Claims analysis method, device, computer equipment and storage medium

Abstract

The embodiments of the present application belong to the field of financial technology and relate to a claims analysis method, apparatus, computer equipment, and storage medium. The method comprises the following steps: obtaining a multimodal claims dataset; performing feature extraction and feature fusion on the multimodal claims dataset to obtain a feature representation of the claims case; performing model training based on the feature representation of the claims case to obtain a case classification model; obtaining a current claims case, classifying the current claims case according to the case classification model to obtain the current case type; obtaining corresponding historical claims amount data according to the current case type, and performing amount distribution calculation based on the historical claims amount data to obtain claim amount distribution characteristics; constructing a claim amount prediction model based on the claim amount distribution characteristics, and inputting the current claim case into the claim amount prediction model to obtain the case claim amount. The present application can achieve accurate claim case amount prediction to improve the accuracy of claim case processing.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, specifically to the field of financial technology, and in particular to a claims analysis method, apparatus, computer equipment and storage medium. Background Art

[0002] In the field of claims analysis, time series data on claims amounts constitutes a crucial basis for insurance companies' risk management, financial planning, and customer service decisions. However, this data often exhibits extremely complex fluctuation patterns and trends, posing significant challenges for accurate analysis. As a key information carrier documenting the evolution of claims amounts over time, time series data possesses inherent characteristics, including long-term trends, cyclical fluctuations, seasonal variations, and the potential impact of unexpected events. These intertwined elements together create a complex picture of claims fluctuations.

[0003] While traditional time series decomposition methods, such as moving averages and exponential smoothing, can reveal certain trends and cyclical components in data to a certain extent, their analytical capabilities are limited when faced with the multi-modal effects of claims data. These methods often struggle to accurately distinguish and characterize the slow evolution of long-term trends, the regularity of cyclical fluctuations, and the specific forms of seasonal variation. This leads to biased analytical results and hinders the ability to provide insurers with accurate risk assessments and business guidance.

[0004] Furthermore, the diversity of claim types further complicates analysis. Different types of claims have vastly different distribution characteristics, ranging from minor vehicle scratches to severe natural disaster losses. This significant variation in amount distribution requires anomaly detection algorithms to be highly flexible and adaptable, automatically adjusting thresholds based on different case types to accurately identify potentially anomalous claims. However, existing single anomaly detection algorithms often lack this adaptability, significantly compromising the accuracy and efficiency of anomaly detection in the face of diverse case types. Summary of the Invention

[0005] The purpose of the embodiments of the present application is to propose a claims analysis method, apparatus, computer equipment and storage medium to solve the problem of being unable to accurately and effectively determine the claim amount of a claims case.

[0006] In order to solve the above technical problems, the present application provides a claims analysis method, which adopts the following technical solutions:

[0007] Acquiring claim case information, preprocessing the claim case information, and obtaining a multimodal claim data set;

[0008] Performing feature extraction and feature fusion on the multimodal claims dataset to obtain a feature representation of the claims case;

[0009] Perform model training based on the feature representation of the claim cases to obtain a case classification model;

[0010] Obtaining a current claim case, and classifying the current claim case according to the case classification model to obtain the current case type;

[0011] Obtaining corresponding historical claim amount data according to the current case type, and performing amount distribution calculation based on the historical claim amount data to obtain claim amount distribution characteristics;

[0012] A claim amount prediction model is constructed based on the claim amount distribution characteristics, and the current claim case is input into the claim amount prediction model to obtain the case claim amount.

[0013] Furthermore, the steps of obtaining claim case information, preprocessing the claim case information, and obtaining a multimodal claim data set specifically include:

[0014] Obtaining a claim case number, and extracting the claim case information from a database according to the claim case number, wherein the claim case information includes case text information, case image information, and case audio information;

[0015] Performing word segmentation and stop word removal on the case text information to obtain standard case text information;

[0016] performing denoising and normalization processing on the case image information to obtain standard case image information;

[0017] Performing noise reduction and endpoint detection processing on the case audio information to obtain standard case audio information;

[0018] The standard case text information, the standard case image information, and the standard case audio information are integrated into the multimodal claims data set.

[0019] Furthermore, the multimodal claims dataset includes standard case text information, standard case image information, and standard case audio information. The step of performing feature extraction and feature fusion on the multimodal claims dataset to obtain a feature representation of the claims case specifically includes:

[0020] Inputting the standard case text information into a preset language processing model for feature extraction to obtain a case text feature vector;

[0021] Extracting features of the standard case image information based on a preset convolutional neural network to obtain a case image feature vector;

[0022] Extracting Mel-frequency cepstral coefficients from the standard case audio information, and using the extracted Mel-frequency cepstral coefficients as the case audio feature vector;

[0023] The case text feature vector, the case image feature vector, and the case audio feature vector are fused to obtain the claim case feature representation.

[0024] Furthermore, the step of fusing the case text feature vector, the case image feature vector, and the case audio feature vector to obtain the claim case feature representation specifically includes:

[0025] Obtain text feature vector weights, image feature vector weights, and audio feature vector weights;

[0026] Performing weighted fusion on the case text feature vector, the case image feature vector, and the case audio feature vector according to the text feature vector weight, the image feature vector weight, and the audio feature vector weight to obtain a fused feature vector;

[0027] Feature extraction is performed on the fused feature vector based on a preset convolutional neural network to obtain the feature representation of the claim case.

[0028] Furthermore, the step of obtaining the current claim case, classifying the current claim case according to the case classification model, and obtaining the current case type specifically includes:

[0029] Obtaining a current case number, and extracting a current claim case from a database according to the current case number;

[0030] Performing feature extraction and feature fusion on the current claim case to obtain a feature representation of the current case;

[0031] The feature representation of the current case is input into the case classification model to obtain the current case type.

