Intelligent contract-driven insurance claim settlement automation system

Abstract

The invention relates to the field of insurance claim settlement, in particular to an intelligent contract-driven insurance claim settlement automation system. The method comprises the following steps: constructing an original data matrix based on collected claim settlement data and non-claim settlement data, and carrying out decentralization and data standardization processing to obtain a standardized data matrix; constructing an adaptive logical reasoning tree model based on the standardized data matrix of the claim settlement data, and generating an output vector of an adaptive logical reasoning tree node; the condition threshold matrix is adjusted, the rule set matrix is evolved, then the total amount of claim settlement is calculated, and the adaptive logical reasoning tree model is evaluated and optimized; and calculating a final total amount of claim settlement based on the standardized data matrix of the non-claim settlement data. The problems that the rule set updating and model optimization process is rigid, and the flexibility and adaptability of claim decision making are insufficient are solved; data standardization and feature extraction lack robustness and accuracy; when a system of a fixed rule set processes multi-dimensional and high-complexity requests, decision deviation occurs.

Description

Technical Field

[0001] The present invention relates to the field of insurance claims, and in particular to an insurance claims automation system driven by smart contracts. Background Art

[0002] Current insurance claims automation technology has widely adopted models based on big data analysis and machine learning, especially in data processing and decision analysis. Through smart contract technology, the automation process is combined with blockchain technology to achieve transparency and credibility of the claims process, reduce manual participation, and improve efficiency and accuracy. In addition, the improvement of the adaptive logic reasoning tree model and covariance matrix further enhances the analysis ability of complex claims scenarios, enabling the insurance claims system to respond to different types of claims requests in real time, and improving the intelligence level of automated decision-making.

[0003] Although the above-mentioned insurance claims systems have made significant progress in automation and data processing, there are still some technical problems: first, the rule set update and model optimization process are relatively rigid, making it difficult to dynamically respond to new claims data and scenarios, resulting in insufficient flexibility and adaptability in claims decisions; second, the data standardization and feature extraction methods still have room for improvement in robustness and accuracy when dealing with extreme data values ​​and complex data correlations; in addition, insurance claims systems based on fixed rule sets will have decision-making biases when processing multi-dimensional and highly complex claims requests, affecting the fairness of the final claim amount. Summary of the invention

[0004] The present invention provides an insurance claims automation system driven by smart contracts, which solves the problems that the rule set update and model optimization process are relatively rigid, making it difficult to dynamically respond to new claims data and scenarios, resulting in insufficient flexibility and adaptability in claims decisions; the data standardization and feature extraction methods lack robustness and accuracy when dealing with extreme data values ​​and complex data correlations; and the insurance claims system based on a fixed rule set will have decision-making biases when processing multi-dimensional and highly complex claims requests, affecting the fairness of the final total claim amount.

[0005] The smart contract-driven insurance claims automation system of the present invention specifically includes the following technical solutions:

[0006] The smart contract driven insurance claims automation method includes the following steps:

[0007] S1: Collect original claims data; construct original data matrices of the paid claims data and unpaid claims data respectively based on the paid claims data and unpaid claims data in the original claims data; decentralize and standardize the original data matrices of the paid claims data and unpaid claims data to obtain standardized data matrices of the paid claims data and unpaid claims data respectively; construct an adaptive logic inference tree model based on the standardized data matrix of the paid claims data, and generate output vectors of the adaptive logic inference tree nodes; adjust the conditional threshold matrix based on the output vectors of the adaptive logic inference tree nodes to obtain the adjusted conditional threshold matrix;

[0008] S2: Based on the adjusted conditional threshold matrix and the output vector of the adaptive logic reasoning tree node, evolve the rule set matrix; based on the evolved rule set matrix and the output vector of the adaptive logic reasoning tree node, calculate the total claim amount; based on the total claim amount, evaluate the adaptive logic reasoning tree model, and optimize the adaptive logic reasoning tree model according to the evaluation results; based on the standardized data matrix of unclaimed data and the optimized adaptive logic reasoning tree model, calculate the final total claim amount.

[0009] Preferably, the S1 specifically includes:

[0010] The original data matrices of the paid claims data and the unpaid claims data are decentralized to obtain the decentralized matrices of the paid claims data and the unpaid claims data respectively; based on the decentralized matrices of the paid claims data and the unpaid claims data, the improved covariance matrices of the paid claims data and the unpaid claims data are calculated respectively.

[0011] Preferably, the S1 specifically includes:

[0012] Based on the improved covariance matrix of the claimed data and the unclaimed data, combined with the decentralized matrix of the claimed data and the unclaimed data, data standardization is performed to obtain the standardized data matrices of the claimed data and the unclaimed data, respectively.

