Insurance product recommendation method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention provides an insurance product recommendation method and device, electronic equipment and a storage medium, belongs to the technical field of artificial intelligence, and is suitable for the fields of financial science and technology and medical treatment. The method comprises the following steps: predicting a health risk level of a target object for an original medical record, and constructing an initial object state feature; and predicting the initial object state characteristics to obtain candidate product recommendation schemes, wherein the candidate product recommendation schemes comprise candidate insurance products and corresponding basic reward scores. And in response to a selection operation of the target object on the candidate insurance product, updating the initial object state feature to obtain a target object state feature. And calculating an execution reward score according to the initial object state feature and the target object state feature. And generating a target product recommendation scheme according to the recommendation scheme prediction model, the target object state characteristics, the basic reward score and the execution reward score, and recommending to a target user based on the scheme. According to the embodiment of the invention, the recommendation accuracy of insurance products is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, and is suitable for the fields of financial technology and medicine, and in particular relates to an insurance product recommendation method and device, an electronic device and a storage medium. BACKGROUND

[0002] At present, the recommendation of insurance products still mainly depends on users to determine suitable insurance products through questionnaire filling, historical medical records or static health information (such as age, gender, occupation, etc.). This recommendation method based on static data ignores the dynamic changes of health status and risk level, for example, the rehabilitation process of patients, the recovery of diseases and the changes of health behaviors, etc., which ultimately leads to the fact that the recommended insurance products may not truly meet the individual needs of users. Therefore, how to accurately recommend insurance products to users has become a technical problem to be solved. SUMMARY

[0003] The main purpose of the embodiments of the present application is to propose an insurance product recommendation method and device, an electronic device and a storage medium, which aims to improve the recommendation accuracy of insurance products.

[0004] To achieve the above-mentioned purpose, a first aspect of the embodiments of the present application proposes an insurance product recommendation method, which comprises:

[0005] Obtaining the original medical record of a target object, and performing risk prediction on the original medical record to obtain the health risk level of the target object;

[0006] Constructing the initial object state feature of the target object according to the original medical record and the health risk level;

[0007] Generating a candidate product recommendation scheme through a preset recommendation scheme prediction model for the initial object state feature, wherein the candidate product recommendation scheme comprises a candidate insurance product recommended to the target object and a basic reward score matched to the candidate insurance product;

[0008] In response to a selection operation of the target object on the candidate insurance product, updating the initial object state feature of the target object to obtain a target object state feature, and performing reward calculation according to the initial object state feature and the target object state feature to obtain an execution reward score of the target object;

[0009] Generating a target product recommendation scheme according to the recommendation scheme prediction model, the target object state feature, the basic reward score and the execution reward score, and recommending an insurance product to the target object based on the target product recommendation scheme.

[0010] In some embodiments, the original medical record includes original sub-data of a plurality of data types, and the initial object state feature of the target object is updated to obtain a target object state feature in response to the selection operation of the target object on the candidate insurance product, including:

[0011] In response to the selection operation of the target object on the candidate insurance product, a target insurance product is determined from at least two candidate insurance products;

[0012] Based on the payment period of the target insurance product, a target type is selected from a plurality of data types;

[0013] Obtain the target sub-data of the target object that matches the target type in the next payment period;

[0014] Update the initial object state feature based on the target sub-data of the target type to obtain the target object state feature.

[0015] In some embodiments, the recommendation scheme generation is performed according to the recommendation scheme prediction model, the target object state feature, the basic reward score and the execution reward score to obtain a target product recommendation scheme, including:

[0016] The basic reward score and the execution reward score are superimposed to obtain a target reward score;

[0017] According to each preset reference score of the original insurance product and the target reward score, a reference insurance product is selected from at least two original insurance products, wherein the target reward score is greater than or equal to the reference score;

[0018] The target object state feature is predicted by the recommendation scheme prediction model to obtain an intermediate insurance product, and the target product recommendation scheme is determined based on the intermediate insurance product and the reference insurance product.

[0019] In some embodiments, before the candidate product recommendation scheme is generated by the preset recommendation scheme prediction model based on the initial object state feature, the method further includes training the recommendation scheme prediction model, specifically including:

[0020] Obtain the first sample state feature of the preset sample object;

[0021] The first sample state feature is predicted by a preset initial prediction model to obtain an initial product recommendation scheme and an execution probability of the initial product recommendation scheme;

[0022] simulate updating the first sample state feature according to the initial product recommendation scheme to obtain a second sample state feature;

[0023] performing rationality evaluation on the initial product recommendation scheme to obtain a rationality score;

[0024] performing state evaluation on the first sample state feature to obtain a first state score, and performing state evaluation on the second sample state feature to obtain a second state score;

[0025] calculating a target loss value according to the execution probability, the rationality score, the first state score and the second state score, and performing parameter updating on the initial prediction model according to the target loss value to obtain the recommendation scheme prediction model.

[0026] In some embodiments, the calculating a target loss value according to the execution probability, the rationality score, the first state score and the second state score comprises:

[0027] performing reward merging calculation according to the rationality score, the first state score and the second state score to obtain state reward data;

[0028] performing integration calculation according to the execution probability and the state reward data to obtain the target loss value.

[0029] In some embodiments, the performing reward calculation according to the initial object state feature and the target object state feature to obtain an execution reward score of the target object comprises:

[0030] extracting a target medical record and a target risk level from the target object state feature;

[0031] performing health state evaluation on the target medical record to obtain a health state score;

[0032] performing risk trend evaluation according to the health risk level and the target risk level to obtain a risk trend score;

[0033] performing weighted calculation according to the health state score and the risk trend score to obtain the execution reward score.