[0032] Furthermore, the step of obtaining corresponding historical claim amount data according to the current case type, and performing amount distribution calculation based on the historical claim amount data to obtain claim amount distribution characteristics specifically includes:

[0033] Extracting the corresponding historical claim amount data from the database according to the current case type;

[0034] Preprocessing the historical claim amount data to obtain a valid historical claim amount;

[0035] Calculate the average of the valid historical claim amounts, and determine the average level of claim amounts for the current case type based on the calculated average;

[0036] Calculating the variance of the valid historical claim amounts and determining the variance as the degree of dispersion of the claim amounts for the current case type;

[0037] Calculate quantiles of the valid historical claim amounts, and characterize the claim amount distribution characteristics of the current case type based on the quantiles;

[0038] The average level of the claim amount, the dispersion degree of the claim amount, and the distribution characteristics of the claim amount are used as the claim amount distribution characteristics.

[0039] Furthermore, the step of constructing a claim amount prediction model based on the claim amount distribution characteristics specifically includes:

[0040] Obtaining the timestamp information of the current claim case, extracting historical case information of the same period from the database based on the timestamp information, and analyzing the claim characteristics of the historical case information of the same period to obtain the claim amount characteristics of the same period;

[0041] Extracting historical case information from a database based on the current case type, and analyzing the claim characteristics of the historical case information to obtain the characteristics of the claim amounts of the same type;

[0042] The claim amount prediction model is constructed based on the claim amount distribution characteristics, the claim amount characteristics of the same period, and the claim amount characteristics of the same type.

[0043] In order to solve the above technical problems, the present application also provides a claims analysis device, which adopts the following technical solutions:

[0044] An information processing module, configured to obtain claim case information, pre-process the claim case information, and obtain a multimodal claim data set;

[0045] A feature processing module, configured to perform feature extraction and feature fusion on the multimodal claims dataset to obtain a feature representation of the claims case;

[0046] A model training module, configured to perform model training based on the feature representation of the claim cases to obtain a case classification model;

[0047] A case classification module is used to obtain a current claim case, classify the current claim case according to the case classification model, and obtain the current case type;

[0048] A feature extraction module is used to obtain corresponding historical claim amount data according to the current case type, and perform amount distribution calculation based on the historical claim amount data to obtain claim amount distribution characteristics;

[0049] The amount prediction module is used to construct a claim amount prediction model based on the claim amount distribution characteristics, and input the current claim case into the claim amount prediction model to obtain the case claim amount.

[0050] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0051] A computer device includes a memory and a processor, wherein the memory stores computer-readable instructions, and the processor implements the steps of any of the above-mentioned claim analysis methods when executing the computer-readable instructions.

[0052] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:

[0053] A computer-readable storage medium having computer-readable instructions stored thereon, wherein the computer-readable instructions, when executed by a processor, implement the steps of the claim analysis method as described in any one of the above items.

[0054] Compared with the prior art, the embodiments of the present application have the following main beneficial effects: the embodiment obtains claim case information, pre-processes the claim case information, and obtains a multimodal claim data set; performs feature extraction and feature fusion on the multimodal claim data set to obtain a claim case feature representation; performs model training based on the claim case feature representation to obtain a case classification model; obtains a current claim case, classifies the current claim case according to the case classification model, and obtains the current case type; obtains corresponding historical claim amount data according to the current case type, and calculates the amount distribution based on the historical claim amount data to obtain the claim amount distribution characteristics; constructs a claim amount prediction model based on the claim amount distribution characteristics, and inputs the current claim case into the claim amount prediction model to obtain the case claim amount. This effectively achieves accurate claim case amount prediction, thereby improving the accuracy of claim case processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] 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.

[0056] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0057] Figure 2A flowchart of an embodiment of a claims analysis method according to the present application;

[0058] Figure 3 yes Figure 2 A flowchart of a specific implementation of step S10;

[0059] Figure 4 yes Figure 2 A flowchart of a specific implementation of step S20;

[0060] Figure 5 yes Figure 4 A flowchart of a specific implementation of step S204;

[0061] Figure 6 yes Figure 2 A flowchart of a specific implementation of step S40;

[0062] Figure 7 yes Figure 2 A flowchart of a specific implementation of step S50;

[0063] Figure 8 yes Figure 2 A flowchart of a specific implementation of step S60;

[0064] Figure 9 is a schematic structural diagram of an embodiment of a claims analysis device according to the present application;

[0065] Figure 10 It is a structural diagram of an embodiment of a computer device according to the present application. DETAILED DESCRIPTION

[0066] 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.

[0067] References to "embodiments" herein 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 refer to unrelated or alternative embodiments that are 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.

[0068] 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.

[0069] 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.

[0070] 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.

[0071] 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.

[0072] 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 .

[0073] It should be noted that the claims analysis method provided in the embodiment of the present application is generally executed by a server / terminal device, and accordingly, the claims analysis device is generally set in the server / terminal device.

[0074] It should be understood that Figure 1The 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.

[0075] Continue to refer Figure 2 , shows a flow chart of an embodiment of a method for claim analysis according to the present application. The method for claim analysis comprises the following steps:

[0076] Step S10: obtaining claim case information, preprocessing the claim case information, and obtaining a multimodal claim data set;

[0077] In this embodiment, claim case information refers to relevant information data related to insurance claim cases in insurance business. Claim case information includes basic case information, specific case details, insured person information, case-related records, etc. Among them, basic case information includes case number, reporting time, reporter information, etc., specific case details include the time, location, cause of the accident, and the loss situation involved (such as personal injury, property damage, etc.), insured person information includes the insured person's identity information, contact information, occupation and other personal background information, and case-related records include reporting materials, investigation reports, medical certificates, legal documents, etc. A multimodal claim data set is a collection of claim data including multiple modalities. In this embodiment, the multimodal claim data set includes text information, image information, and audio information of the claim case.

