed to obtain corresponding first claim data; and the first claim data is subjected to feature derivation processing to construct corresponding second claim data; then a preset claim prediction model is called; wherein, the claim prediction model is a model obtained by training a tuned target lightweight gradient boosting machine model based on pre-constructed sample data; subsequently, the second claim data is predicted based on the claim prediction model to obtain a corresponding claim prediction result; the claim prediction result is further reviewed; if the claim prediction result passes the review, the claim application is processed accordingly based on the claim prediction result. After receiving a claim application submitted by a user through the front-end interface, this application obtains second claim data by performing data preprocessing and feature derivation processing on the multi-dimensional claim data. It then performs prediction processing on the second claim data based on the invoked claim prediction model to obtain a claim prediction result. When the claim prediction result passes review, the application is processed accordingly based on the claim prediction result. In this way, by processing claims based on the use of the claim prediction model, the claims process is automated and intelligent, effectively improving claims efficiency and processing accuracy, and contributing to increased customer satisfaction.

[0058] In some optional implementations, before step S204, the electronic device may further perform the following steps:

[0059] Call the pre-built lightweight gradient boosting machine model.

[0060] In this embodiment, the above-mentioned lightweight gradient boosting machine model is the LightGBM model. The LightGBM model has the following advantages: 1) Histogram-based decision tree splitting: LightGBM discretizes continuous features into histograms, which greatly reduces the amount of calculation, enabling it to perform well when processing large-scale claims data and shortening the training time. 2) Leaf node priority growth strategy: LightGBM prioritizes leaf nodes with the largest gain for expansion, which can better capture complex risk factor relationships, improve prediction accuracy, and accelerate convergence. 3) Automatic feature importance evaluation: Among multi-dimensional risk features, LightGBM can automatically evaluate and screen out features that contribute more to the model, help optimize the model structure, and also assist in identifying core risk factors.

[0061] Get the preset hyperparameter optimization strategy and early stopping strategy.

[0062] In this embodiment, the hyperparameter optimization strategy includes using grid search or Bayesian optimization to tune the hyperparameters of LightGBM, including the learning rate (learning_rate), maximum depth (max_depth), number of leaf nodes (num_leaves), regularization parameters (lambda_1, lambda_1 2), etc. The early stopping strategy includes setting the number of early stopping rounds and terminating training early when the performance on the validation set no longer improves, thereby preventing overfitting and improving the generalization ability of the model.

[0063] The lightweight gradient boosting machine model is tuned based on the hyperparameter optimization strategy and the early stopping strategy to obtain a corresponding tuned model.

[0064] In this embodiment, the lightweight gradient boosting machine model can be tuned according to the strategy content of the above-mentioned hyperparameter optimization strategy and the strategy content of the above-mentioned early stopping strategy, thereby obtaining a tuned model, that is, the above-mentioned tuned model.

[0065] The tuned model is used as the target lightweight gradient boosting machine model.

[0066] In this embodiment, for the generated target lightweight gradient boosting machine model, built-in category feature support can be further used for category features to avoid the dimensionality explosion problem caused by one-hot encoding.

[0067] This application calls a pre-built lightweight gradient boosting machine model; then obtains a preset hyperparameter optimization strategy and early stopping strategy; then tunes the lightweight gradient boosting machine model based on the hyperparameter optimization strategy and the early stopping strategy to obtain a corresponding tuned model; and subsequently uses the tuned model as the target lightweight gradient boosting machine model. This application calls a pre-built lightweight gradient boosting machine model, and then tunes the lightweight gradient boosting machine model based on the use of the obtained hyperparameter optimization strategy and early stopping strategy, thereby achieving intelligent completion of the optimization processing of the lightweight gradient boosting machine model, improving the model effect of the target lightweight gradient boosting machine model obtained, and subsequently training the tuned target lightweight gradient boosting machine model based on the pre-built sample data to obtain a claims prediction model, which can effectively improve the prediction ability of the generated claims prediction model.

[0068] In some optional implementations of this embodiment, before step S204, the electronic device may further perform the following steps:

[0069] Obtain pre-collected historical multi-dimensional claims data.