[0034] In some embodiments, the performing risk prediction on the original medical record to obtain a health risk level of the target object comprises:

[0035] classifying the original medical record according to a data type of the original medical record to obtain at least two original sub-data;

[0036] performing time sequence feature extraction on each of the original sub-data to obtain data time sequence features;

[0037] performing image feature extraction on medical images in the original medical record to obtain medical image features;

[0038] performing risk level division on the data time sequence features and the medical image features through a preset risk prediction model to obtain a health risk level of the target object.

[0039] To achieve the above object, a second aspect of the embodiment of the present application provides an insurance product recommendation device, which comprises:

[0040] an acquisition module configured to acquire original medical records of a target object, and perform risk prediction on the original medical records to obtain a health risk level of the target object;

[0041] a state feature construction module configured to construct initial object state features of the target object according to the original medical records and the health risk level;

[0042] a first recommendation module configured to perform recommendation scheme generation on the initial object state features through a preset recommendation scheme prediction model to obtain a candidate product recommendation scheme; wherein the candidate product recommendation scheme comprises a candidate insurance product recommended to the target object, and a basic reward score matched to the candidate insurance product;

[0043] a state feature updating module configured to update the initial object state features of the target object in response to a selection operation of the target object on the candidate insurance product to obtain target object state features, and perform reward calculation according to the initial object state features and the target object state features to obtain an execution reward score of the target object;

[0044] a second recommendation module configured to perform recommendation scheme generation according to the recommendation scheme prediction model, the target object state features, the basic reward score and the execution reward score to obtain a target product recommendation scheme, and perform insurance product recommendation to the target object based on the target product recommendation scheme.

[0045] To achieve the above object, a third aspect of the embodiment of the present application provides an electronic device, which comprises a memory and a processor, the memory stores a computer program, and the processor implements the method of the first aspect when executing the computer program.

[0046] To achieve the above object, a fourth aspect of the embodiment of the present application provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the method in the first aspect.

[0047] The insurance product recommendation method and device, the electronic device and the storage medium provided by the present application can obtain the original medical record of a target object, predict the health risk level of the target object according to the record, and construct the initial object state feature of the target object according to the original medical record and the health risk level. Then, the initial object state feature is subjected to recommendation scheme generation according to a recommendation scheme prediction model to obtain a candidate product recommendation scheme. When the user determines to select the candidate insurance product, the initial object state feature is updated to obtain a target object state feature, so as to realize dynamic response to the change of the target object condition. The initial object state feature and the target object state feature are used to calculate an execution reward score. Finally, the recommendation scheme prediction model, the target object state feature, the basic reward score and the execution reward score are used to generate a target product recommendation scheme. The method provided by the embodiment of the present application effectively solves the problem of insufficient personalization and inaccurate recommendation caused by the failure to fully consider dynamic factors in the traditional insurance product recommendation, so that the recommendation of the insurance product is more accurate. BRIEF DESCRIPTION OF DRAWINGS

[0048] Figure 1 is a flowchart of the insurance product recommendation method provided by the embodiment of the present application;

[0049] Figure 2 is a flowchart of step S101 in Figure 1 ;

[0050] Figure 3 is another flowchart of the insurance product recommendation method provided by the embodiment of the present application;

[0051] Figure 4 is a flowchart of step S104 in Figure 1 ;

[0052] Figure 5 is another flowchart of step S104 in Figure 1 ;

[0053] Figure 6 is a flowchart of step S105 in Figure 1 ;

[0054] Figure 7 is a structural schematic diagram of the insurance product recommendation device provided by the embodiment of the present application;

[0055] Figure 8 is a hardware structural schematic diagram of the electronic device provided by the embodiment of the present application. DETAILED DESCRIPTION

[0056] In order to make the purposes, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and not intended to limit the present application.

[0057] It should be noted that although the functional modules are divided in the device schematic diagram, and the logical sequence is shown in the flowchart, in some cases, the steps shown or described can be performed in a manner different from the module division in the device or the sequence in the flowchart. The terms "first", "second", and the like in the specification and claims and the above-described drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0058] Unless otherwise defined, 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 application belongs. The terminology used in this application is only for the purpose of describing the embodiments of the present application and is not intended to limit the present application.

[0059] First, the meanings of several terms involved in the present application are analyzed:

[0060] Artificial intelligence (AI): is a new technical science that studies, develops and applies systems for simulating, extending and expanding human intelligence; artificial intelligence is a branch of computer science, artificial intelligence aims to understand the essence of intelligence and produce a new intelligent machine that can react in a similar way to human intelligence. The field of research includes robots, language recognition, image recognition, natural language processing and expert systems. Artificial intelligence can simulate the information process of human consciousness and thinking. Artificial intelligence is also the theory, method, technology and application system of using digital computers or digital computer controlled machines to simulate, extend and expand human intelligence, to perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0061] Blockchain technology: is a distributed ledger technology that stores data in blocks in sequence through encryption algorithms, and connects these blocks through encryption hash functions to form a chain, forming a decentralized, transparent, secure and tamper-proof network. The core of blockchain is its decentralized nature, data is not stored on a centralized server, but distributed among all nodes in the network, which makes blockchain highly reliable and tamper-proof. In addition, blockchain ensures the authenticity and consistency of data through consensus mechanism to ensure that all nodes agree on the verification and addition of new blocks, thereby ensuring the authenticity and consistency of data.

[0062] At present, the recommendation of insurance products still mainly depends on that the user determines the suitable insurance product through questionnaire filling, historical medical records or static health information (such as age, gender, occupation, etc.). This recommendation method based on static data ignores the dynamic changes of health status and risk level, for example, the rehabilitation process of the patient, the recovery of the disease and the change of health behavior, etc., which finally leads to that the recommended insurance product may not truly meet the individual needs of the user and the inaccuracy of the premium adjustment. Therefore, how to accurately recommend the insurance product to the user has become a technical problem to be solved.

