Underwriting multi-dimensional decision-making method for performing intelligent inspection and evaluation on health risk

Abstract

The invention discloses an underwriting multi-dimensional decision-making method for performing intelligent inspection and evaluation on health risks, and the method comprises the steps: collecting image data of a medical data file, and carrying out the preprocessing of the image data; the preprocessing comprises image enhancement, gray level transformation, Gaussian filtering and Hough transform correction; applying an OCR model based on a neural network, performing structured extraction on the image data according to an underwriting multi-dimensional medical decision rule to obtain structured data in a text form, performing standardization processing on the structured data, and transmitting the processed structured data into the underwriting multi-dimensional medical decision rule; and according to the underwriting multi-dimensional medical decision rule, performing multi-dimensional analysis and evaluation on the standardized structured data based on a preset rule flow, and automatically outputting an underwriting result. The multi-dimensional medical decision rule is adopted to effectively overcome multiple technical difficulties in the underwriting and underwriting fields, manual operation is replaced, the underwriting operation efficiency is greatly improved, the underwriting process is simpler, more convenient and faster, and the underwriting conclusion is more accurate.

Description

Technical Field

[0001] This application relates to the field of computer technology, and particularly to a multi-dimensional underwriting decision-making method for intelligent inspection and assessment of health risks. Background Art

[0002] Currently, for customers with health abnormalities, the existing underwriting automated judgment rule technology has limited intelligence and automation. In many cases, manual underwriting intervention is required, especially for medical underwriting cases. This type of business highly depends on manual operations, resulting in low efficiency and slow processing speed of the existing underwriting mode, seriously affecting the customer experience. With the continuous growth of business volume, the workload faced by underwriters has also increased sharply. A large number of business processes still rely on manual operations, which to a certain extent affects the timeliness of policy underwriting and further reduces customer satisfaction.

[0003] With the aggravation of aging in our country, the increase in the population with diseases and the rapid development of the insurance business, the underwriting scale of non-healthy individuals has been continuously expanding, and the number of manual underwriting cases intercepted by the underwriting rule engine has been increasing. Most of the underwriting cases transferred to manual are medical underwriting cases. Underwriters need to manually query past medical records to understand the customer's past medical conditions, then conduct specific risk analysis on the current insurance types, and finally give underwriting opinions. The existing underwriting mode not only increases the workload of underwriters, reduces the automatic underwriting passing rate of insurance companies, but also prolongs the customer's insurance application time and reduces the customer experience.

[0004] In view of the rapid development of current artificial intelligence and the digital economy, promoting digital transformation has become an inevitable choice for commercial insurance companies to enhance their competitiveness. Developing a set of multi-dimensional medical decision-making rules for underwriting is an inevitable result of conforming to the business development trend. It will significantly improve underwriting efficiency, optimize the customer experience, and promote the insurance industry to move towards the direction of intelligence and high efficiency. Summary of the Invention

[0005] This application proposes a multi-dimensional underwriting decision-making method for intelligent inspection and assessment of health risks, which solves the problems that the existing underwriting mode increases the workload of underwriters, reduces the automatic underwriting passing rate of insurance companies, prolongs the customer's insurance application time, and reduces the customer experience.

[0006] An embodiment of this application provides a multi-dimensional underwriting decision-making method for intelligent inspection and assessment of health risks, including:

[0007] Collect image data of medical data files and preprocess the image data; the preprocessing includes image enhancement, grayscale transformation, Gaussian filtering, and Hough transform correction;

[0008] Using the OCR model, structured extraction is performed on the image data according to the multi-dimensional medical decision-making rules for underwriting to obtain structured data in text form. The structured data is standardized and then input into the multi-dimensional medical decision-making rules for underwriting;

[0009] Based on a pre-set rule flow, the multi-dimensional medical decision-making rules for underwriting perform multi-dimensional analysis and evaluation on the standardized structured data and automatically output underwriting conclusions.