[0070] In this embodiment, the historical multi-dimensional claims data includes pre-collected historical policyholder information, historical agent information, and historical case information within a historical time period. The value of the historical time period is not specifically limited; for example, it can be within the past two years. Upon collecting the historical multi-dimensional claims data, missing value processing and outlier processing can be performed to complete data cleaning of the historical multi-dimensional claims data.

[0071] Feature engineering is performed on the historical multi-dimensional claims data to construct corresponding sample data, and the sample data is divided into a training set and a validation set.

[0072] In this embodiment, the above-mentioned feature engineering includes feature standardization and encoding processing, feature interaction and derivation processing, and feature selection processing. Specifically, feature standardization and encoding processing includes: standardizing or normalizing continuous features, and performing one-hot encoding or label encoding on categorical features. Feature interaction and derivation processing includes: mining the interactions between features, and deriving new features (such as historical claims frequency, average claim amount, etc.). Feature selection processing includes: using the built-in feature importance evaluation function of the claims prediction model to screen out features that contribute more to the model and reduce redundant features. In addition, stratified sampling can be used to divide the above-mentioned sample data to obtain training sets and validation sets to ensure that the distribution of target variables in each data set is consistent.

[0073] Call the target lightweight gradient boosting machine model.

[0074] In this embodiment, the construction process of the above-mentioned target lightweight gradient boosting machine model can refer to the aforementioned construction process flow, which will not be repeated here.

[0075] The target lightweight gradient boosting machine model is trained based on the training set to obtain a trained first model.

[0076] In this embodiment, the target lightweight gradient boosting machine model can be trained using the above training set, and the model performance can be improved through early stopping strategy and hyperparameter tuning to obtain a trained first model.

[0077] Performing a performance evaluation on the first model based on the validation set to obtain a performance evaluation result, and optimizing the first model based on the performance evaluation result to obtain a second model that meets the performance requirements;

[0078] In this embodiment, the performance of the first model can be evaluated on the validation set using the AUC and Log Loss metrics, and corresponding performance evaluation results can be obtained. AUC (Area Under Curve) measures the model's ability to distinguish between positive and negative samples, and the closer the value is to 1, the better. Logloss (Logarithmic Loss) measures the accuracy of the model's predicted probability, and the smaller the value, the better. By analyzing the obtained performance evaluation results, if the validation set performance is poor, that is, it does not meet the performance requirements, the training process is returned to adjust the hyperparameters or feature engineering to perform model adjustments until a second model that meets the performance requirements and has good generalization ability is obtained.

[0079] The second model is used as the claim prediction model.

[0080] This application obtains pre-collected historical multi-dimensional claims data; then performs feature engineering on the historical multi-dimensional claims data to construct corresponding sample data, and divides the sample data into a training set and a validation set; then calls the target lightweight gradient boosting machine model; and trains the target lightweight gradient boosting machine model based on the training set to obtain a trained first model; subsequently, a performance evaluation is performed on the first model based on the validation set to obtain a performance evaluation result, and the first model is optimized based on the performance evaluation result to obtain a second model that meets the performance requirements; finally, the second model is used as the claims prediction model. This application constructs sample data by performing feature engineering on pre-collected historical multi-dimensional claims data, and divides the sample data into a training set and a validation set. Then, based on the use of the training set, the target lightweight gradient boosting machine model is trained to obtain a trained first model. Then, based on the use of the validation set, the first model is performance evaluated to obtain a performance evaluation result. Based on the use of the performance evaluation result, the first model is optimized to obtain a claims prediction model, thereby achieving efficient and accurate completion of the model construction of the claims prediction model, improving the construction efficiency of the claims prediction model, and ensuring the predictive ability of the generated claims prediction model.

[0081] In some optional implementations, after the step of using the second model as the claim prediction model, the electronic device may further perform the following steps:

[0082] Regularly collect newly generated specified claims data from the production environment.

[0083] In this embodiment, the multi-dimensional claims data used in the training of the claims prediction model can be obtained as the benchmark data (reference distribution), and newly generated specified claims data (distribution to be compared) in the production environment can be regularly collected according to the preset timed tasks. The above-mentioned timed tasks can be set according to actual business needs.

[0084] Call the preset data drift detection tool.

[0085] In this embodiment, the data drift detection tool is an automated tool that has the function of performing data drift detection, for example, a KS test (Kolmogorov-Smi r nov Test) tool may be used.

