Health insurance policy generation method and device, computer equipment and storage medium

Abstract

The application relates to the technical field of artificial intelligence, and discloses a health insurance policy generation method and device, computer equipment and a storage medium, which comprise the following steps: acquiring multi-source health data, original medical record data, historical disease data and an original health insurance policy of a target object; performing feature extraction on the multi-source health data to obtain multi-source health features; performing feature extraction on the original medical record data to obtain medical record entity features; performing risk assessment on the multi-source health features, the medical record entity features and the historical disease data to obtain health risk data of the target object; and generating a policy based on the health risk data and the original health insurance policy to obtain a target health insurance policy. The application can be applied to a policy generation scene in the field of financial technology, and improves the accuracy of health insurance policy generation.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and is applicable to the financial technology field, particularly to a method, apparatus, computer equipment, and storage medium for generating health insurance policies. Background Technology

[0002] A health insurance policy is generated and delivered by an insurance company after it has conducted an underwriting assessment of the application submitted by the applicant and approved it.

[0003] Currently, risk assessment and premium pricing for health insurance policies primarily rely on statistical models and health data. Insurance companies typically use electronic health records, questionnaire results, and medical examination reports as data inputs, combined with static variables such as age, gender, and body mass index to identify risks and determine policy prices. However, in practical applications, health data often originates from medical databases or user self-reports, resulting in incomplete and inaccurate information. This makes it difficult to fully and accurately reflect the insured's health information and to develop adaptive protection strategies based on individual health trends, thus affecting the accuracy of health insurance policies to some extent.

[0004] Therefore, improving the accuracy of health insurance policy generation has become an urgent technical problem to be solved. Summary of the Invention

[0005] This application provides a method, apparatus, computer equipment, and storage medium for generating health insurance policies using artificial intelligence, aiming to solve the technical problem that health data cannot fully and accurately reflect the health information of the insured and that it is difficult to formulate adaptive protection strategies based on individual health change trends, thereby improving the accuracy of health insurance policy generation.

[0006] Firstly, a method for generating health insurance policies is provided, including: Obtain multi-source health data, original medical records, historical disease data, and original health insurance policies of the target individuals; Feature extraction is performed on the multi-source health data to obtain multi-source health features; Feature extraction is performed on the original medical record data to obtain medical record entity features; A risk assessment is performed on the multi-source health characteristics, the medical record entity characteristics, and the historical disease data to obtain the health risk data of the target object; Based on the health risk data, a policy is generated from the original health insurance policy to obtain the target health insurance policy.

[0007] Secondly, a health insurance policy generation device is provided, including: The health data acquisition module is used to acquire multi-source health data, original medical record data, historical disease data, and original health insurance policies of the target object; A health feature extraction module is used to extract features from the multi-source health data to obtain multi-source health features; The case feature extraction module is used to extract features from the original medical record data to obtain medical record entity features; The health risk assessment module is used to perform risk assessment on the multi-source health characteristics, the medical record entity characteristics, and the historical disease data to obtain the health risk data of the target object. The policy generation module is used to generate a target health insurance policy based on the health risk data from the original health insurance policy.

[0008] Thirdly, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described health insurance policy generation method.

[0009] Fourthly, a computer-readable storage medium is provided, which stores a computer program that, when executed by a processor, implements the steps of the above-described health insurance policy generation method.

[0010] The aforementioned health insurance policy generation method, device, computer equipment, and storage medium provide a comprehensive data foundation for risk assessment by acquiring multi-source health data, original medical record data, historical disease data, and original health insurance policies for the target population. Next, feature extraction is performed on the multi-source health data to obtain multi-source health features; feature extraction is performed on the original medical record data to obtain medical record entity features; and unstructured data is transformed into quantifiable indicators for subsequent model processing. Furthermore, risk assessment is conducted on the multi-source health features, medical record entity features, and historical disease data to obtain the target population's health risk data, achieving risk quantification. Finally, based on the health risk data, the original health insurance policy is generated to obtain the target health insurance policy. This approach provides low-cost, high-quality protection for low-risk individuals while controlling insurance company payout risks through differentiated pricing, simultaneously meeting the health protection needs of the insured population. It solves the technical problem that health data cannot comprehensively and accurately reflect the insured's health information and makes it difficult to formulate adaptive protection strategies based on individual health trends, thus improving the accuracy of health insurance policy generation. Attached Figure Description

[0011] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 This is a schematic diagram of an application environment for a health insurance policy generation method in one embodiment of this application; Figure 2 This is a flowchart illustrating a method for generating a health insurance policy in one embodiment of this application; Figure 3 yes Figure 2 A schematic diagram of a specific implementation method for step S30; Figure 4 yes Figure 2 A schematic diagram of a specific implementation of step S40; Figure 5 yes Figure 4 A flowchart illustrating a specific implementation of step S43; Figure 6 yes Figure 2 A schematic diagram of a specific implementation method for step S50; Figure 7 yes Figure 6 A flowchart illustrating a specific implementation of step S53; Figure 8 yes Figure 6 A flowchart illustrating a specific implementation of step S54; Figure 9 This is a schematic diagram of a health insurance policy generation device in one embodiment of this application; Figure 10 This is a schematic diagram of the structure of a computer device according to one embodiment of this application. Detailed Implementation

[0013] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0014] The health insurance policy generation method provided in this application embodiment can be applied to, for example, Figure 1In this application environment, the client communicates with the server via a network. The server can obtain multi-source health data, original medical record data, historical disease data, and original health insurance policies of the target object from the client; it performs feature extraction on the multi-source health data to obtain multi-source health features; it performs feature extraction on the original medical record data to obtain medical record entity features; it performs risk assessment on the multi-source health features, medical record entity features, and historical disease data to obtain the target object's health risk data; based on the health risk data, it generates the original health insurance policy to obtain the target health insurance policy, and then feeds the target health insurance policy back to the client.