[0078] Step S20, performing feature extraction and feature fusion on the multimodal claims dataset to obtain a feature representation of the claims case;

[0079] In this embodiment, the claim case feature representation is a feature representation generated by integrating the case text feature vector, the case image feature vector, and the case audio feature vector. The claim case feature representation is used to characterize the overall characteristics of the claim case.

[0080] Step S30, performing model training based on the claim case feature representation to obtain a case classification model;

[0081] In this embodiment, the model training steps include: training a first case classification model using a support vector machine algorithm based on the claim case feature representation; if the performance of the first case classification model does not meet the preset requirements, training a second case classification model using a random forest algorithm; if the performance of the second case classification model still does not meet the preset requirements, training a third case classification model using a deep neural network algorithm; and evaluating the performance of the first case classification model, the second case classification model, and the third case classification model through cross-validation, and selecting the model with the best performance as the case classification model.

[0082] Step S40: obtaining a current claim case, classifying the current claim case according to the case classification model, and obtaining the current case type;

[0083] In this embodiment, the current claim case refers to a case that currently requires claim analysis to obtain a claim amount. Feature extraction and feature fusion are performed on the current claim case to obtain a current case feature representation that represents the overall features of the current claim case. The current case feature representation is then input into a case classification model for case classification processing to obtain an accurately identified current case type.

[0084] Step S50, obtaining corresponding historical claim amount data according to the current case type, and performing amount distribution calculation based on the historical claim amount data to obtain claim amount distribution characteristics;

[0085] In this embodiment, historical claim amount data refers to the specific numerical values ​​of claim amounts involved in historical claim cases of the same type as the current claim case (such as auto insurance, medical insurance, critical illness insurance, etc.). It is mainly used to reflect the historical distribution of claim amounts for such cases, including statistical indicators such as the average claim amount, the highest claim amount, and the lowest claim amount. Claim amount distribution characteristics are representative information obtained by calculating the amount distribution of historical claim amount data. Claim amount distribution characteristics include the average claim amount level, the degree of claim amount dispersion, and the claim amount distribution characteristics. These claim amount distribution characteristics can effectively help the model understand the claim amount patterns corresponding to the case.

[0086] Step S60: construct a claim amount prediction model based on the claim amount distribution characteristics, and input the current claim case into the claim amount prediction model to obtain the case claim amount.

[0087] In this embodiment, the claim amount prediction model is a model used to predict the claim amount of a case. The claim amount prediction model can adopt a random forest model. The random forest model is an integrated learning model based on multiple decision trees. It improves the prediction accuracy through voting or averaging to obtain the accurate case claim amount.

[0088] In this embodiment, by obtaining claim case information, preprocessing the claim case information to obtain a multi-modal claim data set; extracting and fusing features of the multi-modal claim data set to obtain a claim case feature representation; training a model based on the claim case feature representation to obtain a case classification model; obtaining a current claim case, distinguishing the type of the current claim case according to the case classification model to obtain the current case type; obtaining corresponding historical claim amount data according to the current case type, and calculating the amount distribution according to the historical claim amount data to obtain the characteristics of the claim amount distribution; constructing a claim amount prediction model based on the characteristics of the claim amount distribution, and inputting the current claim case into the claim amount prediction model to obtain the claim amount of the case. Thus, it effectively realizes accurate prediction of the claim amount of the claim case to improve the accuracy of claim case processing.

[0089] Reference Figure 3 , in some optional implementation manners of this embodiment, step S10 includes the following steps:

[0090] Step S101, obtain a claim case number, and extract the claim case information from the database according to the claim case number, where the claim case information includes case text information, case image information, and case audio information;

[0091] In this embodiment, the claim case number can be obtained by inputting in the system, and traversing and matching queries are performed in the database through the claim case number to effectively obtain the corresponding claim case information. The case text information includes the information of the applicant and the insured, accident description, medical expense list, diagnosis certificate, and reporting time; the case image information includes photos of the accident scene, medical certificate photos, and other relevant photos (such as identity photos of the applicant and the insured, etc., for identity verification); the case audio information includes accident scene recordings and call recordings.

[0092] Step S102, perform word segmentation and stop word removal processing on the case text information to obtain standard case text information;

[0093] In this embodiment, word segmentation is the process of splitting a continuous text string into individual words or phrases, and the word segmentation operation on the case text information can be performed through a word segmentation tool. Stop word removal refers to removing words that frequently appear in the text but contribute little to the meaning of the text, such as "de", "le", "zai", etc. The corresponding stop words can be compared and removed from the case text information through a preset stop word list, such as the Harbin Institute of Technology stop word list, Baidu stop word list, etc.

[0094] Step S103, perform denoising and normalization processing on the case image information to obtain standard case image information;

[0095] In this embodiment, denoising refers to removing noise from an image to improve its clarity and quality. A Gaussian filtering denoising algorithm can be used to denoise the case image information. Normalization refers to adjusting the pixel values ​​of the image corresponding to the case image information to a certain range for subsequent processing or analysis. By normalizing the image pixel values ​​to a range between 0 and 1 or 0 and 255, the case image information is processed into unified image information.

[0096] Step S104, performing noise reduction and endpoint detection processing on the case audio information to obtain standard case audio information;

[0097] In this embodiment, noise reduction refers to removing noise components from the audio signal to improve audio clarity and intelligibility. Noise reduction can be performed using spectral subtraction or a deep learning-based noise reduction algorithm. Endpoint detection accurately identifies the start and end points of valid speech segments from the audio signal to facilitate subsequent speech processing or analysis. Endpoint detection can be implemented using an energy-based endpoint detection algorithm, which identifies speech segments by calculating the short-term energy of the audio signal.