[0013] Preferably, the S1 specifically includes:

[0014] Based on the standardized data matrix of the claims data, an adaptive logic inference tree model is constructed, and the output of the adaptive logic inference tree node is calculated. The specific formula is:

[0015]

[0016] in, is the output of the adaptive logic inference tree node; σ is the sigmoid activation function; tr(·) is the trace operation of the matrix; is an element-wise division operation; is an element-wise product operation; W is the weight matrix of the adaptive logic inference tree node; X is the standardized data matrix of the claims data; B is the bias matrix of the adaptive logic inference tree node; M is the standardized parameter matrix of the adaptive logic inference tree node; is the adaptive adjustment matrix of the adaptive logic reasoning tree node; T is the matrix transpose operation; det is the determinant operation;

[0017] The outputs of all adaptive logic reasoning tree nodes are aggregated to generate an output vector of the adaptive logic reasoning tree node.

[0018] Preferably, the S1 specifically includes:

[0019] Based on the output vector of the adaptive logic reasoning tree node, an adaptive condition node adjustment formula is introduced to adjust the condition threshold matrix in real time to obtain the adjusted condition threshold matrix.

[0020] Preferably, the S2 specifically includes:

[0021] Based on the adjusted conditional threshold matrix and the output vector of the adaptive logic reasoning tree node, the rule set matrix is ​​evolved by introducing determinant and logarithmic operations to obtain the evolved rule set matrix. The evolution formula of the rule set matrix is ​​as follows:

[0022]

[0023] Among them, R new is the evolved rule set matrix; R old is the current rule set matrix; C is the adjusted conditional threshold matrix; Y is the output vector of the adaptive logic inference tree node; R obs is the rule set matrix of the current path; C prev is the historical condition threshold matrix; Y prev is the historical output vector of the adaptive logic inference tree node; R opt is the rule set matrix of the historical optimal path; det(A opt ) is the determinant of the adjustment matrix of the historical optimal path; A opt is the adjustment matrix of the historical optimal path; ∈ is a constant to prevent the logarithm from being zero.

[0024] Preferably, the S2 specifically includes:

[0025] Based on the compensation amount formula stipulated in the policy terms, the basic compensation amount scalar is calculated; based on the basic compensation amount scalar, the basic compensation amount matrix is ​​constructed; based on the evolved rule set matrix and the output vector of the adaptive logic reasoning tree node, combined with the basic compensation amount matrix, the total claim amount is calculated.

[0026] Preferably, the S2 specifically includes:

[0027] Compare and evaluate the total claim amount with the actual claim amount of the claimed data in the original claim data, and calculate the error; when the error exceeds the preset error threshold, optimize the adaptive logic reasoning tree model; input the standardized data matrix of the unclaimed data into the optimized adaptive logic reasoning tree model to calculate the final total claim amount; compare the final total claim amount with the payment limit, payment conditions and customer account status in the policy terms. When the final total claim amount does not exceed the payment limit and the customer account status is valid, trigger payment and generate a claim report.

[0028] The smart contract-driven insurance claims automation system includes the following parts:

[0029] Data collection and preprocessing module, adaptive logic reasoning tree module, rule set evolution module, claim amount calculation module, model evaluation and optimization module, smart contract payment and report generation module;

[0030] The data collection and preprocessing module collects original claim data, which includes paid claim data and unpaid claim data, and outputs the actual claim amount in the paid claim data; constructs original data matrices of paid claim data and unpaid claim data based on the paid claim data and unpaid claim data in the original claim data; performs decentralized processing and data standardization processing on the original data matrices of paid claim data and unpaid claim data, and obtains and outputs standardized data matrices of paid claim data and unpaid claim data respectively; connects the data collection and preprocessing module with the adaptive logic reasoning tree module, the claim amount calculation module and the model evaluation and optimization module through data transmission;

[0031] The adaptive logic reasoning tree module introduces adaptive logic reasoning tree model parameters based on the standardized data matrix of the claims data, constructs an adaptive logic reasoning tree model, and generates output vectors of adaptive logic reasoning tree nodes; adjusts the condition threshold matrix based on the output vectors of the adaptive logic reasoning tree nodes; outputs the adaptive logic reasoning tree model parameters, the adjusted condition threshold matrix and the output vectors of the adaptive logic reasoning tree nodes; the adaptive logic reasoning tree module is connected to the rule set evolution module, the claims amount calculation module and the model evaluation and optimization module by means of data transmission;

[0032] The rule set evolution module evolves the current rule set matrix based on the adjusted condition threshold matrix and the output vector of the adaptive logic reasoning tree node, obtains and outputs the evolved rule set matrix; the rule set evolution module is connected with the claim amount calculation module and the model evaluation and optimization module by means of data transmission;