[0015] In this application, in the scenario of generating health insurance policies, a complete data foundation for risk assessment is laid by acquiring multi-source health data, original medical record data, historical disease data, and original health insurance policies of the target object. Next, feature extraction is performed on the multi-source health data to obtain multi-source health features; feature extraction is performed on the original medical record data to obtain medical record entity features; unstructured data is transformed into quantifiable indicators for subsequent model processing. Furthermore, risk assessment is performed on the multi-source health features, medical record entity features, and historical disease data to obtain the target object's health risk data, achieving risk quantification. Finally, based on the health risk data, the original health insurance policy is generated to obtain the target health insurance policy. This provides low-cost, high-quality protection for low-risk individuals while controlling the insurance company's payout risk through differentiated pricing, simultaneously meeting the health protection needs of the insured population. This solves the technical problem that health data cannot comprehensively and accurately reflect the insured's health information and makes it difficult to formulate adaptive protection strategies based on individual health trends, thus improving the accuracy of health insurance policy generation.

[0016] The client can be, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. The server can be implemented using a standalone server or a server cluster consisting of multiple servers. The following detailed description uses specific embodiments to illustrate this application.

[0017] Please see Figure 2 As shown, Figure 2 A flowchart illustrating the health insurance policy generation method provided in this application embodiment includes the following steps: S10: Obtain multi-source health data, original medical record data, historical disease data, and original health insurance policies of the target object; The health insurance policy generation method provided in this application can be applied to various health insurance policy generation scenarios. It is implemented through a server, which can obtain relevant data of the target object in real time, including but not limited to multi-source health data, original medical record data, historical disease data and original health insurance policies.

[0018] Specifically, the target is the insured person of health insurance or the policyholder of health insurance (if the policyholder and the insured person are the same person).

[0019] The target's multi-source health data comes from wearable devices (such as smart bracelets), medical examination institutions, etc., including but not limited to: medical examination indicators obtained by medical examination institutions from the target's medical examination, heart rate variability and sleep quality collected by wearable devices, frequency of use of health monitoring APP, health check-in records, etc.

[0020] Original medical record data consists of structured or unstructured text such as medical records, examination reports, and diagnostic conclusions recorded by the target individual at the hospital / medical institution, as well as images (such as CT images, X-ray images, etc.).

[0021] Historical disease data includes information such as the target individual's past disease diagnoses and treatment records.

[0022] It should be noted that the distinction between original medical record data and historical disease data is based on time. For example, original medical record data refers to case data generated within the previous month, while historical disease data refers to case data generated before the previous month.

[0023] The original health insurance policy is either a policy that the target individual has already purchased but needs to renew, or a standard policy template provided by the insurance company (for which the target individual has not yet purchased insurance). The original health insurance policy includes the basic terms and conditions of the health insurance product, the scope of coverage, the insurance period, the premium, and the target individual's user information.

[0024] S20: Extract features from multi-source health data to obtain multi-source health features; It is understandable that multi-source health data includes both structured and unstructured content. To improve the convenience and accuracy of subsequent processing, feature extraction is required to obtain multi-source health features. Specifically, feature extraction here refers to using natural language processing technology to transform the raw multi-source health data into quantifiable indicators, forming numerical multi-source health features. This allows for the extraction and standardization of relevant data indicators, reducing redundant information interference and facilitating subsequent model processing.

[0025] For example, extract data such as "average daily steps of 8,000" and "deep sleep duration of 2 hours" from the wearable device data of a target person, and extract data such as "fasting blood glucose of 6.2 mmol / L" from the physical examination report.

[0026] Before step S20 in some embodiments, the structured data (e.g., physical examination indicators such as blood glucose, blood pressure, BMI, blood lipids and other numerical indicators) in the multi-source health data can be preprocessed. The preprocessing process includes, but is not limited to: standardizing each indicator using Z-score or Min-Max method, using KNN imputation, mean / median filling, or using a model to predict missing values ​​(e.g. MICE algorithm).

[0027] S30: Extract features from the original medical record data to obtain the entity features of the medical record; It's important to understand that feature extraction here refers to extracting structured information such as disease name, disease-related indicators, and treatment methods from raw medical record data. For example, Figure 3 As shown, step S30, which involves feature extraction from the original medical record data to obtain the medical record entity features, includes the following steps: S31: Perform text conversion on the original medical record data to obtain the original medical record text; S32: Perform text cleaning on the original medical record text to obtain the initial medical record text; S33: Perform entity recognition on the initial medical record text to obtain the medical record entity features.