[0063] Based on this, the embodiment of the present application provides an insurance product recommendation method and device, an electronic device and a storage medium, aiming to improve the recommendation accuracy of the insurance product.

[0064] The insurance product recommendation method and device, the electronic device and the storage medium provided by the embodiment of the present application are specifically explained by the following embodiment, first, the insurance product recommendation method in the embodiment of the present application is described.

[0065] The embodiment of the present application can acquire and process related data based on artificial intelligence technology. Wherein, artificial intelligence (AI) is to use digital computer or digital computer controlled machine to simulate, extend and expand human intelligence, perceive environment, acquire knowledge and use knowledge to obtain the best results.

[0066] The basic technology of artificial intelligence generally includes technologies such as sensors, special artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction system, mechatronics, etc. The software technology of artificial intelligence mainly includes computer vision technology, robot technology, biometric technology, speech processing technology, natural language processing technology and machine learning / deep learning, etc.

[0067] The insurance product recommendation method provided by the embodiment of the present application relates to the field of artificial intelligence technology. The insurance product recommendation method provided by the embodiment of the present application can be applied in a terminal, can also be applied in a server end, and can also be software running in a terminal or a server end. In some embodiments, the terminal can be a smart phone, a tablet computer, a notebook computer, a desktop computer, etc.; the server end can be configured as an independent physical server, can also be configured as a server cluster or a distributed system composed of multiple physical servers, can also be configured as a cloud server providing cloud service, cloud database, cloud computing, cloud function, cloud storage, network service, cloud communication, middleware service, domain name service, security service, CDN and basic cloud computing services such as big data and artificial intelligence platform; the software can be an application for implementing the insurance product recommendation method, etc., but is not limited to the above forms.

[0068] The application is operable in a variety of general purpose or special purpose computing system environments or configurations. Examples of computing systems, environments, and / or configurations that can be suitable for use with the application include personal computers, server computers, handheld or laptop devices, tablet-type devices, multiprocessor systems, microprocessor-based systems, set top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments that include any of the above systems or devices, and the like. The application can be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, and the like, that perform particular tasks or implement particular abstract data types. The application can also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules can be located in both local and remote computer storage media including memory storage devices.

[0069] It should be noted that in each specific embodiment of the present application, when it is necessary to process relevant data related to the identity or characteristics of the user according to user information, user behavior data, user history data, and user location information, the user's permission or consent will be obtained first, and the collection, use, and processing of these data will comply with relevant laws, regulations, and standards. In addition, when the embodiments of the present application need to obtain sensitive personal information of the user, the separate permission or separate consent of the user will be obtained through a pop-up window or a jump to a confirmation page, and after obtaining the separate permission or separate consent of the user, the necessary user-related data for enabling the embodiments of the present application to normally operate will be obtained.

[0070] Figure 1 is an optional flowchart of the insurance product recommendation method provided by the embodiments of the present application, Figure 1 The method in can include but is not limited to steps S101 to S105.

[0071] In step S101, the original medical record of the target object is obtained, and the original medical record is subjected to risk prediction to obtain the health risk level of the target object.

[0072] In step S102, the initial object state feature of the target object is constructed according to the original medical record and the health risk level.

[0073] In step S103, the initial object state feature is subjected to candidate product recommendation scheme generation by a preset recommendation scheme prediction model to obtain a candidate product recommendation scheme; wherein the candidate product recommendation scheme includes a candidate insurance product recommended to the target object, and a basic reward score matched to the candidate insurance product.

[0074] Step S104, in response to the target object's selection operation on the candidate insurance product, updating the initial object state feature of the target object to obtain the target object state feature, and calculating the reward based on the initial object state feature and the target object state feature to obtain the execution reward score of the target object.

[0075] Step S105, generating the recommendation scheme based on the recommendation scheme prediction model, the target object state feature, the basic reward score and the execution reward score to obtain the target product recommendation scheme, and recommending the insurance product to the target object based on the target product recommendation scheme.

[0076] The steps S101 to S105 shown in the embodiments of the present application, by obtaining the original medical record of the target object, predicting the health risk level of the target object based on the record, and constructing the initial object state feature of the target object based on the original medical record and the health risk level. Then, according to the recommendation scheme prediction model, the initial object state feature is used to generate the recommendation scheme to obtain the candidate product recommendation scheme. When the user determines to select the candidate insurance product, the initial object state feature is updated to obtain the target object state feature, so as to realize the dynamic response to the change of the target object condition. The initial object state feature and the target object state feature are calculated to obtain the execution reward score. Finally, the recommendation scheme prediction model, the target object state feature, the basic reward score and the execution reward score are used to generate the recommendation scheme to obtain the target product recommendation scheme. The method of the embodiments of the present application effectively solves the problem of insufficient personalization and inaccurate recommendation caused by not fully considering dynamic factors in traditional insurance product recommendation, so that the recommendation of insurance product is more accurate.

[0077] In step S101 of some embodiments, the target object refers to the insured person. Specifically, the target object is an individual who needs to receive insurance product recommendation.

[0078] The original medical record can be obtained from a medical institution or other authorized channels to obtain the health record of the target object (i.e. the insured person), covering the past medical history, disease record, physical examination result, drug use, etc. Such data can be obtained through electronic health record system (EHR), or through cooperation with hospitals, clinics, physical examination institutions. It can also be obtained by installing wearable devices (such as smart bracelet, electrocardiogram monitor, etc.).