[0010] Further, the use of the OCR model to perform structured extraction on the image data according to the multi-dimensional medical decision-making rules for underwriting to obtain structured data in text form is specifically as follows:

[0011] Select an OCR model based on the convolutional neural network and recurrent neural network architectures, train the OCR model using a medical field text image dataset, input the pre-processed image into the OCR model, identify the text content, and clean the recognition result to remove garbled characters and duplicate characters;

[0012] Input the text after cleaning and recognition by the OCR model into a large language model. According to the multi-dimensional medical decision-making rules for underwriting, use natural language processing technology to mark key information including but not limited to disease names, symptoms, examination items, treatment methods, and onset times;

[0013] According to the underwriting business requirements and the medical knowledge system, construct a structured data framework, set the framework information categories, subdivide specific fields under each category, and fill the key information marked by the large language model into the corresponding fields to form structured data;

[0014] Perform association verification on the information in each field of the structured data to check logical consistency. For data with logical errors, information missing, or conflicts, re-analyze the original text or use the large language model to interpret it again.

[0015] Further, the standardization process of the structured data is specifically as follows: convert the structured data in text form into standardized numerical values or codes, normalize synonyms, unify the units of numerical results, and standardize the description of text results.

[0016] Further, the multi-dimensional medical decision-making rules for underwriting include:

[0017] According to the business content and scope, compile multiple types of medical dictionaries, standardize medical fields through the medical dictionaries, and set a unique code for each field;

[0018] Standardize inspection results, numerical units, synonyms, and near-synonyms;

[0019] Build a multi-level disease classification system, classify diseases into major categories according to the human body systems, and then further subdivide them according to the severity of the disease, the disease stage, and the treatment situation, and set corresponding underwriting evaluation rules and risk weights for each level.

[0020] Further, the multi-dimensional medical decision-making rules for underwriting perform multi-dimensional analysis and evaluation on the structured data after standardization based on a pre-set rule flow, specifically as follows:

[0021] Receive the structured data after standardization, parse it according to the pre-set data structure, extract key information and classify and store it in the corresponding logical area;

[0022] Based on the pre-constructed disease classification and stratification system, conduct a risk assessment on the diseases suffered by the customer, classify the diseases according to the human body systems, and then further subdivide them according to the severity of the disease, the stage, and the treatment effect. For the subdivided diseases, conduct quantitative scoring according to the corresponding underwriting strategies and risk weights;

[0023] In addition to the disease itself, comprehensively consider other risk factors of the customer, including personal basic information, family medical history, and previous underwriting records, use statistical analysis and machine learning algorithms to assign reasonable weights to each risk factor, and comprehensively calculate the scores of each risk factor and the disease risk score to obtain the comprehensive risk assessment result of the customer.

[0024] Further, it also includes: regularly paying attention to information such as medical research progress, updates of clinical practice guidelines, accumulation of industry underwriting experience, and adjustments of laws, regulations and policies. When new disease diagnosis methods, treatment technologies or risk factors appear, start the rule update process. After the new rules are demonstrated by internal experts, tested and verified, and reviewed for compliance, they are incorporated into the multi-dimensional medical decision-making rule flow for underwriting.

[0025] Further, the underwriting conclusion includes but is not limited to different types of underwriting conclusion texts such as standard body underwriting, additional premium underwriting, excepted liability underwriting, deferred underwriting or rejection of underwriting, and is accompanied by the reasons and basis for the decision.

[0026] Further, it also includes storing the underwriting conclusion and relevant data in the entire underwriting process in the underwriting database, recording the modification history, access logs and operation tracks of the data; the relevant data in the underwriting process includes original medical materials, processing results of each module, and decision-making process logs.

[0027] The embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the method described in any one of the above is implemented.

[0028] An embodiment of the present application further provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the above-mentioned method is implemented.

[0029] At least one of the above technical solutions adopted in the embodiments of the present application can achieve the following beneficial effects:

[0030] The present invention effectively overcomes multiple technical difficulties in the fields of underwriting and risk assessment by adopting multi-dimensional medical decision-making rules. Through methods such as "business structuring", "object standardization", and "rule configuration", it realizes the innovation of the risk assessment model, replaces manual operations, greatly improves the efficiency of risk assessment operations, makes the risk assessment process simpler and faster, and the risk assessment conclusion more accurate. Customers who originally needed to wait for manual risk assessment processing can quickly receive various risk assessment notices or the final risk assessment conclusion, meeting the insurance protection and insurance service needs of the people. It reduces the errors and inconsistent scales that may occur in manual risk assessment and lowers the error rate of risk assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The drawings described herein are used to provide a further understanding of the present application, and constitute a part of the present application. The schematic embodiments of the present application and their descriptions are used to explain the present application, and do not constitute an improper limitation of the present application. In the drawings:

[0032] Figure 1 is a flowchart of the multi-dimensional decision-making method for risk assessment of intelligent inspection and evaluation of the present invention;

[0033] Figure 2 is a structural diagram of the multi-dimensional medical decision-making rule for the application of the multi-dimensional decision-making method for risk assessment of intelligent inspection and evaluation of the present invention in thyroid ultrasound;

[0034] Figure 3 is a schematic structural diagram of the device execution subject of the multi-dimensional decision-making method for risk assessment of intelligent inspection and evaluation of the present invention;

[0035] Figure 4 is a principle block diagram of the device execution subject of the multi-dimensional decision-making method for risk assessment of intelligent inspection and evaluation of the present invention;

[0036] Figure 5 is in accordance with Figure 3 The data flow diagram of the execution subject structure. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0037] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments of this application and the corresponding drawings. Apparently, the described embodiments are only a part of the embodiments of this application, rather than all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in this application without creative efforts belong to the scope of protection of this application.

[0038] The following will detail the technical solutions provided by each embodiment of this application in conjunction with the drawings.

[0039] As Figure 1 shown, the present invention proposes a multi-dimensional underwriting decision-making method for intelligently inspecting and evaluating health risks, including:

[0040] S1. Collect image data of medical data files and preprocess the image data; the preprocessing includes image enhancement, grayscale transformation, Gaussian filtering, and Hough transform correction;

[0041] S2. Use an OCR model based on a neural network to perform structured extraction on the image data according to multi-dimensional medical underwriting decision rules to obtain structured data in text form, perform standardization processing on the structured data, and input it into the multi-dimensional medical underwriting decision rules;

[0042] S3. The multi-dimensional medical underwriting decision rules perform multi-dimensional analysis and evaluation on the standardized structured data based on a preset rule flow, and automatically output an underwriting conclusion.

[0043] The image data corresponding to the collection of medical data files in step S1 is mainly achieved by communicating and docking the server system. For example, by communicating and docking the HIS system on the hospital server with the data server of the insurance application management unit, internal and external data collection and automated collection are realized. Or it can be collected through image collection terminals such as mobile phones and scanners for file materials such as health declarations, physical examination reports, medical records, and previous underwriting materials. The collected data is stored in a data storage medium in the form of image data such as data pictures, PDF files, and image pictures, and is preprocessed in the above manner when the data is called. Specifically:

[0044] (1) Specific data content collected

[0045] Health declaration data: The health declaration information filled in by the customer when applying for insurance, including the current health status, past medical history, family medical history, and living habits (such as smoking, drinking, exercise frequency, etc.). These information are the basis for underwriting and can enable underwriters to initially understand the customer's health risks. For example, if the customer reports a long-term smoking history, this may increase the risk of respiratory diseases and affect the underwriting decision.

[0046] Physical examination report data: including data generated by physical examinations conducted at institutions designated by the insurance company or other medical institutions, such as general physical examinations (height, weight, blood pressure, heart rate, etc.), laboratory tests (blood routine, urine routine, blood sugar, blood lipids, liver function, kidney function, etc.), and instrumental tests (electrocardiogram, chest X-ray, abdominal ultrasound, etc.). Physical examination reports can objectively reflect the physical condition of customers. For example, abnormal blood sugar indicators may indicate that the customer is at risk of diabetes.

[0047] Medical records: Medical records generated when customers visit hospitals, including outpatient medical records and inpatient medical records. Outpatient medical records record customer symptoms, preliminary diagnosis, and treatment recommendations; inpatient medical records include detailed medical records, examination and test reports, surgical records, discharge summaries, etc. Medical records can provide customers with information on disease diagnosis, treatment process, and disease progression. For example, surgical records in inpatient medical records can clarify the severity of the customer's disease and treatment methods.

[0048] Previous underwriting conclusion data: The underwriting results of the customer's previous insurance applications, such as standard coverage, premium coverage, excluded liability coverage, extended coverage, or rejection. This information can help underwriters understand the customer's historical risk assessment. If the customer has been rejected for a certain disease, the current underwriting should focus on the disease and related risks.

[0049] Other medical data: In addition to the above, we can also collect the customer's medical records (registration time, department, diagnosis), medication records (drug name, dosage, medication time), rehabilitation records (rehabilitation treatment items, effects), and genetic test reports and chronic disease management data obtained from third-party medical data platforms. Genetic test reports can reveal the customer's potential genetic disease risks, and chronic disease management data can help understand the customer's chronic disease control status.