[0098] Step S105 : Integrate the standard case text information, the standard case image information, and the standard case audio information into the multimodal claims data set.

[0099] In this embodiment, a multimodal claims dataset is obtained by storing standard case text information, standard case image information, and standard case audio information in a pre-created dataset folder.

[0100] This embodiment obtains a claim case number and extracts the claim case information from a database based on the claim case number, wherein the claim case information includes case text information, case image information, and case audio information; performs word segmentation and stop word removal on the case text information to obtain standard case text information; performs denoising and normalization on the case image information to obtain standard case image information; performs denoising and endpoint detection on the case audio information to obtain standard case audio information; and integrates the standard case text information, the standard case image information, and the standard case audio information into the multimodal claim dataset. This effectively achieves a standard and effective multimodal claim dataset based on claim case information acquisition, facilitating subsequent feature extraction and feature fusion processing.

[0101] refer to Figure 4 In some optional implementations of this embodiment, step S20 includes the following steps:

[0102] Step S201: Inputting the standard case text information into a preset language processing model for feature extraction to obtain a case text feature vector;

[0103] In this embodiment, the preset language processing model can adopt a Transformer model, such as a BERT model, by using word embedding technology (BERT's embedding layer) to convert standard case text information into word vectors, and then input the word vectors into the preset language processing model to obtain the case text feature vector.

[0104] Step S202: extracting features from the standard case image information based on a preset convolutional neural network to obtain a case image feature vector;

[0105] In this embodiment, the preset convolutional neural network is a pre-trained convolutional neural network, which includes an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer. The input layer is responsible for receiving preprocessed text data, the convolutional layer is used to extract local features in the text, the pooling layer is used to reduce the feature dimension and reduce the amount of calculation while retaining important information, the fully connected layer is used to map the extracted features to the target space, and the output layer can be designed as a classifier or feature extractor according to task requirements. Feature extraction of standard case image information is performed through the preset convolutional neural network, thereby accurately extracting the case image feature vector.

[0106] Step S203, extracting Mel-frequency cepstral coefficients from the standard case audio information, and using the extracted Mel-frequency cepstral coefficients as the case audio feature vector;

[0107] In this embodiment, MFCC (Mel-Frequency Cepstral Coefficient) is an audio feature extraction method based on the auditory characteristics of the human ear, which can effectively capture the characteristics of audio signals. The Mel-Frequency Cepstral Coefficient extraction of standard case audio information includes the following steps: performing a fast Fourier transform (FFT) on each audio frame of the standard case audio information to convert the time domain signal into a frequency domain signal; passing the spectrum obtained by FFT through a set of triangular filters uniformly distributed on the Mel scale to obtain a Mel spectrum; calculating the logarithmic energy of the output of each Mel filter; performing a discrete cosine transform on the logarithmic energy spectrum, and taking the first N coefficients after the transform (usually 12 or 13) as MFCC coefficients; combining the MFCC coefficients of each audio frame to form an MFCC sequence; and selecting the entire MFCC sequence as the case audio feature vector.

[0108] Step S204: Fusing the case text feature vector, the case image feature vector, and the case audio feature vector to obtain the claim case feature representation.

[0109] In this embodiment, the fusion method of the case text feature vector, the case image feature vector, and the case audio feature vector adopts weighted summation. By assigning different weights to each feature vector and then performing weighted summation, an overall claim case feature representation is obtained.

[0110] This embodiment extracts features from the standard case text information based on a preset convolutional neural network to obtain a case text feature vector; inputs the standard case image information into a preset language processing model for feature extraction to obtain a case image feature vector; extracts Mel-frequency cepstral coefficients from the standard case audio information, and uses the extracted Mel-frequency cepstral coefficients as the case audio feature vector; and fuses the case text feature vector, the case image feature vector, and the case audio feature vector to effectively obtain a claim case feature representation that characterizes the overall characteristics of the claim case, so as to facilitate subsequent model training processing.

[0111] refer to Figure 5 In some optional implementations of this embodiment, step S204 includes the following steps:

[0112] Step S2041, obtaining text feature vector weights, image feature vector weights, and audio feature vector weights;

[0113] In this embodiment, the text feature vector weight, image feature vector weight, and audio feature vector weight are the weight information corresponding to the case text feature vector, case image feature vector, and case audio feature vector, respectively. The text feature vector weight, image feature vector weight, and audio feature vector weight can be set and adjusted based on heuristic methods (such as based on feature vector dimension, modal importance, etc.).

[0114] Step S2042, performing weighted fusion on the case text feature vector, the case image feature vector, and the case audio feature vector according to the text feature vector weight, the image feature vector weight, and the audio feature vector weight to obtain a fused feature vector;

[0115] In this embodiment, the weights corresponding to the eigenvectors are multiplied by the eigenvectors to obtain weighted eigenvectors, and then the weighted eigenvectors are added and summed to achieve weighted fusion, thereby obtaining a fused eigenvector after fusing multiple eigenvectors.

[0116] Step S2043: extract features from the fused feature vector based on a preset convolutional neural network to obtain a feature representation of the claim case.

[0117] In this embodiment, feature extraction can be implemented based on a pre-trained CNN model (such as VGG, ResNet, etc.). By taking the fused feature vector as the input of the CNN model, and extracting high-level features in the fused feature vector through the CNN's convolutional layer, pooling layer, and fully connected layer structures, the feature representation of the claim case is obtained.

[0118] This embodiment obtains the text feature vector weight, the image feature vector weight, and the audio feature vector weight; performs weighted fusion on the case text feature vector, the case image feature vector, and the case audio feature vector according to the text feature vector weight, the image feature vector weight, and the audio feature vector weight to obtain a fused feature vector; performs feature extraction on the fused feature vector based on a preset convolutional neural network to effectively obtain a claim case feature representation of the fused case text feature vector, the case image feature vector, and the case audio feature vector, so as to provide reliable training data for subsequent case classification model training processing.