[0028] For steps S31 to S33, after obtaining the original medical record data of the target object, the original medical record data is first converted into text to obtain the original medical record text. Next, the original medical record text is cleaned to remove noise (such as spaces), correct errors in expression / grammar, and standardize the text (e.g., unify capitalization, unify full-width and half-width characters, remove punctuation marks, etc.), thus obtaining the initial medical record text and improving the data quality. Further, a pre-trained medical entity detection model (such as PIMCORE-BERT, BlueBert, etc.) is used to perform entity recognition on the initial medical record text to obtain medical record entity features. This helps in health insurance policy scenarios to quickly identify disease names and treatment methods, rapidly determine the health status of the target object, and assist insurance companies in conducting more accurate risk assessments and pricing. For example, based on the patient's age, medical history, and other entity information, combined with big data analysis, their future disease risk can be assessed, thereby formulating more reasonable insurance rates.

[0029] It should be noted that original medical record data includes various formats, such as electronic documents, images, and audio. If the original medical record data consists of handwritten records by medical staff or images of electronic medical records, a pre-trained OCR model (such as CRNN + CTC loss) is used to recognize and extract the text, resulting in the original medical record text. If the original medical record data consists of diagnostic audio from medical staff, a pre-trained speech recognition model (such as Whisper, DeepSpeech, etc.) is used for transcription to obtain the original medical record text. If the original medical record data is an electronic medical record document, the original medical record text is directly extracted.

[0030] Understandably, entity recognition here refers to the process of automatically identifying and labeling various entity features from initial medical record text using a medical entity detection model. Specifically, the medical entity detection model can use medical-specific word segmentation tools (such as Jieba + a medical domain-specific dictionary) or BioNLP toolkits (such as Scispacy) for entity recognition. During training, the medical entity detection model learns from a large amount of labeled medical record text data, mastering the grammatical structure, contextual relationships, and other features of different entities in the text. This knowledge is then used to analyze new initial medical record text and identify the corresponding entities.

[0031] For example, given the text "Patient Li Si, chief complaint of cough and phlegm for 3 days, diagnosed with bronchitis", the model can identify "Li Si" as a person entity, "cough and phlegm" as a symptom entity, and "bronchitis" as a disease entity. These person entity, symptom entity, and disease entity constitute the medical record entity features.

[0032] It should be noted that medical record entity features are entity information with specific meaning and independent semantics identified from the medical record text. In health insurance, common medical record entity features include basic patient information (name, age, gender, etc.), disease name (such as hypertension, diabetes, etc.), symptoms (headache, fever, etc.), examination items (blood routine, CT scan, etc.), and treatment methods (drug treatment, surgery, etc.), but are not limited to these.

[0033] In step S33 of some embodiments, after obtaining the medical record entity features, the medical record entity characteristics can be mapped to the ICD-11 coding system to obtain the coded medical record entity features. For example: "Type 2 diabetes". "Type 2DM" “E11”.

[0034] It should be noted that the ICD-11 coding system is a standardized coding system used to identify diseases, health conditions, and related factors. By mapping the physical characteristics of medical records to the ICD-11 coding system, the accuracy of data recording and collection can be improved, while also facilitating risk assessment.

[0035] S40: Conduct risk assessments on multi-source health characteristics, medical record entity characteristics, and historical disease data to obtain health risk data for the target population; It should be noted that the risk assessment here refers to predicting the probability of the target individual becoming ill or making a claim in the future. In other words, health risk data reflects the target individual's insurable risk and can be used for subsequent pricing or rejection processing. In some embodiments, such as... Figure 4 As shown, step S40, which involves risk assessment of multi-source health characteristics, medical record entity characteristics, and historical disease data to obtain health risk data for the target object, includes the following steps: S41: Time-series alignment is performed based on multi-source health features and medical record entity features to obtain health time-series features; S42: Perform feature fusion on health time-series characteristics and historical disease data to obtain multi-source health fusion data; S43: Risk prediction is performed based on multi-source health fusion data to obtain a predicted disease risk score; wherein, the predicted disease risk score includes the current disease risk score at the current moment and the future disease risk score for a preset period after the current moment; For steps S41 to S43, firstly, by aligning multi-source health features and medical record entity features over time, key health information from different time points is integrated to obtain health time-series features, ensuring data consistency and integrity. Next, the health time-series features and historical disease data are fused to obtain multi-source health fusion data, which can comprehensively consider multiple factors, enrich data dimensions, and uncover deeper health patterns. Finally, risk prediction is performed based on the multi-source health fusion data to obtain a predicted disease risk score. This predicted disease risk score includes the current disease risk score at the current moment and the future disease risk score for a preset period after the current moment. This allows for timely understanding of current health status and advance knowledge of future disease possibilities, providing a comprehensive and accurate basis for the risk assessment of health insurance products.