[0079] In addition, in this embodiment, blockchain technology is also used to manage the collected data. Through the tamper-proof and traceable characteristics of the blockchain, the security of the original medical record is ensured.

[0080] In the embodiment, the health risk level is used to indicate the current health status of the target object, which can include, but is not limited to, "low risk", "medium risk", "high risk", etc.

[0081] Further, for the risk prediction process of the original medical record, please refer to Figure 2 , step S101 can include, but is not limited to, steps S201 to S204:

[0082] Step S201, classifying the original medical record according to the data type of the original medical record, obtaining at least two original sub-data.

[0083] Step S202, time series feature extraction is performed on each original sub-data to obtain data time series features.

[0084] Step S203, image feature extraction is performed on the medical image in the original medical record to obtain medical image features.

[0085] Step S204, the data time series features and the medical image features are divided into risk levels by a preset risk prediction model to obtain the health risk level of the target object.

[0086] In step S201 of some embodiments, the original medical record includes original sub-data of multiple data types. Exemplarily, the data types of the original medical record can include heart rate, blood oxygen, body temperature, MRI, X-ray image data, gait analysis, joint activity, etc., but are not limited thereto. The specific data corresponding to each data type is the original sub-data.

[0087] In step S202 of some embodiments, methods such as sliding window, Fourier transform, wavelet transform, etc. can be used for time series analysis of data to extract time series features such as mean value, variance, maximum value, minimum value, rate of rise and fall, etc. The features obtained after time series feature analysis are the data time series features.

[0088] In step S203 of some embodiments, image processing techniques such as denoising, contrast enhancement, etc. can be used for pre-processing of the medical image. Then, a deep learning method such as convolutional neural network (CNN) is used to extract features from the medical image. Through multi-layer convolution and pooling operations, the neural network can identify important features in the image, such as lesion location, morphological features, tissue density, etc., thereby extracting the medical image features.

[0089] In step S204 of some embodiments, the risk prediction model is a machine learning model trained based on historical data, which can be a logistic regression model, a support vector machine (SVM), a decision tree, a random forest, a neural network, etc., without limitation. The risk prediction model takes the extracted time-series features and image features as input data, performs risk assessment, and divides the health risk level of the target object into different levels, such as "low risk", "medium risk", "high risk", etc.

[0090] The steps S201 to S204 shown in the embodiments of the present application can effectively separate and process multiple types of data by classifying the original medical records according to data types, thereby providing clearer and more structured inputs for subsequent risk prediction. Each category of data is subjected to time-series feature extraction and image feature extraction, ensuring the full use of dynamic health data and static data. Meanwhile, through the preset risk prediction model, the time-series features and image features are jointly analyzed to obtain the health risk level of the target object, thereby accurately and quickly determining the real-time health status of the target object.

[0091] In step S102 of some embodiments, an agent is constructed for each target object, and the state feature of the agent is the initial object state feature. The initial object state feature generally includes factors such as the health status, risk level, age, gender, and medical history of the target object, and is usually expressed as a structured numerical vector to form a state representation for subsequent analysis and processing. For example, the initial state feature of the target object can be represented as a vector: [age = 45, gender = male, risk level = medium risk, blood sugar = normal, blood pressure = slightly high].

[0092] Before step S103 of some embodiments, the insurance product recommendation method further includes pre-training a recommendation scheme prediction model, which is used to predict a recommendation scheme for the target object and can provide multiple insurance products for the target object to choose from.

[0093] Please refer to Figure 3 In some embodiments, the recommendation scheme prediction model is trained based on an Actor-Critic reinforcement learning framework, and its training process can include but is not limited to steps S301 to S306:

[0094] Step S301: Obtain a first sample state feature of a preset sample object.

[0095] Step S302: Perform recommendation scheme prediction on the first sample state feature through a preset initial prediction model to obtain an initial product recommendation scheme and an execution probability of the initial product recommendation scheme.

[0096] Step S303, simulating updating the first sample state feature according to the initial product recommendation scheme to obtain a second sample state feature.

[0097] Step S304, performing rationality evaluation on the initial product recommendation scheme to obtain a rationality score.

[0098] Step S305, performing state evaluation on the first sample state feature to obtain a first state score, and performing state evaluation on the second sample state feature to obtain a second state score.

[0099] Step S306, calculating a target loss value according to the execution probability, the rationality score, the first state score, and the second state score, and performing parameter updating on the initial prediction model according to the target loss value to obtain a recommendation scheme prediction model.

[0100] In step S301 of some embodiments, the preset sample object refers to a user object selected in the model training process. The first sample state feature is a data feature composed of the medical record and the health risk level of the preset sample object. For example, the first sample state feature can be [age = 40, gender = male, risk level = low risk, blood sugar = slightly high].

[0101] In step S302 of some embodiments, the execution probability is used to represent the possibility of the execution of each initial product recommendation scheme under the current object state. The execution probability is usually calculated by the output layer of the initial prediction model, and each strategy score is normalized to a probability distribution by using a Softmax function. The initial product recommendation scheme refers to the insurance product combination that the initial prediction model considers to be the most optimal and possible to be selected under the current first sample state feature.

[0102] In step S303 of some embodiments, based on the initial product recommendation scheme, the first sample state feature is updated through the feedback of the simulation environment, so as to obtain the second sample state feature. Usually, a policy updating algorithm in reinforcement learning is used to predict the change of the health state of the target object after selecting a product through deduction and assumption of the health data of the target object. For example, if the target object selects the insurance product A, the health state of the target object may change, and the blood sugar level of the target object may improve from “slightly high” to “normal”. At this time, the generated second sample state feature can be [age = 40, gender = male, risk level = low risk, blood sugar = normal].