[0086] The data drift detection tool is used to calculate the distribution difference between the designated claim data and the historical multi-dimensional claim data to obtain a corresponding distribution difference result.

[0087] In this embodiment, for each continuous feature contained in the above-mentioned designated claims data and historical multidimensional claims data, the cumulative distribution function (CDF) of the benchmark data (designated claims data) and the new data (historical multidimensional claims data) are calculated respectively, and then the maximum difference between the two CDFs, that is, the KS statistic, is calculated, so as to subsequently determine whether the distribution is significantly different based on the KS statistic.

[0088] Determine whether the distribution difference result is greater than a preset threshold.

[0089] In this embodiment, there is no specific limitation on the selection of the above-mentioned preset threshold, which can be set according to actual business needs, for example, it can be set to 0.1 or 0.2.

[0090] If so, the claim prediction model is adjusted based on a preset adjustment strategy.

[0091] In this embodiment, if the KS statistic is detected to exceed a threshold, it indicates a significant difference between the two distributions, possibly indicating data drift. The aforementioned adjustment strategy may include retraining the model with new data or updating model parameters using an online learning algorithm. Specifically, when data drift is detected, timely model adjustments are required. Specifically, adjustments to the claims model can be made based on the aforementioned adjustment strategy to enable the model to adapt to changes in data distribution and business rules, thereby ensuring model accuracy and stability.

[0092] This application regularly collects newly generated designated claims data in the production environment; then calls a preset data drift detection tool; and based on the data drift detection tool, calculates and processes the distribution differences between the designated claims data and the historical multi-dimensional claims data to obtain corresponding distribution difference results; subsequently determines whether the distribution difference result is greater than a preset threshold; if so, performs model adjustment processing on the claims prediction model based on a preset adjustment strategy. This application regularly collects newly generated designated claims data in the production environment; then, based on the use of a data drift detection tool, calculates and processes the distribution differences between the designated claims data and the historical multi-dimensional claims data to obtain corresponding distribution difference results, and when it is detected that the distribution difference result is greater than a preset threshold, it automatically and intelligently performs model adjustment processing on the claims prediction model based on the use of a preset adjustment strategy, so that the claims prediction model can adapt to changes in data distribution and business rules, thereby ensuring the accuracy and stability of the claims prediction model.

[0093] In some optional implementations, step S202 includes the following steps:

[0094] The multi-dimensional claims data is cleaned to obtain corresponding first processed data.

[0095] In this embodiment, the data cleaning process includes missing value processing and outlier processing. Missing value processing includes filling missing values ​​(e.g., using the mean, median, or mode) or deleting severely missing data. Outlier processing includes identifying and handling outliers, such as using the IQR (interquartile range) method or business rule-based corrections.

[0096] The first processed data is standardized to obtain corresponding second processed data.

[0097] In this embodiment, all continuous features contained in the first processed data are extracted and then normalized to obtain the corresponding second processed data, and the categorical features are one-hot encoded or label encoded.

[0098] The second processed data is encoded to obtain corresponding third processed data.

[0099] In this embodiment, corresponding third processed data can be obtained by extracting all classification features contained in the second processed data and then performing one-hot encoding or label encoding on the classification features.

[0100] The third processed data is used as the first claim settlement data.

[0101] This application performs data cleaning on the multi-dimensional claims data to obtain corresponding first processed data; then performs standardization on the first processed data to obtain corresponding second processed data; then performs encoding on the second processed data to obtain corresponding third processed data; and subsequently uses the third processed data as the first claims data. This application performs data cleaning, standardization, and encoding on the multi-dimensional claims data, thereby achieving efficient and accurate pre-processing of the multi-dimensional claims data, effectively improving the accuracy of the generated first claims data and ensuring the standardization of the data subsequently input into the claims prediction model.

[0102] In some optional implementations of this embodiment, step S203 includes the following steps:

[0103] Obtain interaction information between features of the first claim data.

[0104] In this embodiment, feature interaction can be performed on the first claim data to mine the interactions (interaction information) between the features contained in the first claim data, such as the combined features of the insured's age and occupation, or the difference features between the case submission time and the processing time.

[0105] Based on the interactive information, corresponding feature derivation processing is performed on the first claim data to obtain derived feature data.