[0036] It's important to understand that since the acquisition times of multi-source health features and medical record entity features may differ, temporal alignment here refers to organizing and matching data from these different time points to ensure consistency across time. For example, it involves correlating heart rate data recorded by a wearable device on a particular day with related symptoms recorded in the medical record for that day. Specifically, by analyzing and matching multi-source health features and medical record entity features, they are integrated in chronological order. Insurance companies can gain a more comprehensive and accurate understanding of the target individual's dynamic health changes. For instance, if a target individual's heart rate is consistently high for a period of time, and related symptoms are recorded in their medical record, their health risk can be assessed promptly, providing a more scientific basis for subsequent insurance pricing and claims decisions.

[0037] It's important to clarify that feature fusion here refers to integrating feature data from different types and sources to form a comprehensive and holistic feature dataset. For example, in the health insurance scenario, this involves combining health time-series features with historical disease data features to create multi-source health fusion data, which integrates the target individual's current health dynamics and past disease history.

[0038] Before step S42 in some embodiments, feature extraction can be performed on historical disease data to obtain historical medical record features. The specific implementation steps are basically the same as those shown in steps S31 to S33 above, and will not be repeated here. Next, the health time-series features and historical medical record features are fused to obtain multi-source health fusion features.

[0039] For example, an insurance company possesses time-series health data of a target individual, showing a recent upward trend in their blood pressure, along with historical data on hypertension from five years ago. By integrating these two sets of data through feature fusion, a multi-source health fusion dataset is formed. This multi-source health fusion dataset can reflect a higher risk of hypertension recurrence in the target individual.

[0040] In some embodiments, such as Figure 5 As shown, step S43, which involves risk prediction based on multi-source health fusion data to obtain a predicted disease risk score, includes the following steps: S431: Perform disease identification on multi-source health fusion data to obtain the disease prediction probability data at the current moment, and confirm the disease prediction probability data as the current disease risk score; S432: Perform disease risk assessment on multi-source health fusion data to obtain a future disease risk score.

[0041] Specifically, a pre-trained risk prediction model (Long Short-Term Memory Network (LSTM) or Temporal Fusion Transformer (TFT)) is used to analyze various features and identify diseases in multi-source health fusion data to obtain disease prediction probability data at the current moment. This disease prediction probability data is used to characterize the probability data of whether the target object has chronic diseases (various chronic diseases), abnormal physiological indicators, etc. at the current moment.

[0042] For example, risk prediction models can make predictions based on features from multi-source health fusion data. An example of the reasoning logic is as follows: Continued weight gain increases the risk of fat accumulation. Glycated hemoglobin levels rise year by year, leading to increased insulin resistance. Decreased heart rate variability → impaired autonomic nervous system function; Reduced sleep leads to a decline in healthy behaviors.

[0043] Furthermore, the pre-trained risk prediction model is used to assess the risk of multi-source health fusion data, obtaining a future disease risk score for a preset period after the current time. This future disease risk score represents the probability of developing a serious disease (such as diabetes, coronary heart disease, etc.) within the preset period after the current time.

[0044] It should be noted that the preset time period needs to be determined based on the coverage period of the health insurance product. For example, if the coverage period of a certain health insurance is 1 year, then the preset time period can be set to 1 year; if the coverage period of a certain health insurance is 10 years, then the preset time period can be set to 10 years. Specifically, the preset time period needs to be greater than or equal to the coverage period.

[0045] Understandably, the current disease risk score and future disease risk score are numerical values ​​that quantify the risk of a target individual developing various diseases. A higher score indicates a greater risk. For example, a current disease risk score of 0.3 means a 30% probability of currently developing the disease; a future disease risk score of 0.5 means a 50% probability of developing the disease within the next 3 months.

[0046] It should be noted that if the risk prediction model predicts the probability of the target object having multiple diseases at the current moment, the final disease prediction probability data is obtained by averaging.

[0047] In other embodiments, risk prediction can also be performed for a specific disease to obtain disease prediction probability data for that disease. Similarly, the future disease risk score is also calculated for that specific disease.

[0048] For example, if a person already has hypertension and arteriosclerosis, their current disease risk score is high, and their future disease risk score is also high.

[0049] A person who is currently healthy but has a family history of genetic diseases and a sedentary lifestyle has a low current disease risk score and a high future disease risk score.

[0050] S44: Weight the current disease risk score and the future disease risk score to obtain the health risk data of the target subject.

[0051] Specifically, the current disease risk score and the future disease risk score are weighted and calculated based on preset thresholds to obtain the target object's health risk data. These preset thresholds are pre-defined, for example, set to 0.5 and 0.5, meaning health risk data = 0.5 * current disease risk score + 0.5 * future disease risk score.

[0052] S50: Generate a target health insurance policy based on health risk data from the original health insurance policy.

[0053] It should be noted that the target health insurance policy includes a final quote provided by the insurance company and is offered to the target individual. The target health insurance policy becomes effective after the target individual accepts and pays the premium. In some embodiments, such as... Figure 6 As shown, step S50, which involves generating a target health insurance policy based on health risk data from the original health insurance policy, includes the following steps: S51: Generate candidate pricing discount data based on health risk data; Based on pre-defined rules linking health risks and premiums, various possible pricing discount schemes can be generated. For example: If the health risk data is in the range [0, 0.1), the candidate pricing discount data is -15% or -10%; If the health risk data is in the range of [0.1, 0.2), the candidate pricing discount data is -10% or -5%; If the health risk data is in the range of [0.2, 0.3), the candidate pricing discount data is -5% or 0%; If the health risk data is in the range of [0.3, 0.5), the candidate pricing discount data is +5% or +10%; If the health risk data is in the range of [0.5, 0.7), the candidate pricing discount data is +10% or +15%. If the health risk data is in the range of [0.7, 1], the application will be rejected.