[0103] In step S304 of some embodiments, the rationality evaluation can be implemented by using a rule engine or by performing rationalization evaluation on the recommended insurance product through a deep learning model. The evaluation indicators can include whether the insurance coverage, the insurance amount of the product, and the health risk of the target object match the economic ability of the target object.

[0104] In step S305 of some embodiments, the pre-trained deep learning model can be used as a state evaluation model, which is trained based on clinical guidelines or expert knowledge base. The state evaluation model is used to evaluate the first sample state feature and the second sample state feature respectively. The first state score can represent the health evaluation result of the target object before selecting the product, and the second state score can represent the health evaluation result of the target object after selecting the product. The evaluation process can help measure the health changes of the target object after selecting the recommended product, and adjust the recommendation scheme accordingly.

[0105] In step S306 of some embodiments, the reward merging calculation is performed according to the rationality score, the first state score and the second state score, to obtain state reward data. In this embodiment, a discount factor is also needed, which can be 0.5. The target loss value is calculated according to the execution probability and the state reward data. Further, the calculation process of the target loss value can refer to the following analysis formula:

[0106]

[0107] wherein, L actor (θ) represents the target loss value, π θ (a|s) represents the execution of the initial product recommendation scheme under the first sample state feature, logπ θ (a|s) represents the execution probability of the initial product recommendation scheme. θ represents the model parameter of the initial prediction model. R represents the reward score. V(s) represents the first state score, V(s') represents the second state score, and γ represents the discount factor. (R+γV(s')-V(s)) represents the state reward data.

[0108] Then, the parameter update is performed by the gradient descent algorithm until the target loss value is minimized.

[0109] The steps S301 to S306 shown in the embodiments of the present application can make the insurance product recommendation method accurately generate the candidate product recommendation scheme for the target object based on the health data, risk level and other personalized information of the target object, by pre-training the recommendation scheme prediction model.

[0110] In step S103 of some embodiments, the recommendation scheme prediction model generates the candidate product recommendation scheme by processing the initial object state feature. The candidate recommendation scheme includes a plurality of possible insurance products, and each insurance product is assigned a basic reward score. The basic reward score can be used as a basis for exchanging other insurance products in the future. The specific value of the basic reward score is determined according to the operation rules of the business platform, and is related to the cost of the insurance product.

[0111] Exemplarily, assuming that the target object is a 45-year-old male with overweight, mild hypertension, and a medium health risk level, the recommendation solution prediction model can recommend a medium-risk health insurance product based on his initial state features (such as age, health status, medical history, etc.), and assign a basic reward score, for example, 80, to the product.

[0112] In step S104 of some embodiments, if the target object does not make a selection operation on the candidate insurance product, it means that the target object gives up to purchase the relevant insurance product, at this time, the method terminates, and waits for the target user to start the next round of product recommendation.

[0113] For the specific process of updating the initial object state features of the target object, please refer to Figure 4 In some embodiments, step S104 can include, but is not limited to, steps S401 to S404:

[0114] Step S401, in response to the selection operation of the target object on the candidate insurance product, determining the target insurance product from the at least two candidate insurance products.

[0115] Step S402, filtering the target type from a plurality of data types based on the payment period of the target insurance product.

[0116] Step S403, obtaining the target sub-data of the target object in the next payment period, which matches the target type.

[0117] Step S404, updating the initial object state features based on the target sub-data of the target type, to obtain the target object state features.

[0118] In step S401 of some embodiments, the selection operation refers to the process of the target object selecting the recommended candidate insurance product. In the application scenario of the present embodiment, the operation is usually clicking or confirming by the target object in the system recommendation page, and completing the payment. The target insurance product is the object selected by the target object from the candidate insurance products, for example, if the target object selects and purchases “health insurance A” and “life insurance C” from the candidate insurance products “health insurance A”, “health insurance B”, and “life insurance C”, then “health insurance A” and “life insurance C” are the target insurance products.

[0119] In step S402 of some embodiments, the payment period refers to the premium payment period of the insurance product, which can be a quarter, half a year, or a year, etc., and is not limited thereto. In the present embodiment, the payment period can be divided into a short-term payment period and a long-term payment period. Exemplarily, when the payment period is one month, one quarter, or half a year, it can be determined as a short-term payment period. When the payment period is one year or more, it can be determined as a long-term payment period.

[0120] The type of medical data to be filtered is then determined according to the payment period of the target insurance product (e.g. quarterly, annually, etc.). If the payment period indicates that the target insurance product is a short-term insurance product, the target type is usually matched with the data type related to short-term rehabilitation such as postoperative infection indicators, pain relief speed, and recovery rate of motor function (e.g. wound spread, gait analysis, heart rate). If the payment period indicates that the target insurance product is a long-term insurance product, the target type is usually associated with long-term health data such as chronic disease recurrence probability, organ function recovery trend, and psychological state assessment (e.g. mental illness, blood glucose, and blood pressure).

[0121] In step S403 of some embodiments, according to the payment period of the target insurance product and the target type filtered, the physiological data of the target object in the latest period and type related to the target object are obtained, i.e. the target sub-data. For example, if the payment period of the insurance product of the target object is half a year, and the target type filtered is mental illness, blood glucose, and blood pressure, the mental illness, blood glucose, and blood pressure data of the target object in the next half year are obtained.

[0122] In step S404 of some embodiments, the target sub-data is subjected to risk prediction to obtain the latest health risk level, and the health risk level and the target sub-data are used to construct the state feature of the target object.

[0123] The steps S401 to S404 shown in the embodiments of the present application determine the most suitable target insurance product from a plurality of candidate products in response to the selection operation of the target object. After the corresponding health data type is filtered based on the payment period, the health data of the target object in the next payment period is obtained, which provides an accurate basis for the recommendation of the insurance product. By updating the health data of the target object, the latest state feature of the target object is finally accurately obtained.