[0106] In this embodiment, new features may be derived based on the obtained interactive information to obtain corresponding derived feature data, which may include, for example, the insured's historical claim frequency, the agent's average claim amount, etc.

[0107] The first claim settlement data and the derived characteristic data are combined to obtain corresponding combined data.

[0108] In this embodiment, the combined data is a data set including the first claim data and the derived feature data.

[0109] The combined data is used as the second claim settlement data.

[0110] This application obtains the interactive information between the features of the first claim data; then performs corresponding feature derivation processing on the first claim data based on the interactive information to obtain derived feature data; then combines the first claim data and the derived feature data to obtain corresponding combined data; and subsequently uses the combined data as the second claim data. This application obtains the interactive information between the features of the first claim data; then performs corresponding feature derivation processing on the first claim data based on the use of the interactive information to obtain derived feature data, and then combines the first claim data and the derived feature data, thereby achieving intelligent and accurate feature derivation processing for the first claim data, improving the data diversity of the generated second claim data, and then subsequently using the claim prediction model to predict the second claim data, which can effectively improve the prediction ability of the claim prediction model and improve the accuracy of the generated claim prediction results.

[0111] In some optional implementations of this embodiment, step S207 includes the following steps:

[0112] Get the preset anomaly detection algorithm.

[0113] In this embodiment, the anomaly detection algorithm is a model algorithm capable of identifying potential fraud in claims data. For example, the anomaly detection algorithm can technically detect unusual delays or high-frequency claims processing through time series analysis, thereby identifying potential anomalies or fraudulent behavior.

[0114] Fraud detection is performed on the multi-dimensional claims data based on the anomaly detection algorithm to obtain corresponding fraud detection results.

[0115] In this embodiment, fraud detection can be performed on the multi-dimensional claims data using the aforementioned anomaly detection algorithm to identify whether abnormal behavior exists in the multi-dimensional claims data and generate corresponding fraud detection results. If abnormal behavior is identified in the multi-dimensional claims data, a fraud detection result is generated indicating that the multi-dimensional claims data presents a fraud risk; if no abnormal behavior is identified in the multi-dimensional claims data, a fraud detection result is generated indicating that the multi-dimensional claims data does not present a fraud risk.

[0116] If the fraud detection result shows that there is no fraud risk in the multi-dimensional claims data, the claims prediction result is determined to have passed the review.

[0117] In this embodiment, if the fraud detection result is that there is no fraud risk in the multi-dimensional claims data, it indicates that the generated claims prediction result has passed the review.

[0118] If the fraud detection result indicates that there is a fraud risk in the multi-dimensional claims data, it is determined that the claims prediction result has failed the review.

[0119] In this embodiment, if the fraud detection result is that the multi-dimensional claims data has a fraud risk, it indicates that the generated claims prediction result has not passed the review.

[0120] In addition, the generated claims prediction results can also be fed back to relevant staff through the front-end interface. The staff can review the effectiveness of the claims prediction model and make targeted adjustments to the claims prediction model to make the claims prediction model more accurate.

[0121] This application obtains a preset anomaly detection algorithm; then performs fraud detection on the multi-dimensional claims data based on the anomaly detection algorithm to obtain a corresponding fraud detection result; if the fraud detection result is that the multi-dimensional claims data does not have a fraud risk, then the claim prediction result is determined to have passed the review; and if the fraud detection result is that the multi-dimensional claims data does have a fraud risk, then the claim prediction result is determined to have failed the review. This application performs fraud detection on multi-dimensional claims data based on the use of an anomaly detection algorithm to obtain a corresponding fraud detection result, and then performs content analysis on the fraud detection result, so as to automatically and accurately complete the review processing of the claims prediction result, improve the review efficiency of the claims prediction result, and ensure the accuracy of the obtained review result.

[0122] In some optional implementations, the user information obtained is obtained with the user's consent and complies with relevant laws and policies.

[0123] In addition, any software tools or components not provided by our company that appear in the embodiments of this application are merely examples and do not represent actual use.

[0124] Furthermore, by building a claims prediction model, this application can rapidly respond to claims and significantly increase the proportion of instant claims. While controlling risk, the claims prediction model can promptly determine whether a case qualifies for expedited claims settlement, thereby accelerating the claims process and significantly improving claims efficiency. This allows customers to receive claims faster, which not only reduces wait times but also significantly increases customer satisfaction.