[0033] The claim amount calculation module obtains the evolved rule set matrix, the output vector of the adaptive logic reasoning tree node and the standardized data matrix of the claim data, calculates the total claim amount and outputs it; the claim amount calculation module is connected to the model evaluation and optimization module by data transmission;

[0034] The model evaluation and optimization module obtains the standardized data matrix of the total claim amount, the actual claim amount and the unclaimed data, calculates the error between the total claim amount and the actual claim amount, adjusts the adaptive logic reasoning tree model parameters based on the error and outputs it, and optimizes the adaptive logic reasoning tree model; inputs the standardized data matrix of the unclaimed data into the optimized adaptive logic reasoning tree model, calculates the final total claim amount and outputs it; the model evaluation and optimization module is connected to the smart contract payment and report generation module and the adaptive logic reasoning tree module by data transmission;

[0035] The smart contract payment and report generation module obtains the final claim total amount, compares the final claim total amount with the compensation limit, compensation conditions and customer account status in the policy terms, and triggers payment and generates a claim report when the final claim total amount does not exceed the compensation limit and the customer account status is valid.

[0036] The beneficial effects of the technical solution of the present invention are:

[0037] 1. By decentralizing and standardizing the original data matrix, combining it with an improved covariance matrix calculation formula, and introducing logarithmic and regularization terms, the ability to capture data extreme values ​​and matrix data correlation is enhanced, making the improved covariance matrix robust, and effectively reducing the impact of data anomalies on the claims calculation results, thereby ensuring the accuracy and consistency of subsequent claims decisions.

[0038] 2. The present invention introduces an adaptive logical reasoning tree model, which combines the structured decision-making advantages of traditional decision trees and the probabilistic output capabilities of logistic regression. At the same time, it introduces mathematical operations such as traces and determinants to capture the overall structural characteristics of claims data, so that it can flexibly respond to complex and changeable claims scenarios and improve the responsiveness and adaptability of the decision-making process.

[0039] 3. Based on smart contract technology, the present invention integrates a fully automated claims process from data collection, processing, decision-making to payment. Through the evolution of the rule set matrix, the final total claim amount is calculated, and payment is automatically executed on the premise that the policy terms and customer account status are valid. At the same time, the insurance claims system generates a claims report and automatically sends it to customers and relevant auditors, ensuring the transparency, fairness and efficiency of the claims process, and improving user experience and trust. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is a structural diagram of the insurance claims automation system driven by the smart contract of the present invention;

[0041] Figure 2 This is a flow chart of the smart contract-driven insurance claims automation method described in the present invention. DETAILED DESCRIPTION

[0042] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0043] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0044] The specific scheme of the insurance claims automation system driven by the smart contract provided by the present invention is described in detail below with reference to the accompanying drawings.

[0045] See attached Figure 1 , which shows a structure diagram of an insurance claims automation system driven by a smart contract provided by an embodiment of the present invention, the system includes the following parts:

[0046] Data collection and preprocessing module, adaptive logic reasoning tree module, rule set evolution module, claim amount calculation module, model evaluation and optimization module, smart contract payment and report generation module;

[0047] The data collection and preprocessing module collects original claim data, which includes paid claim data and unpaid claim data, and outputs the actual claim amount in the paid claim data; constructs original data matrices of paid claim data and unpaid claim data based on the paid claim data and unpaid claim data in the original claim data; performs decentralized processing and data standardization processing on the original data matrices of paid claim data and unpaid claim data, and obtains and outputs standardized data matrices of paid claim data and unpaid claim data respectively; connects the data collection and preprocessing module with the adaptive logic reasoning tree module, the claim amount calculation module and the model evaluation and optimization module through data transmission;

[0048] The adaptive logic reasoning tree module introduces adaptive logic reasoning tree model parameters based on the standardized data matrix of the claims data, constructs an adaptive logic reasoning tree model, and generates output vectors of adaptive logic reasoning tree nodes; adjusts the condition threshold matrix based on the output vectors of the adaptive logic reasoning tree nodes; outputs the adaptive logic reasoning tree model parameters, the adjusted condition threshold matrix and the output vectors of the adaptive logic reasoning tree nodes; the adaptive logic reasoning tree module is connected to the rule set evolution module, the claims amount calculation module and the model evaluation and optimization module by means of data transmission;

[0049] The rule set evolution module evolves the current rule set matrix based on the adjusted condition threshold matrix and the output vector of the adaptive logic reasoning tree node, obtains and outputs the evolved rule set matrix; the rule set evolution module is connected with the claim amount calculation module and the model evaluation and optimization module by means of data transmission;