[0050] (2) Data collection implementation method

[0051] Internal system connection and collection: The insurance company's internal business system is connected to the physical examination management system. When a customer completes a physical examination at a designated physical examination institution, the physical examination report is automatically uploaded to the physical examination management system and then transmitted to the underwriting business system. Through data interfaces and transmission protocols, data can be obtained in real time or regularly to ensure data timeliness and accuracy.

[0052] External data exchange collection: Insurance companies connect with the information systems of cooperative hospitals and obtain customer medical records with customer authorization in accordance with data security protocols and medical data exchange standards (such as HL7 protocol). Cooperate with third-party medical data platforms to obtain relevant data through API calls, and conduct data quality assessment and screening during calls to ensure data reliability.

[0053] Automated collection process: Develop an automated data collection program, set collection tasks and frequencies, and collect data from different data sources according to rules. The collection program uses data query statements and ETL (Extract, Transform, Load) tools to associate and extract data by customer ID, and cleans, transforms, and loads the collected data into the underwriting database for subsequent processing and analysis.

[0054] Preprocess the image data; the preprocessing includes image enhancement, grayscale transformation, Gaussian filtering, and Hough transform correction. Collect medical materials such as customers' physical examination reports, medical records, inspection and test sheets, covering both paper documents and electronic documents. If the materials are electronic documents, convert non-image formats (such as PDF, Word, tables) into image formats for convenient OCR processing; if they are paper documents, convert them into images through scanning. For image materials, use image enhancement technology to improve the quality. Use grayscale transformation to enhance the contrast between text and background, making the text clearer; use Gaussian filtering and median filtering to remove noise generated by scanning and avoid interference with recognition; use Hough transform to correct image tilt, ensure that the text is neatly arranged, and improve the accuracy of OCR recognition.

[0055] In step S2, use an OCR model based on a neural network to perform structured extraction on the image data according to the multi-dimensional medical decision-making rules for underwriting to obtain structured data in text form, such as Figure 5 shown. This part is implemented on the corresponding neural network OCR model software of the medical data processor. The input data is the data output after collection, storage, and preprocessing in step 1. A data processing algorithm designed according to the multi-dimensional medical decision-making rules for underwriting is built on the medical data processor to process and obtain the data required by the multi-dimensional medical decision-making rule model on the underwriting unit processor. Specifically:

[0056] Select an OCR model with excellent performance, such as a model based on the architectures of convolutional neural network (CNN) and recurrent neural network (RNN). Use a text image dataset in the medical field to train the OCR model, enabling the model to learn features such as medical terms and special fonts and improving its recognition ability for medical materials. Input the preprocessed image into the OCR model to recognize the text content. Clean the recognition results, remove garbled characters and duplicate characters, unify the text case format, and initially organize the text in order to lay a foundation for subsequent large language model processing.

[0057] Select large language models that perform well in the medical field and fine-tune them using massive amounts of medical literature, medical records, and other data to make them more suitable for medical underwriting scenarios. Input the text after OCR recognition and cleaning into the large language model. Based on the multi-dimensional medical decision-making rules for underwriting, use natural language processing technology to mark key information such as disease names, symptoms, examination items, treatment methods, onset time, etc. For example, identify content such as "hypertension", "dizziness", "blood pressure measurement", "antihypertensive drug treatment", etc., and initially determine their respective categories.

[0058] According to the underwriting business requirements and the medical knowledge system, construct a structured data framework. Set up framework information categories such as "Personal Basic Information", "Disease Information", "Examination Information", "Treatment Information", etc. Subdivide specific fields under each category. For example, under "Disease Information", fields include disease name, disease type, onset time, disease severity, etc. Based on the key information marked by the large language model, accurately fill it into the corresponding fields. For example, fill in the patient's name, age, etc. into "Personal Basic Information"; fill in the disease name, onset time, etc. into "Disease Information" to initially form the prototype of structured data.