[0054] Understandably, setting multiple options for candidate pricing discount data is intended to further predict whether the target audience will be motivated to improve their health when they receive the discount, thereby determining whether to increase the discount or decrease the price increase.

[0055] Furthermore, by setting multiple pricing discount levels based on health risk data, the product fully considers the differences in health risks among target groups, providing diversified pricing discount options to meet the needs of customers with different risk levels and enhancing the product's market adaptability. For example, a higher discount can be offered to younger customers with lower health risks to attract them to purchase or renew their policies; for customers with higher risks, a reasonable price increase can be given based on their specific circumstances, avoiding a "one-size-fits-all" pricing approach.

[0056] S52: Perform health prediction based on candidate pricing discount data to obtain health prediction data; wherein, the health prediction data is used to characterize the health status of the target object after obtaining the candidate pricing discount data; Specifically, pre-trained health prediction models can be used, taking candidate pricing discount data as one of the input variables and combining it with the target audience's existing health risk data. This simulates the target audience's health trends under different discount schemes, considers the impact of discount schemes on customer health management investments (such as providing health consultations and physical examinations), and predicts the probability of customers developing certain diseases in the future, thereby generating health prediction data. This allows for advance understanding of the impact of different pricing discount schemes on the target audience's health, providing a basis for subsequent selection of reasonable pricing discounts. It can also prevent unreasonable discount schemes from increasing customer health risks and provides a reference for insurance companies to assess the risks and benefits of the schemes.

[0057] It should be noted that the pre-trained health prediction model is trained based on the insurance company's existing health insurance product underwriting data and the insured's health information, and is capable of accurately predicting the health status of the target object after obtaining the candidate pricing discount data; For example, an insurance company offered three pricing options—70%, 75%, and 80%—to a 40-year-old customer with a history of hypertension. Using a health prediction model, combined with data on the customer's age, medical history, and lifestyle habits, the company predicted that the customer's risk of developing cardiovascular disease over the next 5 years was 30% under the 70% discount option, 25% under the 75% discount option, and 20% under the 80% discount option.

[0058] S53: Based on health prediction data and health risk data, candidate pricing discount data are filtered to obtain target pricing discount data; It should be noted that the filtering here refers to using health prediction data and health risk data to filter multiple candidate pricing discount data to obtain the most suitable discount data. In some embodiments, such as Figure 7 As shown, step S53, which involves filtering candidate pricing discount data based on health prediction data and health risk data to obtain target pricing discount data, includes the following steps: S531: Based on health prediction data and health risk data, difference calculations are performed to obtain predicted health change data; S532: Filter candidate pricing discount data based on predicted health change data to obtain target pricing discount data.

[0059] For steps S41 to S43, firstly, by calculating the difference between health prediction data and health risk data, the expected changes in the target object's health status can be accurately quantified, providing a key basis for assessing the impact of pricing discounts. Next, based on the predicted health change data, multiple candidate pricing discount data are selected to identify target pricing discount data that both aligns with the health trends of the target customers and ensures reasonable returns for the insurance company, thus achieving precise pricing and risk control.

[0060] Specifically, using specific algorithms or models, health prediction data and health risk data are compared and analyzed to calculate the difference between the two. For example, regression analysis can be used, with health risk data as the independent variable and health prediction data as the dependent variable. A mathematical model is then established to calculate the difference, thereby obtaining data predicting changes in health. This allows for a more accurate understanding of the target individual's health status trends after receiving candidate pricing discount data, helping insurance companies to more accurately assess risk and providing a scientific basis for subsequently developing reasonable pricing strategies.

[0061] In other embodiments, the difference between health prediction data and health risk data can be calculated directly to obtain predicted health change data.

[0062] After obtaining the predicted health change data, based on the predicted health change data corresponding to each candidate pricing discount data, the candidate pricing discount data with the largest predicted health change data value is selected as the target pricing discount data.

[0063] Understandably, for the target group, it's crucial to ensure that the health insurance coverage they receive aligns with their evolving health trends. If predicted health changes indicate a potential increase in health risk, the selected pricing discount data can guide the target group in managing their health risks and raising awareness. For insurance companies, by using predicted health change data to select pricing discounts, they can avoid placing excessive payout burdens on customers with excessively high health risks and unreasonable pricing discounts. This helps in accurately controlling risk and optimizing returns, thereby enhancing the competitiveness of health insurance products and the company's profitability.

[0064] For example, in a health insurance scenario, a 45-year-old male customer's health risk data, including a history of hypertension, smoking 20 cigarettes a day, and rarely exercising, results in a health risk score of 30% after risk assessment. Correspondingly, the candidate pricing discounts are +5% and +10%.

[0065] For the candidate pricing discount data of +5%, the health prediction model shows that the probability of this customer developing cardiovascular disease in the next year is 24% even with a +5% discount.