[0119] Continue to refer Figure 6 In some optional implementations of this embodiment, step S40 includes the following steps:

[0120] Step S401: obtaining a current case number, and extracting a current claim case from a database according to the current case number;

[0121] In this embodiment, the current case number is information corresponding to the current claim case. By using the current case number as a query condition, a traversal query is performed in the database, and the queried information is extracted to obtain the current claim case.

[0122] Step S402: extracting and fusing features of the current claim case to obtain a feature representation of the current case;

[0123] In this embodiment, the current claim case includes the corresponding current claim case text information, current claim case image information, and current claim case audio information. The feature extraction and feature fusion of the current claim case are the same as the feature extraction and feature fusion method of the multimodal claim data set mentioned above. By performing feature extraction and feature fusion on the current claim case, the current case feature representation that characterizes the overall characteristics of the current claim case can be effectively obtained.

[0124] Step S403: Input the feature representation of the current case into the case classification model to obtain the current case type.

[0125] In this embodiment, the current case features are input into the case classification model to determine the case type corresponding to the current claim case. The case type is formed based on the characteristics and feature classification of the case. The case types may include traffic accident claims, medical health claims, property damage claims, life insurance claims, accidental injury claims, travel insurance claims, liability insurance claims, and other special claims (including pet insurance, legal expenses insurance, credit insurance and other special types of insurance claims).

[0126] This embodiment obtains the current case number and extracts the current claim case from the database according to the current case number; performs feature extraction and feature fusion on the current claim case to obtain the current case feature representation; inputs the current case feature representation into the case classification model, thereby effectively obtaining the current case type corresponding to the current claim case, so as to facilitate the subsequent extraction of historical claim amount data.

[0127] Continue to refer Figure 7 In some optional implementations of this embodiment, step S50 includes the following steps:

[0128] Step S501, extracting the corresponding historical claim amount data from the database according to the current case type;

[0129] In this embodiment, a traversal query is performed in the database using the current case type as a query condition, and the queried data is extracted to obtain historical claim amount data.

[0130] Step S502: pre-process the historical claim amount data to obtain a valid historical claim amount;

[0131] In this embodiment, preprocessing includes outlier processing and missing value processing. Outliers refer to values ​​that significantly deviate from the majority of data points. They may be caused by data entry errors, measurement errors, or extreme events. Outliers can be detected through statistical methods such as Z-score, IQR (interquartile range), etc., and the detected outliers are then deleted, replaced, or retained to form a preprocessing of historical claim amount data.

[0132] Step S503, calculating the mean of the valid historical claim amounts, and determining the average level of claim amounts for the current case type based on the calculated mean;

[0133] In this embodiment, the mean calculation formula is: mean = (sum of all valid claim amounts) / (number of valid claim amounts). The mean calculation formula is used to calculate the mean of valid historical claim amounts to obtain the average level of claim amounts for the current case type.

[0134] Step S504: Calculate the variance of the valid historical claim amounts, and determine the variance as the dispersion of the claim amounts of the current case type;

[0135] In this embodiment, variance is a statistic that measures the degree of data dispersion. It represents the average of the squares of the differences between data points and the mean. The formula for calculating variance is: variance = Σ[(x-mean)^2] / n, where x is the valid historical claim amount data point, mean is the average of the valid historical claim amounts, and n is the number of valid historical claim amount data points. A larger variance indicates a higher degree of claim amount dispersion, i.e., greater differences between data points; a smaller variance indicates a lower degree of claim amount dispersion, i.e., smaller differences between data points. The variance of valid historical claim amounts is used to represent the degree of claim amount dispersion for the current case type.

[0136] Step S505: Calculate quantiles of the valid historical claim amounts, and characterize the claim amount distribution characteristics of the current case type based on the quantiles;

[0137] In this embodiment, the quantile is the value at a specific position after sorting a set of data from small to large. Common quantiles include the 25th percentile (lower quartile), the 50th percentile (median), the 75th percentile (upper quartile), etc. In spreadsheet software (such as Excel), you can use the built-in PERCENTILE function to calculate the value of a specific percentile to obtain the quantile. Characterizing the distribution characteristics of the claim amount according to the quantile includes the distribution form: by comparing the values ​​of different quantiles, you can preliminarily judge the distribution form of the claim amount. For example, if the median is much lower than the upper quartile and much higher than the lower quartile, it may indicate that the claim amount has a large volatility; Central tendency: the median, as the center position of the distribution, can reflect the central tendency of the claim amount. If the median is low, it means that most claim amounts are low; conversely, if the median is high, it means that most claim amounts are high. Dispersion: By comparing the differences between different quantiles (such as the interquartile range, which is the difference between the upper and lower quartiles), the dispersion of claim amounts can be assessed. The larger the difference, the greater the dispersion of claim amounts; the smaller the difference, the lower the dispersion of claim amounts. Outlier detection: Extremely low or high quantiles (such as the 1st or 99th percentiles) can help identify outliers or extreme cases in claim amounts.

[0138] Step S506: The average level of the claim amount, the dispersion degree of the claim amount, and the claim amount distribution characteristics are used as the claim amount distribution characteristics.

[0139] In this embodiment, data integration is performed on the average level of claim amounts, the dispersion of claim amounts, and the distribution characteristics of claim amounts to obtain the distribution characteristics of claim amounts.