[0059] Perform correlation verification on the information in each field of the structured data to check logical consistency. Confirm whether the correlation between the disease and related symptoms, treatment methods is reasonable. For example, for customers with diabetes, their treatment methods should be related to blood sugar control; check whether the results of examination items match the disease diagnosis. For example, the chest X-ray examination results of pneumonia patients should show corresponding lesions in the lungs. For data with logical errors, missing information, or conflicts, re-analyze the original text or use the large language model to interpret it again. Supplement and perfect the missing information, correct the error information, and optimize and adjust the data to ensure that the structured data is accurate and reliable and meets the requirements for underwriting decision-making.

[0060] Next, perform standardization processing on the structured medical materials extracted, specifically: convert the text-form examination results into standardized numerical values or codes, normalize synonyms, unify the units of numerical results, and standardize the description of text results.

[0061] After the standardization processing is completed, input the data into the multi-dimensional medical decision-making rules for underwriting. The multi-dimensional medical decision-making rules for underwriting are specifically as follows:

[0062] 1. Business structuring

[0063] According to the business content and scope, multiple types of medical dictionaries are compiled. The medical fields are standardized through the medical dictionaries, and a unique code is set for each field. In this embodiment, according to the business content and scope, five major types of medical dictionaries are compiled, specifically including medical record dictionaries, disease + symptom ICD dictionaries, surgical dictionaries, inspection + examination dictionaries, etc. Through the above five major types of dictionaries, the medical fields are standardized, and a unique code is set for each field to ensure accurate correspondence in the underwriting multi-dimensional medical decision-making rules, thereby improving the accuracy of underwriting conclusions. This standardized dictionary system can not only improve the efficiency of data processing but also enhance the reliability of decision-making, and it is the basis for building an efficient digital underwriting system.

[0064] The construction of this system enables each field to be accurately matched with the key fields in the underwriting rule multi-dimensional decision table, ensuring that when assessing health risks, decisions can be made based on the most accurate and comprehensive medical information. The structuring and coding of this process greatly improve the accuracy and scientific nature of the underwriting process, laying a solid foundation for achieving a higher level of underwriting services.

[0065] 2. Object standardization

[0066] Standardize the inspection results, numerical units, synonyms, and near-synonyms. Medical professional knowledge is vast, and there are different synonyms for each hospital, each project, each disease, each indicator, and each description. Standardization needs to be carried out according to the already standardized medical data. On the one hand, standardize the inspection results and units: standardize and normalize the inspection results, units, and enumerated values of medical items. If the inspection result of an inspection item is a numerical value, the unit of the numerical value needs to be normalized; if the inspection result of an inspection item is in text form, all text enumerated values need to be sorted out. On the other hand, standardize synonyms: for inspection item indicators, such as the full name of alanine aminotransferase is glutamate-pyruvate transaminase, the alias is alanine transferase, and the abbreviation is ALT; the English name 1 is ALT; the English name 2 is GPT. For disease names, such as synonyms of diabetes include sugar disease, consumptive thirst disorder, glycosuria, insulin-dependent diabetes / non-insulin-dependent diabetes, adult-onset diabetes, and juvenile-onset diabetes. Such synonyms need to be sorted out and standardized to obtain a set of standardized medical field information.

[0067] 3. Rule configuration

[0068] Construct a multi-level disease classification system, divide diseases into major categories according to the human body system, and then subdivide them according to the severity of the disease, disease stage, and treatment situation, and set corresponding underwriting evaluation point rules and risk weights for each level.

[0069] For example, according to the underwriting scoring guidelines for common diseases of China Life Insurance, for more than 30,000 diseases corresponding to ICD10, the multi-dimensional medical decision-making rules for underwriting are compiled according to the logic and hierarchy of the underwriting scoring rules. The first question in the first layer of the underwriting rules for thyroid nodules is to judge the benignity and malignancy of the thyroid nodules. If it is malignant (the enumerated values are cancer, malignancy, etc.), the underwriting scoring will be carried out according to the multi-dimensional decision-making rules for thyroid cancer. If malignancy is not mentioned, the multi-dimensional decision-making rules for benign thyroid nodules will be followed, and then it will be judged whether the customer has had surgical resection. The only unit for the time since the surgical resection is "year", and the judgment will be made based on "year". And so on, the final multi-dimensional decision-making rules will give an accurate underwriting conclusion for each insurance type, and its structural diagram is as Figure 2 shown.

[0070] Through the above "business structuring", "object standardization", and "rule configuration", the goals of improving operation ability and decision-making ability are achieved, so as to ensure that all eligible customers are insured, and those who can be insured are insured quickly, improving the customer's insurance purchase experience.