[0050] The claim amount calculation module obtains the evolved rule set matrix, the output vector of the adaptive logic reasoning tree node and the standardized data matrix of the claim data, calculates the total claim amount and outputs it; the claim amount calculation module is connected to the model evaluation and optimization module by data transmission;

[0051] The model evaluation and optimization module obtains the standardized data matrix of the total claim amount, the actual claim amount and the unclaimed data, calculates the error between the total claim amount and the actual claim amount using the mean square error or the mean absolute error, adjusts the adaptive logic reasoning tree model parameters based on the error and outputs it, and optimizes the adaptive logic reasoning tree model; inputs the standardized data matrix of the unclaimed data into the optimized adaptive logic reasoning tree model, calculates the final total claim amount and outputs it; the model evaluation and optimization module is connected to the smart contract payment and report generation module and the adaptive logic reasoning tree module by means of data transmission;

[0052] The smart contract payment and report generation module obtains the final claim total amount, compares the final claim total amount with the compensation limit, compensation conditions and customer account status in the policy terms, ensures that the final claim total amount is within the specified range and the customer account status is valid. If the conditions are met, the smart contract automatically triggers payment and transfers the final claim total amount to the customer's bank account or e-wallet. At the same time, a claim report is generated and automatically sent to the customer and insurance company reviewers to ensure the transparency and fairness of the claims process.

[0053] See attached Figure 2 , which shows a flow chart of an insurance claim automation method driven by a smart contract provided by an embodiment of the present invention, the method comprising the following steps:

[0054] S1: Collect original claims data; construct original data matrices of the paid claims data and unpaid claims data respectively based on the paid claims data and unpaid claims data in the original claims data; decentralize and standardize the original data matrices of the paid claims data and unpaid claims data to obtain standardized data matrices of the paid claims data and unpaid claims data respectively; construct an adaptive logic inference tree model based on the standardized data matrix of the paid claims data, and generate output vectors of the adaptive logic inference tree nodes; adjust the conditional threshold matrix based on the output vectors of the adaptive logic inference tree nodes to obtain the adjusted conditional threshold matrix;

[0055] Original claim data is collected from information uploaded by users, third-party interfaces (such as hospitals, traffic management systems) or Internet of Things devices (such as vehicle recorders), and the original claim data includes claimed data and unclaimed data; based on the claimed data and unclaimed data, original data matrices of the claimed data and unclaimed data are constructed respectively, and the rows of the original data matrices of the claimed data and unclaimed data represent specific claim records (such as detailed claim data of a traffic accident), and the columns represent the characteristics of the claim records, that is, the attributes describing the accident or claim request (such as the accident time, location and loss amount of the traffic accident).

[0056] The original data matrix of the claims data is decentralized, and the mean of each column in the original data matrix of the claims data is subtracted to obtain the decentralized matrix of the claims data; based on the decentralized matrix of the claims data, the improved covariance matrix of the claims data is calculated; the improved covariance matrix calculation formula enhances the robustness of the improved covariance matrix to extreme data values ​​by introducing logarithms and regularization terms, and improves the ability to capture the correlation of matrix data.

[0057] The calculation formula of the improved covariance matrix is ​​as follows:

[0058]

[0059] Where E is the improved covariance matrix of the claims data; n is the number of rows of the original data matrix of the claims data; D is the original data matrix of the claims data; It is a decentralized matrix of claims data; is the covariance feature weight matrix, which is used to adjust the influence of each feature on the improved covariance matrix of the claims data, and is set according to the specific implementation scenario; α is the regularization coefficient, which is a scalar used to adjust the stability of the improved covariance matrix of the claims data; β is the adjustment coefficient, which is a scalar used to control the influence of the logarithmic term. The regularization coefficient and the adjustment coefficient are set according to the specific implementation scenario; I is the unit matrix, which is used to maintain the positive definiteness of the improved covariance matrix of the claims data; is the eigenvector matrix of the initial covariance matrix of the claims data, which is obtained by calculating the initial covariance matrix of the claims data using the traditional method and then performing eigendecomposition on the initial covariance matrix of the claims data; det is a determinant operation; is the determinant of the decentralized matrix of the claims data, reflecting the global variability of the decentralized matrix of the claims data; ∈ is a constant to prevent the logarithm from being zero; T is the matrix transpose operation.