[0066] For the candidate pricing discount data of +10%, the health prediction model shows that the customer has a 20% probability of developing cardiovascular disease in the next year if they receive a +10% discount.

[0067] If the difference calculation method is used, the predicted health change data is 10% when the candidate pricing discount data is +10%, and the predicted health change data is 6% when the candidate pricing discount data is +5%.

[0068] Furthermore, the largest value is selected from the predicted health change data of 10% and 6%, and its corresponding candidate pricing discount data, which is +10%, is used as the target pricing discount data.

[0069] In some embodiments, in addition to filtering candidate pricing discount data based on health prediction data and health risk data, the profit of the insurance product can be further combined to obtain the target pricing discount data. The calculation of profit depends on the actual operating scenario of the insurance company, but can generally be based on the following logic: Profit = Premium Income - Expected Claims Costs. The expected claims costs vary among different insurance companies and different health insurance products, and are not limited here.

[0070] Finally, the selection criteria for candidate pricing discount data are obtained by weighting health prediction data, health risk data (calculated to obtain predicted health change data) and profit.

[0071] S54: Adjust the original health insurance policy based on the target pricing discount data to obtain the target health insurance policy.

[0072] It should be noted that the product adjustment here refers to the adjustment of the premium data in the original health insurance policy.

[0073] In some embodiments, such as Figure 8 As shown, step S54, which involves adjusting the original health insurance policy based on the target pricing discount data to obtain the target health insurance policy, includes the following steps: S541: Obtain the original premium data of the original health insurance policy; S542: Aggregate and calculate the original premium data based on the target pricing discount data to obtain the target premium data; S543: Adjust the pricing of the original health insurance policy based on the target premium data to obtain the target health insurance policy.

[0074] For steps S541 to S543, firstly, the original premium data of the original health insurance policy is obtained, and then the original premium data is aggregated and calculated (e.g., multiplication) based on the target pricing discount data to accurately derive the target premium data that matches the customer's health risk and discount. Further, the original health insurance policy is priced and adjusted based on the target premium data to obtain the target health insurance policy, which not only aligns with the customer's actual situation and protects their rights, but also facilitates reasonable pricing by the insurance company, enhancing product competitiveness and market adaptability.

[0075] Specifically, the original premium information recorded in the original health insurance policy is located and extracted through the insurance company's business system, database, or relevant document management channels. This provides accurate benchmark data for subsequent premium adjustments, ensuring that adjustments are made based on the original reasonable pricing and guaranteeing the consistency and accuracy of insurance pricing.

[0076] Furthermore, the obtained raw premium data is multiplied by the target pricing discount data to calculate the target premium data. For example, if the target pricing discount is +10%, the raw premium data is multiplied by (1+10%). If the target pricing discount is -15%, the raw premium data is multiplied by (1-15%).

[0077] Understandably, by recalculating premiums based on target pricing discount data, premiums can be precisely adjusted according to the health status of the target individuals, making the premiums more reasonably reflect the actual risk level of customers and improving the fairness and competitiveness of insurance products.

[0078] Finally, after obtaining the target premium data, the premium field in the original health insurance policy is modified according to the target premium data to update the premium amount information, while ensuring that other relevant terms remain unchanged.

[0079] For example: Suppose the original premium data in a certain health insurance policy is 5,000 yuan, and the calculated target premium data is 4,500 yuan. Then the annual premium amount in the original health insurance policy data is adjusted from 5,000 yuan to 4,500 yuan, forming the target health insurance policy.

[0080] After step S50 in some embodiments, the target health insurance policy is pushed to the target object, and the target health insurance policy becomes effective after the target object approves and pays the premium.

[0081] In some embodiments, the health insurance policy generation method illustrated above will be automatically executed once a period of time before the policy expires (such as one month or one week before), thereby dynamically adjusting the policy premium. This can provide low-cost, high-quality protection for low-risk groups, control the insurance company's payout risk through differentiated pricing, and meet the health protection needs of the insured population.

[0082] As can be seen, in the above solution, in the scenario of generating health insurance policies, a complete data foundation is laid for risk assessment by acquiring multi-source health data, original medical record data, historical disease data, and original health insurance policies of the target object. Next, feature extraction is performed on the multi-source health data to obtain multi-source health features; feature extraction is performed on the original medical record data to obtain medical record entity features; unstructured data is transformed into quantifiable indicators for subsequent model processing. Furthermore, risk assessment is performed on the multi-source health features, medical record entity features, and historical disease data to obtain the target object's health risk data, achieving risk quantification. Finally, based on the health risk data, the original health insurance policy is generated to obtain the target health insurance policy. This provides low-cost, high-quality protection for low-risk groups, controls the insurance company's payout risk through differentiated pricing, and meets the health protection needs of the insured population. It solves the technical problem that health data cannot comprehensively and accurately reflect the insured's health information and makes it difficult to formulate adaptive protection strategies based on individual health trends, thus improving the accuracy of health insurance policy generation.