[0125] Furthermore, business needs, market conditions, and data distributions change over time. Claims prediction models can adapt to these changes, undergoing regular tuning and updates to address new data distributions and risk factors. For example, new user behavior characteristics, case types, or external data sources can be quickly incorporated into the claims prediction model to improve prediction accuracy.

[0126] It should be understood that the order of execution of the steps in the above embodiments does not necessarily mean the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0127] It should be emphasized that in order to further ensure the privacy and security of the above-mentioned claim prediction results, the above-mentioned claim prediction results can also be stored in a node of a blockchain.

[0128] The blockchain referred to in this application refers to a new application model for computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms. Blockchain is essentially a decentralized database, a series of data blocks generated using cryptographic methods. Each data block contains information about a batch of network transactions, which is used to verify the validity of the information (to prevent counterfeiting) and generate the next block. Blockchain can include the blockchain underlying platform, the platform product service layer, and the application service layer.

[0129] 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 achieve optimal results.

[0130] Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interaction systems, and mechatronics. AI software technologies primarily encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0131] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing related hardware via computer-readable instructions. The computer-readable instructions can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes in the above-described method embodiments. The aforementioned storage medium can be a non-volatile storage medium such as a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0132] It should be understood that although the steps in the flowcharts of the accompanying drawings are shown in sequence as indicated by the arrows, these steps are not necessarily executed 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 some of the steps in the flowcharts 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 portion of the sub-steps or stages of other steps.

[0133] Further references Figure 3 , as a response to the above Figure 2 The present application provides an embodiment of a claims processing device based on artificial intelligence, which is similar to Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0134] like Figure 3 As shown, the AI-based claims processing device 300 of this embodiment includes: a receiving module 301, a pre-processing module 302, a derivation module 303, a first calling module 304, a prediction module 305, an audit module 306, and a processing module 307. Among them:

[0135] Receiving module 301, for receiving a claim application submitted by a user through a front-end interface; wherein the claim application carries multi-dimensional claim data, and the multi-dimensional claim data includes at least policyholder information, agent information, and case information;

[0136] A preprocessing module 302 is configured to perform data preprocessing on the multi-dimensional claims data to obtain corresponding first claims data;

[0137] A derivation module 303 is configured to perform feature derivation processing on the first claim data to construct corresponding second claim data;

[0138] A first calling module 304 is configured to call a preset claim prediction model; wherein the claim prediction model is a model obtained by training a tuned target lightweight gradient boosting machine model based on pre-built sample data;

[0139] Prediction module 305, configured to perform prediction processing on the second claim settlement data based on the claim settlement prediction model to obtain a corresponding claim settlement prediction result;

[0140] An audit module 306 is used to audit the claim prediction result;

[0141] The processing module 307 is used to perform corresponding claim processing on the claim application based on the claim prediction result if the claim prediction result passes the review.

[0142] In this embodiment, the operations performed by the above modules or units correspond one-to-one to the steps of the artificial intelligence-based claims processing method in the aforementioned embodiment, and will not be repeated here.

[0143] In some optional implementations of this embodiment, the artificial intelligence-based claims processing device further includes:

[0144] The second calling module is used to call the pre-built lightweight gradient boosting machine model;

[0145] The first acquisition module is used to obtain the preset hyperparameter optimization strategy and early stopping strategy;

[0146] A tuning module, configured to perform tuning processing on the lightweight gradient boosting machine model based on the hyperparameter optimization strategy and the early stopping strategy to obtain a corresponding tuned model;

[0147] A first determining module is configured to use the tuned model as the target lightweight gradient boosting machine model.

[0148] In this embodiment, the operations performed by the above modules or units correspond one-to-one to the steps of the artificial intelligence-based claims processing method in the aforementioned embodiment, and will not be repeated here.