[0060] Perform eigendecomposition on the improved covariance matrix of the claims data to obtain the eigenvector matrix and eigenvalue diagonal matrix of the improved covariance matrix of the claims data, and perform data standardization processing in combination with the decentralized matrix of the claims data to calculate and obtain the standardized data matrix of the claims data;

[0061] The specific calculation formula for the standardized data matrix of claims data is as follows:

[0062]

[0063] Among them, X is the standardized data matrix of the claims data; Person is the eigenvalue diagonal matrix of the improved covariance matrix of the claims data; U is the eigenvector matrix of the improved covariance matrix of the claims data.

[0064] The original data matrix of unclaimed data is also subjected to the above-mentioned decentralization and data standardization processing to obtain a standardized data matrix of unclaimed data; the standardized data matrix of claimed data is used to construct an adaptive logical reasoning tree model, and the standardized data matrix of unclaimed data is used for subsequent calculation of the final total claim amount.

[0065] The adaptive logical reasoning tree model is an improvement on the traditional decision tree and logistic regression model. It combines the structured decision-making advantages of the traditional decision tree and the probabilistic output capability of the logistic regression model. It dynamically adjusts the weight matrix, conditional threshold matrix and bias matrix to achieve real-time analysis and judgment of claims data. It also introduces mathematical operations such as trace and determinant to capture the overall structural characteristics of claims data, improve decision-making flexibility, and is suitable for handling complex claims scenarios.

[0066] The specific calculation formula for the output of the adaptive logic reasoning tree node is as follows:

[0067]

[0068] in, is the output of the adaptive logic inference tree node, indicating the degree of response of the adaptive logic inference tree node to a feature of the input data (such as accident time, location, loss amount, etc.). The larger the value of the output of the adaptive logic inference tree node, the higher the probability that the adaptive logic inference tree node believes that the input data meets a certain claim settlement rule or condition; σ is the sigmoid activation function; tr(·) is the trace operation of the matrix, taking the sum of the diagonal elements of the matrix; is an element-wise division operation; is an element-wise product operation; W is the weight matrix of the adaptive logic reasoning tree node, which indicates the sensitivity of the adaptive logic reasoning tree node to the standardized data matrix of the claims data; B is the bias matrix of the adaptive logic reasoning tree node, which is used for nonlinear adjustment; M is the standardized parameter matrix of the adaptive logic reasoning tree node, which is used to control the output range and stability of the adaptive logic reasoning tree node; It is the adaptive adjustment matrix of the adaptive logic reasoning tree node, which is used to adjust the weight of the input data, making the adaptive logic reasoning tree model more flexible in processing different types of claims data.

[0069] Specifically, the above-mentioned standardized parameter matrix, weight matrix, bias matrix and adaptive adjustment matrix are generated by selecting the initialization method (such as random initialization, uniform initialization, mean normalization initialization, etc.) according to the expert experience method at the beginning of training, and then adjusted and optimized by the back propagation algorithm and gradient descent method based on the claims data in the original claims data. The initialization method, back propagation algorithm and gradient descent method are all existing technologies and will not be elaborated here.

[0070] Aggregate the outputs of all nodes in the adaptive logic reasoning tree to generate an output vector Y of the adaptive logic reasoning tree node, with a length of n, which is the same as the number of claims records in the currently used standardized data matrix; Y represents the overall compliance level of each claim record under the evaluation of the entire adaptive logic reasoning tree model, and is used to adjust the rule set matrix to better predict and handle different claims scenarios and ensure the accuracy and effectiveness of claims decisions.

[0071] Based on the output vector of the adaptive logic reasoning tree node, the adaptive condition node adjustment formula is used to adjust the condition threshold matrix in real time to flexibly respond to different types of claims requests. The adaptive condition node adjustment formula calculates the difference between the output vector of the current adaptive logic reasoning tree node and the historical output vector of the adaptive logic reasoning tree node, and uses the maximum eigenvalue of the covariance matrix of the adjustment control matrix and the minimum eigenvalue of the covariance matrix of the standardized adjustment parameter matrix to control the adjustment range and avoid over- or under-adjustment.

[0072] The adaptive condition node adjustment formula is as follows:

[0073]

[0074] Among them, C is the adjusted conditional threshold matrix, which is used for the evolution of the subsequent rule set matrix; C prev is the historical condition threshold matrix, which comes from the own database and is initialized to an all-zero vector when it is first calculated; prev is the historical output vector of the adaptive logic inference tree node, that is, the previous output vector of the adaptive logic inference tree node, which is initialized to an all-zero vector when it is first calculated; max (V T V) is the covariance matrix V of the adjustment control matrix T The maximum eigenvalue of V reflects the dominant change direction; is the covariance matrix of the standardized adjustment parameter matrix The minimum eigenvalue of reflects the direction of secondary changes; V is the adjustment control matrix, which is used to control the adjustment rate and intensity of the condition threshold matrix; It is a standardized adjustment parameter matrix, which is used to control the scale consistency during the threshold adjustment process. During the first calculation, the adjustment control matrix and the standardized adjustment parameter matrix are generated by random initialization or uniform initialization, and are subsequently optimized by the gradient descent method, which will not be described in detail here.