[0140] This embodiment calculates the degree of influence of the characteristics of the crop growth monitoring dataset on the crop growth status assessment result according to a preset feature importance analysis method; extracts a preset influence degree threshold and determines whether the characteristic influence degree is greater than or equal to the influence degree threshold; if the characteristic influence degree value is greater than or equal to the influence degree threshold, the feature corresponding to the characteristic influence degree value is used as the key influencing factor; if the characteristic influence degree value is less than the influence degree threshold, the feature corresponding to the characteristic influence degree value is used as the non-key influencing factor; the crop growth monitoring dataset is screened according to the key influencing factors to obtain a key influence dataset; a key time series dataset is constructed according to the key influence dataset; features are extracted from the key time series dataset according to a time series analysis method to obtain a growth trend feature vector; the growth trend feature vector and the crop growth status assessment result are input into a pre-trained key growth assessment model, thereby effectively obtaining a growth trend prediction result that accurately predicts crop growth, so as to facilitate subsequent crop growth claims analysis.

[0141] Continue to refer Figure 8 In some optional implementations of this embodiment, step S60 includes the following steps:

[0142] Step S601: obtaining the timestamp information of the current claim case, extracting historical case information of the same period from the database based on the timestamp information, and analyzing the claim characteristics of the historical case information of the same period to obtain the claim amount characteristics of the same period;

[0143] In this embodiment, the timestamp automatically generated when the current claim is submitted is queried to obtain timestamp information. This timestamp information is then used as a query condition to perform a traversal query in the database to obtain historical case information corresponding to the current claim. In this embodiment, the historical case information corresponds to a time range corresponding to the case. This time range is initially set to one month and can be adjusted accordingly based on actual circumstances. Analysis of the claims characteristics of historical case information from the same period includes statistical distribution: calculating statistics such as the mean, median, maximum, and minimum values ​​of the claim amounts for the historical cases from the same period to understand the distribution of claim amounts; trend analysis: plotting a trend chart of claim amounts over time to observe whether there are significant fluctuations or trends; and classification analysis: classifying cases by claim type or claim reason, calculating the claim amounts for each category, and then comparing and analyzing them. Based on these analysis results, the claims characteristics of the historical cases from the same period can be summarized. For example, it can be determined which types of cases have higher claim amounts and which reasons lead to larger fluctuations in claim amounts, thereby obtaining a characteristic of the claim amounts for the same period.

[0144] Step S602: extracting historical case information from a database based on the current case type, and analyzing the claim characteristics of the historical case information to obtain the characteristics of the claim amounts of the same type;

[0145] In this embodiment, by using the current case type as a query condition to traverse the database and query, historical case information corresponding to the current case type is obtained. The claim characteristics analysis of historical case information includes claim amount statistics: calculating the average, median, maximum, minimum and other statistical quantities of the same type of claim amounts to understand the distribution of the claim amounts, drawing a frequency distribution histogram or box plot of the claim amounts to intuitively display the distribution form and dispersion of the claim amounts; claim amount trend analysis: if the historical case data contains information at multiple time points, a trend chart of the claim amount changing over time can be drawn to observe whether there are obvious fluctuations or trends, and analyze the reasons for the claim amount trend, such as policy changes, changes in the market environment, changes in customer behavior, etc.; claim amount and case characteristics correlation analysis: analyzing the correlation between the claim amount and case characteristics (such as claim reasons, case complexity, claim amount, etc.), using tools such as scatter plots, line graphs or correlation matrices for visual analysis to reveal potential correlations and trends; claim amount characteristics summary: based on the above analysis, summarize the characteristics of the same type of claim amounts. For example, it can be pointed out which claim reasons lead to higher claim amounts, which case characteristics are significantly correlated with claim amounts, etc., in order to obtain the characteristics of claim amounts in the same period.

[0146] Step S603: construct the claim amount prediction model based on the claim amount distribution characteristics, the claim amount characteristics of the same period, and the claim amount characteristics of the same type.

[0147] In this embodiment, the claim amount prediction model adopts a random forest model. By adjusting the parameters of the random forest model, such as the number of trees, maximum depth, minimum number of samples, etc., and using the characteristics of the claim amount distribution, the characteristics of the claim amount in the same period, and the characteristics of the claim amount of the same type as training data, multiple training subsets are generated through bootstrap sampling. For each training subset, a decision tree is trained. During the training process, nodes are split by randomly selecting feature subsets to increase the diversity of the model, thereby effectively constructing an initial claim amount prediction model. Cross-validation and feature importance evaluation are used to identify the features that contribute most to the claim amount prediction, and feature importance evaluation results and cross-validation results are obtained. Finally, based on the feature importance evaluation results and cross-validation results, the parameters of the random forest are adjusted, such as increasing or decreasing the number of trees, adjusting the maximum depth, etc., to optimize the performance of the model and obtain a claim amount prediction model.

[0148] This embodiment obtains claims analysis environmental information; determines the risk status based on the claims analysis environmental information, the crop growth status assessment results, and the growth trend prediction results to obtain the risk status type; extracts the corresponding risk warning model from the database based on the risk status type; inputs the claims analysis environmental information, the crop growth status assessment results, and the growth trend prediction results into the risk warning model, thereby effectively obtaining risk warning information for accurate claims analysis based on the crop growth status assessment results and the growth trend prediction results, so as to facilitate risk avoidance based on the risk warning information.

[0149] 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).

[0150] 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.