[0124] The specific process of calculating the reward score can be referred to in Figure 5 In some embodiments, step S104 can further include but is not limited to steps S501 to S504:

[0125] Step S501, extracting the target medical record and the target risk level from the state feature of the target object.

[0126] Step S502, performing health state assessment on the target medical record to obtain a health state score.

[0127] Step S503, performing risk trend assessment according to the health risk level and the target risk level to obtain a risk trend score.

[0128] Step S504, according to the health status score and the risk trend score, a weighted calculation is performed to obtain an execution reward score.

[0129] In step S501 of some embodiments, the target medical record refers to relevant medical data matching the aforementioned target type. The target risk level is the health risk level obtained by risk prediction according to the target medical record.

[0130] In step S502 of some embodiments, a pre-trained scoring model or a multi-dimensional assessment method can be used to implement health status assessment. For example, a weighted scoring system can be used, in which different weights are assigned to each health indicator (such as blood glucose, blood pressure, body weight, etc.) of the target object, and a comprehensive score is calculated. For example, based on the standard range of each health indicator, the score of each health data of the target object is calculated, and the total health status score is calculated by weighting.

[0131] In step S503 of some embodiments, a support vector regression (SVR), random forest regression, or other model can be used for risk trend assessment. The higher the risk trend score, the better the improvement in the health status of the target object. If the risk trend score is lower, it means that the health status of the target object may be deteriorating.

[0132] In step S504 of some embodiments, by setting appropriate weights for the health status score and the risk trend score, and combining the current health data of the target object, a weighted sum is obtained to generate an execution reward score. The weights of the health status score and the risk trend score can be set according to actual needs. For example, the weight of the health status score can be 0.7, and the weight of the risk trend score can be 0.3.

[0133] The steps S501 to S504 shown in the embodiments of the present application extract medical records and health risk levels from the state characteristics of the target object by dynamically analyzing the health data of the target object, and perform health status assessment and risk trend assessment based on these data, thereby achieving more accurate and personalized insurance product recommendations. By extracting the health status score from the medical records of the target object, and combining the health risk level and the target risk level to predict the risk trend, the technical solution can dynamically adjust the health status of the target object, and provide a reference for insurance product recommendations according to the health changes. In addition, by weighting the health status score and the risk trend score, the health risk of the target object can be more accurately assessed, thereby providing more accurate reward scores for subsequent recommendation schemes.

[0134] In step S105 of some embodiments, please refer to Figure 6, step S105 includes but is not limited to steps S601 to S603:

[0135] Step S601, superimposing the basic reward score and the execution reward score to obtain a target reward score.

[0136] Step S602, according to the reference score of each preset original insurance product and the target reward score, filtering out a reference insurance product from at least two original insurance products, wherein the target reward score is greater than or equal to the reference score.

[0137] Step S603, product prediction is performed on the target object state feature by the recommendation scheme prediction model to obtain an intermediate insurance product, and the target product recommendation scheme is determined based on the intermediate insurance product and the reference insurance product.

[0138] In step S601 of some embodiments, the basic reward score and the execution reward score can be directly added or weighted summed, and the final calculation result is the target reward score. The target reward score can be used by the target object for subsequent exchange of other insurance products when determining the final target product recommendation scheme.

[0139] In step S602 of some embodiments, for the premium discount and the gift of value-added services (such as free extension of insurance period) and other user benefits, they can be pushed to the target object in the form of insurance products to encourage the target object to purchase the insurance products. Such insurance schemes containing discounts and value-added services are original insurance products, and the reference score is a standard for measuring whether the target object can receive the original insurance product. Further, the target reward score is compared with the reference score of each original insurance product, and the original insurance product whose target reward score is greater than or equal to the reference score is selected as the reference insurance product.

[0140] In step S603 of some embodiments, the target object state feature is input into the recommendation scheme prediction model of step S103, and the output insurance product is the intermediate insurance product. The specific prediction process is consistent with that of step S103, and only the input is different. The intermediate insurance product and the reference insurance product are pushed to the target object as the target product recommendation scheme.

[0141] The steps S601 to S603 shown in the embodiments of the present application can comprehensively consider the health status, health behavior and historical health data of the target object by superimposing and calculating the basic reward score and the execution reward score to obtain the target reward score. Then, on the basis of the target reward score, the reference insurance product matched with the target reward score is screened to further determine the insurance product type suitable for the target object. Finally, the product prediction model of the recommendation scheme combines the state characteristics of the target object to predict the product, generate an intermediate insurance product, and determine the final target product recommendation scheme by comparing with the reference insurance product, so as to ensure that the recommended insurance product is highly matched with the health needs of the target object.

[0142] Based on the finally generated target product recommendation scheme, the intermediate insurance product and the reference insurance product are displayed to the user through the man-machine interaction interface, and detailed product information, preferential content and reward mechanism are attached.

[0143] Please refer to Figure 7 The embodiments of the present application also provide an insurance product recommendation device, which can implement the above-mentioned insurance product recommendation method. The device comprises:

[0144] The acquisition module 701 is configured to acquire the original medical record of the target object, and perform risk prediction on the original medical record to obtain the health risk level of the target object.

[0145] The state characteristic construction module 702 is configured to construct the initial object state characteristic of the target object according to the original medical record and the health risk level.

[0146] The first recommendation module 703 is configured to perform recommendation scheme generation on the initial object state characteristic by using the preset recommendation scheme prediction model to obtain a candidate product recommendation scheme. The candidate product recommendation scheme comprises a candidate insurance product recommended to the target object and a basic reward score matched with the candidate insurance product.