[0149] In some optional implementations of this embodiment, the artificial intelligence-based claims processing device further includes:

[0150] The second acquisition module is used to obtain pre-collected historical multi-dimensional claims data;

[0151] A construction module is used to perform feature engineering on the historical multi-dimensional claims data to construct corresponding sample data, and divide the sample data into a training set and a validation set;

[0152] A third calling module is used to call the target lightweight gradient boosting machine model;

[0153] A training module, configured to train the target lightweight gradient boosting machine model based on the training set to obtain a trained first model;

[0154] an optimization module, configured to perform a performance evaluation on the first model based on the validation set to obtain a performance evaluation result, and to optimize the first model based on the performance evaluation result to obtain a second model that meets the performance requirements;

[0155] The second determining module is used to use the second model as the claim prediction model.

[0156] In this embodiment, the operations performed by the above modules or units correspond one-to-one to the steps of the artificial intelligence-based claims processing method in the aforementioned embodiment, and will not be repeated here.

[0157] In some optional implementations of this embodiment, the artificial intelligence-based claims processing device further includes:

[0158] The collection module is used to regularly collect newly generated specified claims data in the production environment;

[0159] The fourth calling module is used to call a preset data drift detection tool;

[0160] a calculation module, configured to calculate the distribution difference between the specified claim data and the historical multi-dimensional claim data based on the data drift detection tool, and obtain a corresponding distribution difference result;

[0161] A judgment module, configured to judge whether the distribution difference result is greater than a preset threshold;

[0162] The adjustment module is used to, if yes, perform model adjustment processing on the claim prediction model based on a preset adjustment strategy.

[0163] In this embodiment, the operations performed by the above modules or units correspond one-to-one to the steps of the artificial intelligence-based claims processing method in the aforementioned embodiment, and will not be repeated here.

[0164] In some optional implementations of this embodiment, the preprocessing module 302 includes:

[0165] A first processing submodule is configured to perform data cleaning on the multi-dimensional claims data to obtain corresponding first processed data;

[0166] a second processing submodule, configured to perform standardization processing on the first processed data to obtain corresponding second processed data;

[0167] a third processing submodule, configured to perform encoding processing on the second processed data to obtain corresponding third processed data;

[0168] The first determining submodule is configured to use the third processed data as the first claim settlement data.

[0169] In this embodiment, the operations performed by the above modules or units correspond one-to-one to the steps of the artificial intelligence-based claims processing method in the aforementioned embodiment, and will not be repeated here.

[0170] In some optional implementations of this embodiment, the derivation module 303 includes:

[0171] A first acquisition submodule, configured to acquire interaction information between features of the first claim data;

[0172] a derivation submodule, configured to perform corresponding feature derivation processing on the first claim data based on the interaction information to obtain derived feature data;

[0173] a combining submodule, configured to combine the first claim settlement data and the derived characteristic data to obtain corresponding combined data;

[0174] The second determining submodule is configured to use the combined data as the second claim settlement data.

[0175] In this embodiment, the operations performed by the above modules or units correspond one-to-one to the steps of the artificial intelligence-based claims processing method in the aforementioned embodiment, and will not be repeated here.

[0176] In some optional implementations of this embodiment, the processing module 307 includes:

[0177] The second acquisition submodule is used to obtain a preset anomaly detection algorithm;

[0178] a detection submodule, configured to perform fraud detection on the multi-dimensional claims data based on the anomaly detection algorithm to obtain corresponding fraud detection results;

[0179] A first determination submodule is configured to determine that the claim prediction result has passed the review if the fraud detection result indicates that the multi-dimensional claim data does not have a fraud risk;

[0180] The second determination submodule is configured to determine that the claim prediction result has failed the review if the fraud detection result indicates that the multi-dimensional claim data has a fraud risk.

[0181] In this embodiment, the operations performed by the above modules or units correspond one-to-one to the steps of the artificial intelligence-based claims processing method in the aforementioned embodiment, and will not be repeated here.

[0182] 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.

[0183] The computer device 4 includes a memory 41, a processor 42, and a network interface 43 that are interconnected through a system bus. It should be noted that the figure only shows a computer device 4 having components 41-43, 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 microprocessors, application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0184] The computer device may be a desktop computer, notebook computer, PDA, cloud server, etc. The computer device may interact with the user via a keyboard, mouse, remote control, touchpad, or voice control device.

[0185] The memory 41 includes at least one type of readable storage medium, including flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 41 can be an internal storage unit of the computer device 4, such as the hard disk or memory of the computer device 4. In other embodiments, the memory 41 can also be an external storage device of the computer device 4, such as a plug-in hard disk equipped on the computer device 4, a smart media card (SMC), a secure digital (SD) card, a flash memory card, etc. 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 for the artificial intelligence-based claims processing method. 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.