[0151] Further references Figure 9 , as a response to the above Figure 1 In order to realize the method shown in the figure, the present application provides an embodiment of a claim analysis device. Figure 1 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0152] like Figure 9 As shown, the claim analysis device 700 of this embodiment includes: an information processing module 701, a feature processing module 702, a model training module 703, a case classification module 704, a feature extraction module 705, and an amount prediction module 706. Among them:

[0153] An information processing module 701 is used to obtain claim case information, pre-process the claim case information, and obtain a multimodal claim data set;

[0154] A feature processing module 702 is used to perform feature extraction and feature fusion on the multimodal claims dataset to obtain a feature representation of the claims case;

[0155] A model training module 703 is used to perform model training based on the claim case feature representation to obtain a case classification model;

[0156] The case classification module 704 is used to obtain a current claim case, classify the current claim case according to the case classification model, and obtain the current case type;

[0157] A feature extraction module 705 is configured to obtain corresponding historical claim amount data according to the current case type, and perform amount distribution calculation based on the historical claim amount data to obtain claim amount distribution characteristics;

[0158] The amount prediction module 706 is used to build a claim amount prediction model based on the claim amount distribution characteristics, and input the current claim case into the claim amount prediction model to obtain the case claim amount.

[0159] By employing the above-mentioned claims analysis device, this embodiment can obtain claims case information, pre-process the claims case information to obtain a multimodal claims dataset; perform feature extraction and feature fusion on the multimodal claims dataset to obtain a claims case feature representation; perform model training based on the claims case feature representation to obtain a case classification model; obtain a current claim case, classify the current claim case according to the case classification model to obtain the current case type; obtain corresponding historical claim amount data based on the current case type, and calculate the amount distribution based on the historical claim amount data to obtain the claim amount distribution characteristics; construct a claim amount prediction model based on the claim amount distribution characteristics, and input the current claim case into the claim amount prediction model to obtain the case claim amount. This effectively achieves accurate claim case amount prediction, thereby improving the accuracy of claims case processing.

[0160] To solve the above technical problems, the present application also provides a computer device. Figure 10 , Figure 10 This is a basic structural block diagram of the computer device in this embodiment.

[0161] The computer device 8 includes a memory 81, a processor 82, and a network interface 83 that are interconnected through a system bus. It should be noted that the figure only shows a computer device 8 with components 81-83, 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.

[0162] 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.

[0163] The memory 81 includes at least one type of readable storage medium, including flash memory, a hard disk, a multimedia card, a 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 storage, a magnetic disk, an optical disk, etc. In some embodiments, the memory 81 may be an internal storage unit of the computer device 8, such as the hard disk or memory of the computer device 8. In other embodiments, the memory 81 may also be an external storage device of the computer device 8, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash memory card, etc. Of course, the memory 81 may also include both the internal storage unit of the computer device 8 and its external storage device. In this embodiment, the memory 81 is generally used to store the operating system and various application software installed on the computer device 8, such as computer-readable instructions for the claims analysis method. In addition, the memory 81 can also be used to temporarily store various types of data that have been output or are to be output.

[0164] In some embodiments, the processor 82 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip. The processor 82 is generally used to control the overall operation of the computer device 8. In this embodiment, the processor 82 is used to execute computer-readable instructions stored in the memory 81 or process data, such as computer-readable instructions for executing the claims analysis method.

[0165] The network interface 83 may include a wireless network interface or a wired network interface. The network interface 83 is generally used to establish a communication connection between the computer device 8 and other electronic devices.

[0166] By using the above-mentioned computer device, this embodiment can obtain claim case information, pre-process the claim case information to obtain a multimodal claim data set; perform feature extraction and feature fusion on the multimodal claim data set to obtain a claim case feature representation; perform model training based on the claim case feature representation to obtain a case classification model; obtain a current claim case, classify the current claim case according to the case classification model to obtain the current case type; obtain corresponding historical claim amount data based on the current case type, and calculate the amount distribution based on the historical claim amount data to obtain the claim amount distribution characteristics; construct a claim amount prediction model based on the claim amount distribution characteristics, and input the current claim case into the claim amount prediction model to obtain the case claim amount. This effectively achieves accurate claim case amount prediction, thereby improving the accuracy of claim case processing.

[0167] The present application also provides another embodiment, namely, providing a computer-readable storage medium, which stores computer-readable instructions. The computer-readable instructions can be executed by at least one processor to enable the at least one processor to perform the steps of the claim analysis method as described above.

[0168] By using the computer-readable storage medium, this embodiment can obtain claim case information, pre-process the claim case information to obtain a multimodal claim data set, perform feature extraction and feature fusion on the multimodal claim data set to obtain a claim case feature representation, perform model training based on the claim case feature representation to obtain a case classification model, obtain a current claim case, classify the current claim case according to the case classification model to obtain the current case type, obtain corresponding historical claim amount data based on the current case type, calculate the amount distribution based on the historical claim amount data to obtain claim amount distribution characteristics, construct a claim amount prediction model based on the claim amount distribution characteristics, and input the current claim case into the claim amount prediction model to obtain the case claim amount. This effectively achieves accurate claim case amount prediction, thereby improving the accuracy of claim case processing.

[0169] 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.

[0170] 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 analysis method, characterized in that: The steps include: Acquiring claim case information, preprocessing the claim case information, and obtaining a multimodal claim data set; Performing feature extraction and feature fusion on the multimodal claims dataset to obtain a feature representation of the claims case; Perform model training based on the feature representation of the claim cases to obtain a case classification model; Obtaining a current claim case, and classifying the current claim case according to the case classification model to obtain the current case type; Obtaining corresponding historical claim amount data according to the current case type, and performing amount distribution calculation based on the historical claim amount data to obtain claim amount distribution characteristics; Building a claim amount prediction model based on the claim amount distribution characteristics, and inputting the current claim case into the claim amount prediction model to obtain the claim amount of the case; The multimodal claims dataset includes standard case text information, standard case image information, and standard case audio information. The step of performing feature extraction and feature fusion on the multimodal claims dataset to obtain a feature representation of the claims case specifically includes: Inputting the standard case text information into a preset language processing model for feature extraction to obtain a case text feature vector; Extracting features of the standard case image information based on a preset convolutional neural network to obtain a case image feature vector; Extracting Mel-frequency cepstral coefficients from the standard case audio information, and using the extracted Mel-frequency cepstral coefficients as the case audio feature vector; Fusing the case text feature vector, the case image feature vector, and the case audio feature vector to obtain the claim case feature representation; The step of obtaining corresponding historical claim amount data according to the current case type, and performing amount distribution calculation based on the historical claim amount data to obtain claim amount distribution characteristics specifically includes: Extracting the corresponding historical claim amount data from the database according to the current case type; Preprocessing the historical claim amount data to obtain a valid historical claim amount; Calculate the average of the valid historical claim amounts, and determine the average level of claim amounts for the current case type based on the calculated average; Calculating the variance of the valid historical claim amounts and determining the variance as the degree of dispersion of the claim amounts for the current case type; Calculate quantiles of the valid historical claim amounts, and characterize the claim amount distribution characteristics of the current case type based on the quantiles; The average level of the claim amount, the dispersion degree of the claim amount, and the distribution characteristics of the claim amount are used as the claim amount distribution characteristics.