[0075] S2: Based on the adjusted conditional threshold matrix and the output vector of the adaptive logic reasoning tree node, evolve the rule set matrix; based on the evolved rule set matrix and the output vector of the adaptive logic reasoning tree node, calculate the total claim amount; based on the total claim amount, evaluate the adaptive logic reasoning tree model, and optimize the adaptive logic reasoning tree model according to the evaluation results; based on the standardized data matrix of unclaimed data and the optimized adaptive logic reasoning tree model, calculate the final total claim amount.

[0076] According to the adjusted conditional threshold matrix and the output vector of the adaptive logic inference tree node, the rule set matrix is ​​evolved to obtain the evolved rule set matrix. The rule set matrix is ​​adjusted comprehensively by logarithm and determinant to enhance the sensitivity of the rule set matrix to data differences, so that the evolution of the rule set matrix has directionality and stability, thereby improving the adaptability of the rule set matrix.

[0077] The evolution formula of the rule set matrix is ​​as follows:

[0078]

[0079] Among them, R new is the evolved rule set matrix; R old is the current rule set matrix, which comes from the current rule set state; R obs is the rule set matrix of the current path, used to compare the current rules; R optis the rule set matrix of the historical optimal path, which is used to compare the current path with the historical optimal path; det(A opt ) is the determinant of the adjustment matrix of the historical optimal path, which is used to adjust the sensitivity of the rule set matrix; A opt It is the adjustment matrix of the historical optimal path.

[0080] Specifically, the current rule set matrix R old The initial value of is obtained from the insurance company's preset rules or the basic rule set set in the historical claims data; the rule set matrix R of the current path obs The initial value of is based on the current claim record (such as accident type, liability ratio, loss amount, etc.), and is generated using the rule mapping preset by the expert experience method (such as a logical mapping table based on accident type and severity); the rule set matrix R of the historical optimal path opt It is generated based on historical claims data (such as accident type, compensation amount, claim time, etc.) using supervised learning algorithms in existing technologies (such as regression analysis or decision trees) to train and reflect the rules for achieving optimal claims decisions in past cases; the adjustment matrix A of the historical optimal path opt It is generated based on historical claims data by using the gradient descent method to perform weighted adjustments on the features that affect claims.

[0081] Evolved rule set matrix R new , which integrates current data (adjusted condition threshold matrix, output vector of adaptive logic reasoning tree node, rule set matrix of current path) and historical data (historical condition threshold matrix, historical output vector of adaptive logic reasoning tree node, rule set matrix of historical optimal path, determinant of adjustment matrix in historical optimal path) to provide a basis for the final calculation of claim amount.

[0082] Finally, the total claim amount is calculated based on the evolved rule set matrix and the output vector of the adaptive logic reasoning tree node, combined with the basic compensation amount matrix. The basic compensation amount matrix is ​​generated by arranging the basic compensation amount scalars, which are calculated based on the compensation amount formula specified in the policy terms for each claim record request according to different condition combinations (such as accident type, loss amount, and liability ratio). The total claim amount calculation formula introduces global statistics (such as traces and determinants) to control the rationality and volatility of the total claim amount, making the payment more fair.

[0083] The total claim amount is calculated as follows:

[0084]

[0085] Among them, A is the total claim amount, which means the claim amount finally paid to the user, expressed in the form of a scalar; is the basic compensation amount matrix; is the trace of the basic compensation amount matrix, reflecting the sum of the basic compensation amounts stipulated in the policy terms, expressed in the form of a scalar; m is the number of columns in the currently used standardized data matrix, that is, the number of features contained in the claim records; n is the number of rows in the currently used standardized data matrix, that is, the total number of claim records; R new，ij is the specific element in the evolved rule set matrix, corresponding to the jth feature of the i-th claim record; Y j is the jth element in the output vector of the adaptive logic inference tree node; N is the number of elements with a value of 1 in the conditional satisfaction matrix, expressed in scalar form; is the element of the conditional matrix. For each feature, the claim threshold is set according to the expert experience method. The element X in the currently used standardized data matrix ij Compared with the predefined claim threshold of the jth feature, if X ij Greater than or equal to the claim threshold, set Less than the claim threshold, set

[0086] Next, the evaluation operation is performed. The total claim amount is compared with the actual claim amount of the claimed data in the original claim data, and the error is calculated using the mean square error (MSE) or the mean absolute error (MAE). The error threshold is set according to the specific implementation scenario. If the error is less than the error threshold, it indicates that the performance of the adaptive logic reasoning tree model has reached the expected standard and can be applied to actual claims. If it exceeds the error threshold, it is necessary to adjust the adaptive logic reasoning tree model parameters (such as weight matrix, bias matrix, condition threshold matrix, etc.) and optimize the adaptive logic reasoning tree model until the error is less than the error threshold.