[0083] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0084] In one embodiment, a health insurance policy generation device is provided, which corresponds one-to-one with the health insurance policy generation method described in the above embodiments. For example... Figure 9 As shown, the health insurance policy generation device includes a health data acquisition module 101, a health feature extraction module 102, a medical record feature extraction module 103, a health risk assessment module 104, and a policy generation module 105. Detailed descriptions of each functional module are as follows: The health data acquisition module 101 is used to acquire multi-source health data, original medical record data, historical disease data and original health insurance policies of the target object; The health feature extraction module 102 is used to extract features from multi-source health data to obtain multi-source health features; The case feature extraction module 103 is used to extract features from the original medical record data to obtain the entity features of the medical record; The health risk assessment module 104 is used to conduct risk assessments on multi-source health characteristics, medical record entity characteristics and historical disease data to obtain health risk data of the target object. The policy generation module 105 is used to generate a target health insurance policy based on health risk data from the original health insurance policy.

[0085] In one embodiment, the health risk assessment module 104 is specifically used for: Health time-series features are obtained by time-series alignment based on multi-source health features and medical record entity features. By fusing health time-series characteristics and historical disease data, multi-source health fusion data is obtained; Risk prediction is performed based on multi-source health fusion data to obtain a predicted disease risk score; the predicted disease risk score includes the current disease risk score at the current moment and the future disease risk score for a preset period after the current moment. The current disease risk score and the future disease risk score are weighted and calculated to obtain the health risk data of the target population.

[0086] In one embodiment, the health risk assessment module 104 is specifically used for: Disease identification is performed on multi-source health fusion data to obtain the disease prediction probability data at the current moment, and the disease prediction probability data is confirmed as the current disease risk score; Disease risk assessment is performed on multi-source health fusion data to obtain future disease risk scores.

[0087] In one embodiment, the policy generation module 105 is specifically used for: Candidate pricing discount data are generated based on health risk data; Health prediction is performed based on candidate pricing discount data to obtain health prediction data; the health prediction data is used to characterize the health status of the target object after obtaining the candidate pricing discount data. Candidate pricing discount data are filtered based on health prediction data and health risk data to obtain target pricing discount data; The original health insurance policy is adjusted based on the target pricing discount data to obtain the target health insurance policy.

[0088] In one embodiment, the policy generation module 105 is specifically used for: Based on the difference calculation of health prediction data and health risk data, the predicted health change data is obtained; Candidate pricing discount data are filtered based on predicted health change data to obtain target pricing discount data.

[0089] In one embodiment, the policy generation module 105 is specifically used for: Obtain the original premium data from the original health insurance policy; The target premium data is obtained by aggregating and calculating the original premium data based on the target pricing discount data; The original health insurance policy is priced and adjusted based on the target premium data to obtain the target health insurance policy.

[0090] In one embodiment, the case feature extraction module 103 is specifically used for: The original medical record data is converted into text to obtain the original medical record text; The original medical record text is cleaned to obtain the initial medical record text. Entity recognition is performed on the initial medical record text to obtain the entity features of the medical record.

[0091] This application provides a health insurance policy generation device that lays a complete data foundation for risk assessment by acquiring multi-source health data, original medical record data, historical disease data, and original health insurance policies of the target object. Next, feature extraction is performed on the multi-source health data to obtain multi-source health features; feature extraction is performed on the original medical record data to obtain medical record entity features; unstructured data is transformed into quantifiable indicators for subsequent model processing. Further, risk assessment is performed on the multi-source health features, medical record entity features, and historical disease data to obtain the target object's health risk data, achieving risk quantification. Finally, based on the health risk data, the original health insurance policy is generated to obtain the target health insurance policy. This provides low-cost, high-quality protection for low-risk groups while controlling insurance company payout risks through differentiated pricing, simultaneously meeting the health protection needs of the insured population. It solves the technical problem that health data cannot comprehensively and accurately reflect the insured's health information and makes it difficult to formulate adaptive protection strategies based on individual health trends, thus improving the accuracy of health insurance policy generation.

[0092] Specific limitations regarding the health insurance policy generation device can be found in the limitations of the health insurance policy generation method described above, and will not be repeated here. Each module in the aforementioned health insurance policy generation device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0093] Please see Figure 10 , Figure 10 The hardware structure of a computer device according to another embodiment is illustrated. The computer device includes: The processor 1001 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application. The memory 1002 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 1002 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 1002 and is called and executed by the processor 1001 to execute the health insurance policy generation method of the embodiments of this application, including: Obtain multi-source health data, original medical records, historical disease data, and original health insurance policies of the target individuals; Feature extraction is performed on multi-source health data to obtain multi-source health features; Feature extraction is performed on the original medical record data to obtain the entity features of the medical records; Risk assessments are conducted on multi-source health characteristics, medical record entity characteristics, and historical disease data to obtain health risk data for the target population. Based on health risk data, the original health insurance policy is used to generate a target health insurance policy.

[0094] Input / output interface 1003 is used to implement information input and output; The communication interface 1004 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 1005 transmits information between various components of the device (e.g., processor 1001, memory 1002, input / output interface 1003, and communication interface 1004); The processor 1001, memory 1002, input / output interface 1003 and communication interface 1004 are connected to each other within the device via bus 1005.