2. The claim analysis method according to claim 1, characterized in that: The steps of obtaining claim case information, preprocessing the claim case information, and obtaining a multimodal claim data set specifically include: Obtaining a claim case number, and extracting the claim case information from a database according to the claim case number, wherein the claim case information includes case text information, case image information, and case audio information; Performing word segmentation and stop word removal on the case text information to obtain standard case text information; performing denoising and normalization processing on the case image information to obtain standard case image information; Performing noise reduction and endpoint detection processing on the case audio information to obtain standard case audio information; The standard case text information, the standard case image information, and the standard case audio information are integrated into the multimodal claims data set.

3. The claim analysis method according to claim 2, characterized in that: The step of fusing the case text feature vector, the case image feature vector, and the case audio feature vector to obtain the claim case feature representation specifically includes: Obtain text feature vector weights, image feature vector weights, and audio feature vector weights; Performing weighted fusion on the case text feature vector, the case image feature vector, and the case audio feature vector according to the text feature vector weight, the image feature vector weight, and the audio feature vector weight to obtain a fused feature vector; Feature extraction is performed on the fused feature vector based on a preset convolutional neural network to obtain the feature representation of the claim case.

4. The claim analysis method according to claim 1, characterized in that: The step of obtaining the current claim case, classifying the current claim case according to the case classification model, and obtaining the current case type specifically includes: Obtaining a current case number, and extracting a current claim case from a database according to the current case number; Performing feature extraction and feature fusion on the current claim case to obtain a feature representation of the current case; The feature representation of the current case is input into the case classification model to obtain the current case type.

5. The claim analysis method according to claim 1, characterized in that: The step of constructing a claim amount prediction model based on the claim amount distribution characteristics specifically includes: Obtaining the timestamp information of the current claim case, extracting historical case information of the same period from the database based on the timestamp information, and analyzing the claim characteristics of the historical case information of the same period to obtain the claim amount characteristics of the same period; Extracting historical case information from a database based on the current case type, and analyzing the claim characteristics of the historical case information to obtain the characteristics of the claim amounts of the same type; The claim amount prediction model is constructed based on the claim amount distribution characteristics, the claim amount characteristics of the same period, and the claim amount characteristics of the same type.

6. A claims analysis device, characterized in that: include: An information processing module, configured to obtain claim case information, pre-process the claim case information, and obtain a multimodal claim data set; A feature processing module, configured to perform feature extraction and feature fusion on the multimodal claims dataset to obtain a feature representation of the claims case; A model training module, configured to perform model training based on the feature representation of the claim cases to obtain a case classification model; A case classification module is used to obtain a current claim case, classify the current claim case according to the case classification model, and obtain the current case type; A feature extraction module is used to obtain corresponding historical claim amount data according to the current case type, and perform amount distribution calculation based on the historical claim amount data to obtain claim amount distribution characteristics; An amount prediction module is used to construct a claim amount prediction model based on the claim amount distribution characteristics, and input the current claim case into the claim amount prediction model to obtain the claim amount of the case; The multimodal claims dataset includes standard case text information, standard case image information, and standard case audio information. The step of performing feature extraction and feature fusion on the multimodal claims dataset to obtain a feature representation of the claims case specifically includes: Inputting the standard case text information into a preset language processing model for feature extraction to obtain a case text feature vector; Extracting features of the standard case image information based on a preset convolutional neural network to obtain a case image feature vector; Extracting Mel-frequency cepstral coefficients from the standard case audio information, and using the extracted Mel-frequency cepstral coefficients as the case audio feature vector; Fusing the case text feature vector, the case image feature vector, and the case audio feature vector to obtain the claim case feature representation; The step of obtaining corresponding historical claim amount data according to the current case type, and performing amount distribution calculation based on the historical claim amount data to obtain claim amount distribution characteristics specifically includes: Extracting the corresponding historical claim amount data from the database according to the current case type; Preprocessing the historical claim amount data to obtain a valid historical claim amount; Calculate the average of the valid historical claim amounts, and determine the average level of claim amounts for the current case type based on the calculated average; Calculating the variance of the valid historical claim amounts and determining the variance as the degree of dispersion of the claim amounts for the current case type; Calculate quantiles of the valid historical claim amounts, and characterize the claim amount distribution characteristics of the current case type based on the quantiles; The average level of the claim amount, the dispersion degree of the claim amount, and the distribution characteristics of the claim amount are used as the claim amount distribution characteristics.

7. A computer device, characterized in that: The method comprises a memory and a processor, wherein the memory stores computer-readable instructions, and the processor implements the steps of the claim analysis method according to any one of claims 1 to 5 when executing the computer-readable instructions.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the claim analysis method according to any one of claims 1 to 5.

Images