[0186] In some embodiments, the processor 42 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip. 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 execute computer-readable instructions or process data stored in the memory 41, such as computer-readable instructions for executing the artificial intelligence-based claims processing method.

[0187] 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.

[0188] Compared with the prior art, the embodiments of the present application have the following beneficial effects:

[0189] In an embodiment of the present application, a claim application submitted by a user through a front-end interface is first received; wherein, the claim application carries multi-dimensional claim data, and the multi-dimensional claim data includes at least insurance information, agent information and case information; then, data preprocessing is performed on the multi-dimensional claim data to obtain corresponding first claim data; and feature derivation processing is performed on the first claim data to construct corresponding second claim data; then a preset claim prediction model is called; wherein, the claim prediction model is a model obtained by training a tuned target lightweight gradient boosting machine model based on pre-constructed sample data; subsequently, the second claim data is predicted based on the claim prediction model to obtain a corresponding claim prediction result; the claim prediction result is further reviewed; if the claim prediction result passes the review, corresponding claim processing is performed on the claim application based on the claim prediction result. After receiving a claim application submitted by a user through the front-end interface, this application obtains second claim data by performing data preprocessing and feature derivation processing on the multi-dimensional claim data. It then performs prediction processing on the second claim data based on the invoked claim prediction model to obtain a claim prediction result. When the claim prediction result passes review, the application is processed accordingly based on the claim prediction result. In this way, by processing claims based on the use of the claim prediction model, the claims process is automated and intelligent, effectively improving claims efficiency and processing accuracy, and contributing to increased customer satisfaction.

[0190] The present application also provides another embodiment, namely, providing a computer-readable storage medium, which 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 above-mentioned artificial intelligence-based claims processing method.

[0191] Compared with the prior art, the embodiments of the present application have the following beneficial effects:

[0192] In an embodiment of the present application, a claim application submitted by a user through a front-end interface is first received; wherein, the claim application carries multi-dimensional claim data, and the multi-dimensional claim data includes at least insurance information, agent information and case information; then, data preprocessing is performed on the multi-dimensional claim data to obtain corresponding first claim data; and feature derivation processing is performed on the first claim data to construct corresponding second claim data; then a preset claim prediction model is called; wherein, the claim prediction model is a model obtained by training a tuned target lightweight gradient boosting machine model based on pre-constructed sample data; subsequently, the second claim data is predicted based on the claim prediction model to obtain a corresponding claim prediction result; the claim prediction result is further reviewed; if the claim prediction result passes the review, corresponding claim processing is performed on the claim application based on the claim prediction result. After receiving a claim application submitted by a user through the front-end interface, this application obtains second claim data by performing data preprocessing and feature derivation processing on the multi-dimensional claim data. It then performs prediction processing on the second claim data based on the invoked claim prediction model to obtain a claim prediction result. When the claim prediction result passes review, the application is processed accordingly based on the claim prediction result. In this way, by processing claims based on the use of the claim prediction model, the claims process is automated and intelligent, effectively improving claims efficiency and processing accuracy, and contributing to increased customer satisfaction.

[0193] 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 the necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of 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 enabling 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.

[0194] Obviously, the embodiments described above are only some of the embodiments of the present application, rather than all of the 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 has been described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions described in the aforementioned specific embodiments, or to make equivalent replacements for some of the technical features therein. Any equivalent structure made using the contents of the present application specification and the accompanying drawings, directly or indirectly used in other related technical fields, is also within the scope of patent protection of the present application.

Claims

1. A claims processing method based on artificial intelligence, characterized in that: The steps include: Receive a claim application submitted by a user through a front-end interface; wherein the claim application carries multi-dimensional claim data, and the multi-dimensional claim data includes at least policyholder information, agent information, and case information; Performing data preprocessing on the multi-dimensional claims data to obtain corresponding first claims data; Performing feature derivation processing on the first claim data to construct corresponding second claim data; Calling a preset claims prediction model; wherein the claims prediction model is a model obtained by training a tuned target lightweight gradient boosting machine model based on pre-constructed sample data; Performing prediction processing on the second claim data based on the claim prediction model to obtain a corresponding claim prediction result; Reviewing the claims forecast results; If the claim prediction result passes the review, the claim application will be processed accordingly based on the claim prediction result.