[0095] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor: Obtain multi-source health data, original medical records, historical disease data, and original health insurance policies of the target individuals; Feature extraction is performed on multi-source health data to obtain multi-source health features; Feature extraction is performed on the original medical record data to obtain the entity features of the medical records; Risk assessments are conducted on multi-source health characteristics, medical record entity characteristics, and historical disease data to obtain health risk data for the target population. Based on health risk data, the original health insurance policy is used to generate a target health insurance policy.

[0096] It should be noted that the functions or steps that can be implemented by the computer-readable storage medium or computer device described above can be referred to the relevant descriptions on the server side and client side in the foregoing method embodiments. To avoid repetition, they will not be described one by one here.

[0097] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0098] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0099] It should be noted that any AI models, software tools, or components not belonging to this company appearing in the embodiments of this application are merely illustrative examples and do not represent actual use. All user personal information involved in the embodiments of this application has been authorized (with the knowledge and consent) by the relevant parties or has been fully authorized by all parties, and the executing entity may obtain it through various legal and compliant means. The collection, storage, use, processing, transmission, provision, and disclosure of the information, data, and signals involved all comply with relevant laws and regulations and do not violate public order and good morals.

[0100] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for generating a health insurance policy, characterized in that, include: Obtain multi-source health data, original medical records, historical disease data, and original health insurance policies of the target individuals; Feature extraction is performed on the multi-source health data to obtain multi-source health features; Feature extraction is performed on the original medical record data to obtain medical record entity features; A risk assessment is performed on the multi-source health characteristics, the medical record entity characteristics, and the historical disease data to obtain the health risk data of the target object; Based on the health risk data, a policy is generated from the original health insurance policy to obtain the target health insurance policy.

2. The method for generating a health insurance policy as described in claim 1, characterized in that, The step of performing risk assessment on the multi-source health characteristics, the medical record entity characteristics, and the historical disease data to obtain the health risk data of the target object includes: Based on the multi-source health features and the medical record entity features, time-series alignment is performed to obtain health time-series features; The health time-series features and the historical disease data are fused to obtain multi-source health fusion data; Risk prediction is performed based on the multi-source health fusion data to obtain a predicted disease risk score; wherein, the predicted disease risk score includes the current disease risk score at the current moment and the future disease risk score for a preset period after the current moment; The current disease risk score and the future disease risk score are weighted and calculated to obtain the health risk data of the target object.

3. The method for generating a health insurance policy as described in claim 2, characterized in that, The risk prediction based on the multi-source health fusion data, to obtain a predicted disease risk score, includes: Disease identification is performed on the multi-source health fusion data to obtain the disease prediction probability data at the current moment, and the disease prediction probability data is confirmed as the current disease risk score; The multi-source health fusion data is used to conduct a disease risk assessment to obtain the future disease risk score.

4. The method for generating a health insurance policy as described in claim 1, characterized in that, The process of generating a target health insurance policy based on the health risk data from the original health insurance policy includes: Candidate pricing discount data are generated based on the aforementioned health risk data; Based on the candidate pricing discount data, a health prediction is performed to obtain health prediction data; wherein, the health prediction data is used to characterize the health status of the target object after obtaining the candidate pricing discount data; Based on the health prediction data and the health risk data, the candidate pricing discount data are filtered to obtain the target pricing discount data; The original health insurance policy is adjusted based on the target pricing discount data to obtain the target health insurance policy.

5. The method for generating a health insurance policy as described in claim 4, characterized in that, The step of filtering the candidate pricing discount data based on the health prediction data and the health risk data to obtain the target pricing discount data includes: Based on the difference calculation between the health prediction data and the health risk data, the predicted health change data is obtained. The candidate pricing discount data is filtered based on the predicted health change data to obtain the target pricing discount data.

6. The method for generating a health insurance policy as described in claim 4, characterized in that, The step of adjusting the original health insurance policy based on the target pricing discount data to obtain the target health insurance policy includes: Obtain the original premium data of the original health insurance policy; The original premium data is aggregated and calculated based on the target pricing discount data to obtain the target premium data; The original health insurance policy is priced and adjusted based on the target premium data to obtain the target health insurance policy.

7. The method for generating a health insurance policy as described in any one of claims 1 to 6, characterized in that, The step of extracting features from the original medical record data to obtain medical record entity features includes: The original medical record data is converted into text to obtain the original medical record text; The original medical record text is cleaned to obtain the initial medical record text; Entity recognition is performed on the initial medical record text to obtain the entity features of the medical record.

8. A health insurance policy generation device, characterized in that, include: The health data acquisition module is used to acquire multi-source health data, original medical record data, historical disease data, and original health insurance policies of the target object; A health feature extraction module is used to extract features from the multi-source health data to obtain multi-source health features; The case feature extraction module is used to extract features from the original medical record data to obtain medical record entity features; The health risk assessment module is used to perform risk assessment on the multi-source health characteristics, the medical record entity characteristics, and the historical disease data to obtain the health risk data of the target object. The policy generation module is used to generate a target health insurance policy based on the health risk data from the original health insurance policy.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the health insurance policy generation method as described in 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 steps of the health insurance policy generation method as described in any one of claims 1 to 7.

Images