[0087] The standardized data matrix of unclaimed data is input into the optimized adaptive logic inference tree model to calculate the final total claim amount. The final total claim amount is compared with the compensation limit, compensation conditions and customer account status in the policy terms to ensure that the final total claim amount does not exceed the compensation limit and the customer account status is valid. If an abnormal situation occurs, such as incomplete data or non-compliance with the policy regulations, the smart contract will automatically suspend the payment process and notify the relevant parties for further review and processing.

[0088] If the conditions are met, the smart contract automatically triggers payment and transfers the final claim amount to the customer's bank account or e-wallet. At the same time, a claim report is generated, which includes the calculation process of the final claim amount, the data involved, the rule set used, and the final claim amount. The claim report is automatically sent to the customer and the auditors of the relevant insurance company to facilitate transparent communication and ensure the fairness and justice of the claims process.

[0089] In summary, the smart contract-driven insurance claims automation system has been completed.

[0090] The order of the embodiments of the invention is for description only and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0091] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0092] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. An automated insurance claims settlement method driven by smart contracts, characterized in that: The following steps are involved: S1: Collect original claims data; Based on the paid claims data and unpaid claims data in the original claims data, the original data matrices of paid claims data and unpaid claims data are constructed respectively; The original data matrices of the paid claims data and the unpaid claims data are decentralized and standardized to obtain the standardized data matrices of the paid claims data and the unpaid claims data respectively; Based on the standardized data matrix of the claims data, an adaptive logic inference tree model is constructed, and the output vectors of the adaptive logic inference tree nodes are generated; Based on the output vector of the adaptive logic reasoning tree node, the conditional threshold matrix is ​​adjusted to obtain an adjusted conditional threshold matrix; S2: Based on the adjusted condition threshold matrix and the output vector of the adaptive logic reasoning tree node, the rule set matrix is ​​evolved; based on the evolved rule set matrix and the output vector of the adaptive logic reasoning tree node, the total claim amount is calculated; Based on the total amount of claims, the adaptive logic reasoning tree model is evaluated, and the adaptive logic reasoning tree model is optimized according to the evaluation results; The final total claim amount is calculated based on the standardized data matrix of unclaimed data and the optimized adaptive logic reasoning tree model.

2. The smart contract driven insurance claims automation method according to claim 1, characterized in that: The S1 specifically includes: The original data matrices of the paid claims data and the unpaid claims data are decentralized to obtain the decentralized matrices of the paid claims data and the unpaid claims data respectively; based on the decentralized matrices of the paid claims data and the unpaid claims data, the improved covariance matrices of the paid claims data and the unpaid claims data are calculated respectively.

3. The smart contract driven insurance claims automation method according to claim 2, characterized in that: The S1 specifically includes: Based on the improved covariance matrix of the claimed data and the unclaimed data, combined with the decentralized matrix of the claimed data and the unclaimed data, data standardization is performed to obtain the standardized data matrices of the claimed data and the unclaimed data, respectively.

4. The smart contract driven insurance claims automation method according to claim 3, characterized in that: The S1 specifically includes: Based on the standardized data matrix of the claims data, an adaptive logic inference tree model is constructed, and the output of the adaptive logic inference tree node is calculated. The specific formula is: in, is the output of the adaptive logic inference tree node; σ is the sigmoid activation function; tr(·) is the trace operation of the matrix; is an element-wise division operation; is an element-wise product operation; W is the weight matrix of the adaptive logic inference tree node; X is the standardized data matrix of the claims data; B is the bias matrix of the adaptive logic inference tree node; M is the standardized parameter matrix of the adaptive logic inference tree node; is the adaptive adjustment matrix of the adaptive logic reasoning tree node; T is the matrix transpose operation; det is the determinant operation; The outputs of all adaptive logic reasoning tree nodes are aggregated to generate an output vector of the adaptive logic reasoning tree node.

5. The smart contract driven insurance claims automation method according to claim 4, characterized in that: The S1 specifically includes: Based on the output vector of the adaptive logic reasoning tree node, an adaptive condition node adjustment formula is introduced to adjust the condition threshold matrix in real time to obtain the adjusted condition threshold matrix.