[0071] In step S3, the multi-dimensional medical decision-making rules for underwriting are based on a pre-set rule flow, and perform multi-dimensional analysis and evaluation on the structured data after the standardization process, and automatically output an underwriting conclusion, such as Figure 5 shown. The implementation of this step is achieved on the insurance data processor. Its input data is the structured and standardized data processed by the data processing algorithm on the medical data processor. A multi-dimensional medical decision-making rule model for underwriting is built on the insurance data processor, and the structured and standardized data is subjected to multi-dimensional analysis and evaluation according to the pre-set rule flow to automatically generate an underwriting conclusion. Specifically:

[0072] The underwriting system receives the standardized medical structured data after the preliminary processing. These data cover various aspects such as the customer's basic information, disease diagnosis, examination and test results, and treatment conditions. The system performs a preliminary analysis on the input data according to the pre-set data structure, extracts the key information, and classifies and stores it in the corresponding logical area for subsequent analysis. Taking the customer's physical examination report data as an example, the basic information such as age and gender, the examination results such as blood pressure and blood sugar, and the disease diagnosis names are classified respectively, which is convenient for quick calling and processing.

[0073] Based on the pre-built disease classification and stratification system, the risk of the diseases suffered by the customer is evaluated. The diseases are classified according to the human body system, such as the cardiovascular system, the respiratory system, etc., and then further subdivided according to the disease severity, stage, treatment effect, etc. For the subdivided diseases, the system performs quantitative scoring according to the corresponding underwriting strategies and risk weights. Taking hypertension as an example, in the case of grade 1 hypertension without complications, a lower risk score is given according to the rules; while grade 3 hypertension with serious complications is given a higher risk score, so as to reflect the disease risk differences.

[0074] In addition to the disease itself, the system also comprehensively considers other risk factors of the customer, such as personal basic information (age, gender, occupation, lifestyle), family medical history, previous underwriting records, etc. Statistical analysis and machine learning algorithms are used to assign reasonable weights to each risk factor. Customers with older age, engaged in high-risk occupations, having a family genetic history, or previous records of declination or premium loading for underwriting have higher weights for their corresponding risk factors; customers with good lifestyles have lower weights for relevant risk factors. The scores of each risk factor are comprehensively calculated with the disease risk score to obtain the comprehensive risk assessment result of the customer. The underwriting conclusions include but are not limited to different types of underwriting conclusion texts such as standard underwriting, premium loading underwriting, exclusion underwriting, deferred underwriting, or declination underwriting, and are accompanied by reasons and bases for the decisions. And the underwriting conclusions and relevant data in the entire underwriting process are stored in the underwriting database, recording the modification history, access logs, and operation traces of the data; the relevant data in the underwriting process includes original medical materials, processing results of each module, and decision-making process logs.

[0075] In addition, due to the continuous changes in medical knowledge, industry experience, and regulatory policies, the underwriting rules need to be continuously updated. The system regularly pays attention to information such as medical research progress, updates of clinical practice guidelines, accumulation of industry underwriting experience, and adjustments of laws, regulations, and policies. When new disease diagnosis methods, treatment technologies, or risk factors emerge, the rule update process is promptly initiated. After the new rules are demonstrated by internal experts, tested and verified, and reviewed for compliance, they are incorporated into the multi-dimensional medical decision-making rule flow for underwriting. When analyzing and evaluating the input data, the system automatically operates according to the latest rules to ensure the scientificity and rationality of underwriting decisions. For example, if the development of gene detection technology provides a more accurate basis for the risk assessment of certain genetic diseases, the system will accordingly adjust the underwriting rules for related diseases, and adopt the updated risk assessment criteria when evaluating customers carrying such genetic risks.

[0076] To sum up, the present invention effectively overcomes multiple technical difficulties in the fields of underwriting and risk assessment by using multi-dimensional medical decision-making rules, realizes the innovation of the underwriting model through means such as "business structuring", "object standardization", and "rule configuration", replaces manual operations, greatly improves the efficiency of underwriting operations, makes the underwriting process simpler and faster, and the underwriting conclusions more accurate. Customers who originally had to wait for manual underwriting processing can quickly receive various underwriting notices or final underwriting conclusions, meeting the insurance protection and insurance service needs of the people. It reduces the errors and inconsistent scales that may occur in manual underwriting and lowers the error rate of underwriting.