[0147] The state characteristic updating module 704 is configured to update the initial object state characteristic of the target object in response to the selection operation of the target object on the candidate insurance product to obtain a target object state characteristic, and perform reward calculation according to the initial object state characteristic and the target object state characteristic to obtain an execution reward score of the target object.

[0148] The second recommendation module 705 is configured to perform recommendation scheme generation according to the recommendation scheme prediction model, the target object state characteristic, the basic reward score and the execution reward score to obtain a target product recommendation scheme, and perform insurance product recommendation to the target object based on the target product recommendation scheme.

[0149] The specific implementation of the insurance product recommendation device is basically the same as the specific embodiments of the above-mentioned insurance product recommendation method, and will not be repeated here.

[0150] The embodiment of the present application further provides an electronic device, which comprises a memory and a processor, the memory stores a computer program, and the processor executes the computer program to realize the insurance product recommendation method. The electronic device can be any intelligent terminal, such as a tablet computer or a vehicle-mounted computer.

[0151] Please refer to Figure 8 , Figure 8 The hardware structure of the electronic device of another embodiment is shown, which comprises:

[0152] The processor 801 can be implemented in the form of a general-purpose CPU (Central Processing Unit), a microprocessor, an ASIC (Application Specific Integrated Circuit), or one or more integrated circuits, and is used to execute related programs to implement the technical solutions provided by the embodiments of the present application.

[0153] The memory 802 can be implemented in the form of a ROM (Read Only Memory), a static storage device, a dynamic storage device, or a RAM (Random Access Memory). The memory 802 can store an operating system and other application programs. When the technical solutions provided by the embodiments of the present application are implemented by software or firmware, the related program codes are stored in the memory 802 and are called and executed by the processor 801 to implement the insurance product recommendation method of the embodiments of the present application.

[0154] The input / output interface 803 is used to realize information input and output.

[0155] The communication interface 804 is used to realize the communication interaction between the device and other devices, and can realize communication through a wired manner (such as a USB, a network cable, etc.) or a wireless manner (such as a mobile network, WIFI, Bluetooth, etc.).

[0156] The bus 805 is used to transmit information between various components (such as the processor 801, the memory 802, the input / output interface 803, and the communication interface 804) of the device.

[0157] The processor 801, the memory 802, the input / output interface 803, and the communication interface 804 are connected to each other through the bus 805 to realize communication connection between them in the device.

[0158] The embodiment of the present application further provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the insurance product recommendation method.

[0159] The memory, as a non-transitory computer readable storage medium, can be used to store non-transitory software programs and non-transitory computer executable programs. In addition, the memory can include a high-speed random access memory, and can also include a non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state memory device. In some embodiments, the memory can optionally include a memory remotely arranged relative to the processor, and these remote memories can be connected to the processor through a network. Examples of the above network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0160] The insurance product recommendation method, the insurance product recommendation device, the electronic equipment and the storage medium provided by the embodiment of the present application are used to obtain the original medical record of the target object, predict the health risk level of the target object according to the record, and construct the initial object state feature of the target object according to the original medical record and the health risk level. Then, the initial object state feature is generated according to the recommendation scheme prediction model to obtain a candidate product recommendation scheme. When the user determines the selection of the candidate insurance product, the initial object state feature is updated to obtain the target object state feature, so as to realize the dynamic response to the change of the target object condition. The initial object state feature and the target object state feature are used to calculate an execution reward score. Finally, the target product recommendation scheme is generated according to the recommendation scheme prediction model, the target object state feature, the basic reward score and the execution reward score. The method of the embodiment of the present application effectively solves the problems of insufficient personalization and inaccurate recommendation caused by the failure to fully consider dynamic factors in the traditional insurance product recommendation, so that the recommendation of the insurance product is more accurate.

[0161] The embodiments described in the embodiments of the present application are used to more clearly illustrate the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided by the embodiments of the present application. Those skilled in the art can know that, with the evolution of technology and the appearance of new application scenarios, the technical solutions provided by the embodiments of the present application are also applicable to similar technical problems.

[0162] Those skilled in the art can understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and can include more or fewer steps than the figures shown, or combine certain steps, or different steps.

[0163] The apparatus embodiments described above are merely exemplary, and the units described as separate units can or can not be physically separate, i.e., can be located in one place, or can be distributed over multiple network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the embodiment.

[0164] Those skilled in the art can understand that all or some of the steps in the method disclosed above, the functional modules / units in the system and the device can be implemented as software, firmware, hardware and appropriate combinations thereof.

[0165] The terms "first", "second", "third", "fourth" and the like in the description of the application and in the claims of the foregoing drawings, if any, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the use of the terms so

[0166] It should be understood that in this application, "at least one" means one or more, and "multiple" means two or more. "And / or" is used to describe the relationship between the associated objects, which means that there can be three relationships, for example, "A and / or B" can mean that there are three cases: only A, only B, and A and B at the same time, where A and B can be singular or plural. The character " / " generally represents that the associated objects before and after are in an "or" relationship. "At least one of the following" or the like means any combination of these items, including any combination of single or multiple items. For example, at least one of a, b or c, can mean a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0167] In several embodiments provided in the present application, it should be understood that the disclosed apparatus and method can be implemented by other manners. For example, the apparatus embodiments described above are merely illustrative, for example, the division of the above units is merely a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units or components shown or discussed can be indirect coupling or communication connection through some interfaces, apparatuses or units, and can be electrical, mechanical or other forms.

[0168] The units described above as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they can be located in one place or distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0169] In addition, the functional units in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0170] If the integrated unit is realized in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the part of the prior art that makes a contribution or the whole or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, including a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method of each embodiment of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various program storage media.