6. The smart contract driven insurance claims automation method according to claim 1, characterized in that: The S2 specifically includes: Based on the adjusted conditional threshold matrix and the output vector of the adaptive logic reasoning tree node, the rule set matrix is ​​evolved by introducing determinant and logarithmic operations to obtain the evolved rule set matrix. The evolution formula of the rule set matrix is ​​as follows: Among them, R new is the evolved rule set matrix; R old is the current rule set matrix; C is the adjusted conditional threshold matrix; Y is the output vector of the adaptive logic inference tree node; R obs is the rule set matrix of the current path; C prev is the historical condition threshold matrix; Y prev is the historical output vector of the adaptive logic inference tree node; R opt is the rule set matrix of the historical optimal path; det(A opt ) is the determinant of the adjustment matrix of the historical optimal path; A opt is the adjustment matrix of the historical optimal path; ∈ is a constant to prevent the logarithm from being zero.

7. The smart contract driven insurance claims automation method according to claim 6, characterized in that: The S2 specifically includes: Based on the compensation amount formula stipulated in the policy terms, the basic compensation amount scalar is calculated; based on the basic compensation amount scalar, the basic compensation amount matrix is ​​constructed; based on the evolved rule set matrix and the output vector of the adaptive logic reasoning tree node, combined with the basic compensation amount matrix, the total claim amount is calculated.

8. The smart contract driven insurance claims automation method according to claim 7, characterized in that: The S2 specifically includes: Compare and evaluate the total claim amount with the actual claim amount of the claimed data in the original claim data, and calculate the error; when the error exceeds the preset error threshold, optimize the adaptive logic reasoning tree model; input the standardized data matrix of the unclaimed data into the optimized adaptive logic reasoning tree model to calculate the final total claim amount; compare the final total claim amount with the payment limit, payment conditions and customer account status in the policy terms. When the final total claim amount does not exceed the payment limit and the customer account status is valid, trigger payment and generate a claim report.

9. An insurance claims automation system driven by a smart contract, applied to the insurance claims automation method driven by a smart contract as claimed in claim 1, characterized in that: Includes the following parts: Data collection and preprocessing module, adaptive logic reasoning tree module, rule set evolution module, claim amount calculation module, model evaluation and optimization module, smart contract payment and report generation module; The data collection and preprocessing module collects original claim data, which includes paid claim data and unpaid claim data, and outputs the actual claim amount in the paid claim data; Based on the paid claims data and unpaid claims data in the original claims data, the original data matrices of paid claims data and unpaid claims data are constructed respectively; The original data matrices of the paid claims data and the unpaid claims data are decentralized and standardized to obtain and output the standardized data matrices of the paid claims data and the unpaid claims data respectively; the data collection and preprocessing module is connected with the adaptive logic reasoning tree module, the claim amount calculation module and the model evaluation and optimization module through data transmission; The adaptive logic reasoning tree module introduces adaptive logic reasoning tree model parameters based on the standardized data matrix of the claims data, constructs an adaptive logic reasoning tree model, and generates output vectors of adaptive logic reasoning tree nodes; Adjusting the conditional threshold matrix based on the output vector of the adaptive logic reasoning tree node; Outputting adaptive logic inference tree model parameters, adjusted conditional threshold matrix and output vector of adaptive logic inference tree nodes; The adaptive logic reasoning tree module is connected with the rule set evolution module, the claim amount calculation module and the model evaluation and optimization module through data transmission; The rule set evolution module evolves the current rule set matrix based on the adjusted condition threshold matrix and the output vector of the adaptive logic reasoning tree node, obtains and outputs the evolved rule set matrix; the rule set evolution module is connected with the claim amount calculation module and the model evaluation and optimization module by means of data transmission; The claim amount calculation module obtains the evolved rule set matrix, the output vector of the adaptive logic reasoning tree node and the standardized data matrix of the claim data, calculates the total claim amount and outputs it; the claim amount calculation module is connected to the model evaluation and optimization module by data transmission; The model evaluation and optimization module obtains a standardized data matrix of the total claim amount, the actual claim amount, and the unclaimed data, calculates the error between the total claim amount and the actual claim amount, adjusts the adaptive logic inference tree model parameters based on the error and outputs them, and optimizes the adaptive logic inference tree model; The standardized data matrix of unclaimed data is input into the optimized adaptive logic reasoning tree model, and the final total claim amount is calculated and output; the model evaluation and optimization module is connected with the smart contract payment and report generation module and the adaptive logic reasoning tree module through data transmission; The smart contract payment and report generation module obtains the final claim total amount, compares the final claim total amount with the compensation limit, compensation conditions and customer account status in the policy terms, and triggers payment and generates a claim report when the final claim total amount does not exceed the compensation limit and the customer account status is valid.

Images