[0077] In practical applications, the multi-dimensional medical decision-making rules achieve real-time interpretation of complex data such as customers' health declarations, physical examination reports, medical records, and previous underwriting conclusions, and are applied to actual underwriting operations. As of the first half of 2024, for the full-process intelligent underwriting, the substitution rate of underwriters' work has reached 28.5%, and the automatic underwriting review rate has reached 96.2%, leading the industry. The multi-dimensional medical decision-making rules have achieved partial manual substitution in three scenarios, such as customers' previous claims and physical examination reports, improving processing consistency and timeliness, saving a large amount of manual underwriting time, and thus bringing faster customer underwriting speed and better underwriting services.

[0078] It should be noted that the execution subject of each step of the method provided in the above embodiments can be the same device, or the method can also be executed by different devices. For example, the execution subjects of steps S1 and S2 can be device 1, and the execution subject of step S3 can be device 2; or, the execution subject of step S1 can be device 1, and the execution subjects of steps S2 and S3 can be device 2; and so on. As Figure 3 shown, devices 11, 12, and 13 are data processing devices of different subjects. Device 11 is an image acquisition terminal for collecting users' medical data, such as a mobile phone, which is used to collect image data of medical data files; device 12 is a medical data processing server of a medical institution, which is used to perform structured extraction on the image data based on the OCR model of the neural network according to the multi-dimensional medical decision-making rules for underwriting to obtain structured data in text form; device 13 is an underwriting data processing server of an insurance company. The multi-dimensional medical decision-making rules model on the underwriting data processing server performs multi-dimensional analysis and evaluation on the standardized structured data based on a pre-set rule flow and automatically outputs an underwriting conclusion.

[0079] As Figure 5 shown, it is a data flow diagram of the execution subject structure according to Figure 3 . First, device 11 (such as an image acquisition terminal) collects image data of medical data files. The data is medical data including but not limited to health declarations, physical examination reports, medical records, and previous underwriting conclusion data, and these data are all collected in the form of pictures to facilitate subsequent OCR recognition. At the same time, preprocessing is performed on the image data; the preprocessing includes image enhancement, grayscale transformation, Gaussian filtering, and Hough transform correction. After the medical data image data is collected, it is sent to the medical data storage server. The OCR model based on the neural network is used to perform structured extraction on the image data according to the multi-dimensional medical decision-making rules for underwriting to obtain structured data in text form, perform standardized processing on the structured data, and then transfer it to the underwriting data processing server. The multi-dimensional medical decision-making rules model on the underwriting data processing server performs multi-dimensional analysis and evaluation on the standardized structured data based on a pre-set rule flow and automatically outputs an underwriting conclusion.

[0080] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) that contain computer-usable program code.

[0081] Therefore, the present application also proposes a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the method described in any one of the embodiments of the present application.

[0082] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the specified functions in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0083] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means implements the specified functions in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0084] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the specified functions in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0085] Furthermore, the present application also provides an electronic device (or computing device), including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the method described in any embodiment of the present application is implemented.

[0086] In a typical configuration, a computing device includes one or more processors (CPUs), an input / output interface, a network interface, and memory. The memory may include non-permanent memory in the form of computer-readable media, random access memory (RAM), and / or non-volatile memory such as read-only memory (ROM) or flash memory (flash RAM). Memory is an example of computer-readable media. Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology for information storage. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transitory media such as modulated data signals and carrier waves.

[0087] It should also be noted that the term "comprising", "including" or any other variation thereof is intended to cover a non-exclusive inclusion, such that a process, method, article or apparatus comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or apparatus. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or apparatus comprising the element.

[0088] The above are only embodiments of the present application and are not used to limit the present application. For those skilled in the art, various changes and modifications can be made to the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included within the scope of the claims of the present application.

Claims

1. A multi-dimensional underwriting decision-making method for intelligent review and evaluation of health risks, characterized in that: include: Collect image data from medical documents and preprocess the image data; the preprocessing includes image enhancement, grayscale transformation, Gaussian filtering or Hough transformation correction; Using the OCR model, the image data is structured extracted according to the underwriting multidimensional medical decision-making rules to obtain structured data in text form, and the structured data is standardized and transmitted to the underwriting multidimensional medical decision-making rules; The multi-dimensional medical decision-making rules for underwriting are based on a pre-set rule flow, perform multi-dimensional analysis and evaluation on the standardized structured data, and automatically output underwriting conclusions.