[0171] The preferred embodiments of the embodiments of the present application are described above with reference to the accompanying drawings, but this does not limit the scope of the embodiments of the present application. Any modifications, equivalent replacements and improvements made by those skilled in the art without departing from the scope and essence of the embodiments of the present application shall be within the scope of the embodiments of the present application.

Claims

1. A method for recommending insurance products, characterized in that, The method includes: Obtain the original medical records of the target object, and perform risk prediction on the original medical records to obtain the health risk level of the target object; The initial object state characteristics of the target object are constructed based on the original medical records and the health risk level. A recommendation scheme is generated by using a preset recommendation scheme prediction model to predict the initial object's state features, resulting in a candidate product recommendation scheme; wherein, the candidate product recommendation scheme includes candidate insurance products recommended to the target object, and a basic reward score matching the candidate insurance products; In response to the target object's selection operation of the candidate insurance product, the initial object state characteristics of the target object are updated to obtain the target object state characteristics, and a reward calculation is performed based on the initial object state characteristics and the target object state characteristics to obtain the target object's execution reward score; The recommendation scheme is generated based on the prediction model of the recommendation scheme, the state characteristics of the target object, the basic reward score and the execution reward score, to obtain the target product recommendation scheme, and insurance products are recommended to the target object based on the target product recommendation scheme.

2. The method according to claim 1, characterized in that, The original medical record includes original sub-data of multiple data types. In response to the target object's selection operation of the candidate insurance product, the initial object state characteristics of the target object are updated to obtain the target object state characteristics, including: In response to the target object's selection operation of the candidate insurance products, a target insurance product is determined from at least two candidate insurance products; The target type is obtained by filtering from multiple data types based on the payment cycle of the target insurance product; Obtain the target sub-data of the target object that matches the target type in the next payment cycle; The initial object state features are updated based on the target sub-data of the target type to obtain the target object state features.

3. The method according to claim 1, characterized in that, The step of generating a recommendation scheme based on the prediction model of the recommendation scheme, the state features of the target object, the basic reward score, and the execution reward score to obtain a target product recommendation scheme includes: The target reward score is obtained by superimposing the base reward score and the execution reward score. Based on the reference score of each preset original insurance product and the target reward score, a reference insurance product is selected from at least two original insurance products, wherein the target reward score is greater than or equal to the reference score; The recommended scheme prediction model is used to predict the product based on the state characteristics of the target object to obtain an intermediate insurance product, and the recommended scheme for the target product is determined based on the intermediate insurance product and the reference insurance product.

4. The method according to claim 1, characterized in that, Before generating recommendation schemes from the initial object state features using a preset recommendation scheme prediction model to obtain candidate product recommendation schemes, the method further includes training the recommendation scheme prediction model, specifically including: Obtain the first sample state features of the preset sample object; The initial product recommendation scheme and the execution probability of the initial product recommendation scheme are obtained by predicting the recommendation scheme based on the state features of the first sample through a preset initial prediction model. The state features of the first sample are simulated and updated according to the initial product recommendation scheme to obtain the state features of the second sample. The initial product recommendation scheme is evaluated for reasonableness, and a reasonableness score is obtained. The state features of the first sample are evaluated to obtain a first state score, and the state features of the second sample are evaluated to obtain a second state score. Based on the execution probability, the rationality score, the first state score, and the second state score, a target loss value is calculated. The parameters of the initial prediction model are then updated based on the target loss value to obtain the recommendation scheme prediction model.

5. The method according to claim 4, characterized in that, The step of calculating the target loss value based on the execution probability, the rationality score, the first state score, and the second state score includes: The reward is calculated by combining the rationality score, the first state score, and the second state score to obtain the state reward data. The target loss value is obtained by integrating and calculating the execution probability and the state reward data.

6. The method according to any one of claims 1 to 4, characterized in that, The step of calculating the reward based on the initial object state characteristics and the target object state characteristics to obtain the execution reward score of the target object includes: Extract the target medical record and target risk level from the state characteristics of the target object; A health status assessment is performed on the target medical records to obtain a health status score; A risk trend assessment is performed based on the stated health risk level and the stated target risk level to obtain a risk trend score; The execution reward score is obtained by weighting the health status score and the risk trend score.

7. The method according to any one of claims 1 to 4, characterized in that, The step of performing risk prediction on the original medical records to obtain the health risk level of the target object includes: The original medical records are classified according to their data type to obtain at least two original sub-data. Temporal features are extracted from each of the original sub-data to obtain the data temporal features; Image features are extracted from the medical images in the original medical records to obtain medical image features; The health risk level of the target object is obtained by classifying the time-series features of the data and the features of the medical images into risk levels using a preset risk prediction model.

8. An insurance product recommendation device, characterized in that, The device includes: The acquisition module is used to acquire the original medical records of the target object, perform risk prediction on the original medical records, and obtain the health risk level of the target object; A state feature construction module is used to construct the initial object state features of the target object based on the original medical records and the health risk level. The first recommendation module is used to generate a recommendation scheme based on the initial object state features using a preset recommendation scheme prediction model, thereby obtaining a candidate product recommendation scheme; wherein, the candidate product recommendation scheme includes candidate insurance products recommended to the target object, and a basic reward score matching the candidate insurance products; The state feature update module is used to update the initial state features of the target object in response to the target object's selection operation of the candidate insurance product, to obtain the target object state features, and to calculate the reward based on the initial object state features and the target object state features, to obtain the target object's execution reward score; The second recommendation module is used to generate a recommendation scheme based on the recommendation scheme prediction model, the target object's state characteristics, the basic reward score, and the execution reward score, to obtain a target product recommendation scheme, and to recommend insurance products to the target object based on the target product recommendation scheme.

9. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

Images