2. The claim processing method based on artificial intelligence according to claim 1, characterized in that: Before the step of calling the preset claim prediction model, the method further includes: Call the pre-built lightweight gradient boosting machine model; Get the preset hyperparameter optimization strategy and early stopping strategy; Tuning the lightweight gradient boosting machine model based on the hyperparameter optimization strategy and the early stopping strategy to obtain a corresponding tuned model; The tuned model is used as the target lightweight gradient boosting machine model.

3. The claim processing method based on artificial intelligence according to claim 1, characterized in that: Before the step of calling the preset claim prediction model, the method further includes: Obtain pre-collected historical multi-dimensional claims data; Performing feature engineering on the historical multi-dimensional claims data to construct corresponding sample data, and dividing the sample data into a training set and a validation set; Calling the target lightweight gradient boosting machine model; Training the target lightweight gradient boosting machine model based on the training set to obtain a trained first model; Performing a performance evaluation on the first model based on the validation set to obtain a performance evaluation result, and optimizing the first model based on the performance evaluation result to obtain a second model that meets the performance requirements; The second model is used as the claim prediction model.

4. The claim processing method based on artificial intelligence according to claim 3 is characterized in that: After the step of using the second model as the claim prediction model, the method further includes: Regularly collect newly generated designated claims data in the production environment; Call the preset data drift detection tool; Calculating the distribution difference between the specified claim data and the historical multi-dimensional claim data based on the data drift detection tool to obtain a corresponding distribution difference result; Determine whether the distribution difference result is greater than a preset threshold; If so, the claim prediction model is adjusted based on a preset adjustment strategy.

5. The claim processing method based on artificial intelligence according to claim 1, characterized in that: The step of preprocessing the multi-dimensional claims data to obtain corresponding first claims data specifically includes: Performing data cleaning on the multi-dimensional claims data to obtain corresponding first processed data; performing standardization processing on the first processed data to obtain corresponding second processed data; performing encoding processing on the second processed data to obtain corresponding third processed data; The third processed data is used as the first claim settlement data.

6. The claim processing method based on artificial intelligence according to claim 1, characterized in that: The step of performing feature derivation processing on the first claim data to construct corresponding second claim data specifically includes: Obtaining interaction information between features of the first claim data; performing corresponding feature derivation processing on the first claim data based on the interactive information to obtain derived feature data; Combining the first claim data with the derived characteristic data to obtain corresponding combined data; The combined data is used as the second claim settlement data.

7. The claim processing method based on artificial intelligence according to claim 1, characterized in that: The step of reviewing the claim prediction result specifically includes: Get the preset anomaly detection algorithm; performing fraud detection on the multi-dimensional claims data based on the anomaly detection algorithm to obtain corresponding fraud detection results; If the fraud detection result indicates that there is no fraud risk in the multi-dimensional claims data, then the claims prediction result is determined to have passed the review; If the fraud detection result indicates that there is a fraud risk in the multi-dimensional claims data, it is determined that the claims prediction result has failed the review.

8. An artificial intelligence-based claims processing device, characterized in that: include: A receiving module, configured to receive a claim application submitted by a user through a front-end interface; wherein the claim application carries multi-dimensional claim data, and the multi-dimensional claim data includes at least policyholder information, agent information, and case information; A preprocessing module, configured to perform data preprocessing on the multi-dimensional claims data to obtain corresponding first claims data; a derivation module, configured to perform feature derivation processing on the first claim data to construct corresponding second claim data; A first calling module is used to call a preset claim prediction model; wherein the claim prediction model is a model obtained by training a tuned target lightweight gradient boosting machine model based on pre-constructed sample data; A prediction module, configured to perform prediction processing on the second claim data based on the claim prediction model to obtain a corresponding claim prediction result; An audit module, used for auditing the claim prediction result; A processing module is used to perform corresponding claim processing on the claim application based on the claim prediction result if the claim prediction result passes the review.

9. A computer device, characterized in that: The method 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 claims processing method based on artificial intelligence are implemented as described in any one of claims 1 to 7.

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

Images