2. According to claim 1, a multi-dimensional underwriting decision-making method for intelligent review and evaluation of health risks is characterized in that: The OCR model is used to extract the image data in a structured manner according to the multi-dimensional medical decision-making rules for underwriting to obtain structured data in text form, specifically: Select an OCR model based on convolutional neural network and recurrent neural network architecture, use the medical field text image dataset to train the OCR model, input the preprocessed image into the OCR model, recognize the text content, and clean the recognition results to remove garbled and repeated characters; The texts recognized and cleaned by the OCR model are input into the big language model. Based on the multi-dimensional medical decision-making rules of underwriting, natural language processing technology is used to mark key information including but not limited to the disease name, symptoms, examination items, treatment methods, and onset time. According to the underwriting business needs and medical knowledge system, a structured data framework is constructed, and framework information categories are set. Each category is subdivided into specific fields, and the key information marked by the large language model is filled into the corresponding fields to form structured data; Perform correlation verification on the information of each field in the structured data to check the logical consistency. For data with logical errors, missing or conflicting information, re-analyze the original text or interpret it again with the help of a large language model.

3. According to claim 1, a multi-dimensional underwriting decision-making method for intelligent review and evaluation of health risks is characterized in that: The standardization of structured data is specifically: converting the structured data in text form into standardized numerical values ​​or codes, normalizing synonyms, unifying numerical result units, and standardizing text result descriptions.

4. According to claim 1, a multi-dimensional underwriting decision-making method for intelligent review and evaluation of health risks is characterized in that: The multi-dimensional medical decision-making rules for underwriting include: Compile multiple medical dictionaries according to the business content and scope, standardize medical fields through the medical dictionaries, and set unique codes for each field; Standardize test results, numerical units, synonyms, and antonyms; Build a multi-level disease classification system, classify diseases into major categories according to the human body system, and then subdivide them according to the severity of the disease, the stage of the disease, and the treatment situation, and set corresponding underwriting evaluation rules and risk weights for each level.

5. According to claim 1, a multi-dimensional underwriting decision-making method for intelligent review and evaluation of health risks is characterized in that: The underwriting multi-dimensional medical decision rule performs multi-dimensional analysis and evaluation on the standardized structured data based on a pre-set rule flow, specifically: Receive standardized structured data, parse it according to the pre-set data structure, extract key information and classify and store it in the corresponding logical area; Based on the pre-built disease classification and stratification system, we conduct risk assessment on the diseases suffered by customers, classify diseases according to human body systems, and then subdivide them according to the severity, stage, and treatment effect of the disease. For the subdivided diseases, we quantify and score them according to the corresponding underwriting strategy and risk weight; In addition to the disease itself, we comprehensively consider other risk factors of the customer, including basic personal information, family medical history, and previous underwriting records. We use statistical analysis and machine learning algorithms to assign reasonable weights to each risk factor, and comprehensively calculate the scores of each risk factor and the disease risk score to obtain a comprehensive risk assessment result for the customer.

6. The multi-dimensional underwriting decision-making method for intelligent review and evaluation of health risks according to claim 5 is characterized in that: Also includes: We regularly pay attention to information including but not limited to medical research progress, clinical diagnosis and treatment guideline updates, industry underwriting experience accumulation, and adjustments to laws, regulations, and policies. When new disease diagnosis methods, treatment technologies, or risk factors emerge, we initiate the rule update process. After internal expert demonstration, testing and verification, and compliance review, the new rules are integrated into the underwriting multi-dimensional medical decision-making rule flow.

7. According to claim 1, a multi-dimensional underwriting decision-making method for intelligent review and evaluation of health risks is characterized in that: The underwriting conclusion includes but is not limited to different types of underwriting conclusion texts, such as standard coverage, premium coverage, excluded coverage, deferred coverage or rejection of coverage, with the reasons and basis for the decision attached.

8. The multi-dimensional underwriting decision-making method for intelligent review and evaluation of health risks according to claim 7 is characterized in that: It also includes storing the underwriting conclusion and relevant data in the entire underwriting process into the underwriting database, recording the data modification history, access log and operation track; the relevant data in the underwriting process includes original medical information, processing results of each module, and decision-making process log.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

10. An electronic 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, the method according to any one of claims 1 to 8 is implemented.

Images