Physical examination add-on recommendation method and device

Abstract

The application discloses a physical examination item recommendation method and device, and relates to the technical field of medical treatment and artificial intelligence. A specific embodiment of the method comprises the following steps: in response to a recommendation request transmitted by a client, collecting multi-source information of a user to construct a portrait snapshot, inputting the portrait snapshot into a health intelligent agent to determine a health risk part; searching a preset item library for a plurality of physical examination items corresponding to the health risk part and satisfying a preset recommendation condition, and sorting the plurality of physical examination items based on a single customer price hierarchical label in the portrait snapshot; generating a recommendation card corresponding to each physical examination item, and sending the health risk part and the recommendation card of each physical examination item associated with the health risk part to the client for sequential display; and in response to a transaction instruction for a target physical examination item in one or more recommendation cards transmitted by the client, executing a transaction process for the target physical examination item. The embodiment improves the personalized recommendation conversion rate of physical examination items and the user satisfaction.

Description

Technical Field

[0001] This invention relates to the fields of medical and artificial intelligence technologies, and in particular to a method and apparatus for recommending additional items in a physical examination. Background Technology

[0002] In the health checkup service sector, to increase the average order value and package coverage, users are often offered the option to purchase individual health checkup items (i.e., add-ons) in addition to the basic package. However, existing health checkup add-on recommendation schemes generally suffer from a low degree of matching between the recommended results and the user's actual health risk areas and consumption preferences; at the same time, the recommended results lack a structured information display medium and fail to form an effective closed loop with the actual add-on transaction process, resulting in high online decision-making costs for users and difficulty in improving the overall personalized recommendation conversion rate. Summary of the Invention

[0003] In view of this, embodiments of the present invention provide a method and apparatus for recommending additional items in physical examinations, which can at least solve the problems in the prior art where the recommendation results have a low degree of matching with the user's actual health risk sites and consumption preferences, lack a structured information display carrier, and fail to form an effective closed loop with the actual additional item transaction link.

[0004] To achieve the above objectives, according to one aspect of the present invention, a method for recommending additional items in a physical examination is provided, comprising: In response to the recommendation request transmitted by the client, multi-source information of the user is collected to construct a profile snapshot, and the profile snapshot is input into the health intelligent agent to determine the health risk sites; In the preset add-on item library, multiple physical examination items that correspond to the health risk sites and meet the preset recommendation conditions are retrieved, and the multiple physical examination items are sorted based on the customer price stratification tags in the profile snapshot; Generate recommendation cards corresponding to each of the aforementioned physical examination items, and send the health risk sites and the associated recommendation cards for each of the aforementioned physical examination items to the client for sequential display; In response to a transaction instruction transmitted by the client for a target health checkup item in one or more recommended cards, the transaction process for the target health checkup item is executed.

[0005] To achieve the above objectives, according to another aspect of the present invention, a device for recommending additional medical examination items is provided, comprising: The risk assessment module is used to respond to recommendation requests transmitted by the client, collect multi-source information of users to construct a profile snapshot, and input the profile snapshot into the health intelligent agent to determine the health risk sites; The add-on recommendation module is used to retrieve multiple physical examination items that correspond to the health risk sites and meet the preset recommendation conditions from the preset add-on item library, and to sort the multiple physical examination items based on the customer price stratification tags in the profile snapshot; The recommendation card module is used to generate recommendation cards corresponding to each of the physical examination items, and send the health risk sites and the associated recommendation cards of each of the physical examination items to the client for sequential display. The add-item transaction module is used to execute the transaction process for the target physical examination item in response to the transaction instruction transmitted by the client for one or more recommended cards.

[0006] To achieve the above objectives, according to another aspect of the present invention, an electronic device for recommending additional items in physical examinations is provided.

[0007] The electronic device of this invention includes: one or more processors; and a storage device for storing one or more programs, which, when executed by the one or more processors, cause the one or more processors to implement any of the above-described methods for recommending additional items for physical examinations.

[0008] To achieve the above objectives, according to another aspect of the present invention, a computer-readable medium is provided having a computer program stored thereon, which, when executed by a processor, implements any of the above-described methods for recommending additional medical examination items.

[0009] To achieve the above objectives, according to another aspect of the present invention, a computing program product is provided. One such computing program product includes a computer program that, when executed by a processor, implements the physical examination supplement recommendation method provided in this embodiment of the invention.

[0010] According to the solution provided by the present invention, one embodiment of the invention has the following advantages or beneficial effects: by integrating multi-source information to construct a user profile snapshot, introducing a health intelligence agent to accurately locate the user's health risk areas, and introducing a tiered tag based on average order value to intelligently sort recommended physical examination items, it not only ensures that the recommended add-ons are highly consistent with the user's actual health needs and consumption intentions, but also significantly reduces the user's online decision-making costs through structured card display and seamless transaction links, realizing an efficient business closed loop from "risk discovery" to "add-on conversion", and significantly improving the personalized recommendation conversion rate of physical examination add-ons and user satisfaction.

[0011] The further effects of the aforementioned unconventional alternative methods will be explained below in conjunction with specific implementation methods. Attached Figure Description

[0012] The accompanying drawings are provided to better understand the invention and are not intended to unduly limit the scope of the invention. Wherein: Figure 1 This is a schematic diagram of the main process of a method for recommending additional items in a physical examination according to an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the specific process for recommending additional items during a physical examination; Figure 3(a) is a schematic diagram showing the recommended cards for health risk sites and related physical examination items; Figure 3(b) is a schematic diagram showing the health risk sites and the entry points for additional items; Figure 3(c) is a schematic diagram showing the recommendation cards for each physical examination item after clicking the "Add Items" entry; Figure 4 This is a schematic diagram of the overall framework for recommending additional items in a physical examination; Figure 5 This is a schematic diagram of the main modules of a physical examination supplement recommendation device according to an embodiment of the present invention; Figure 6 This is an exemplary system architecture diagram in which embodiments of the present invention can be applied; Figure 7 This is a schematic diagram of the structure of a computer system suitable for implementing the embodiments of the present invention, such as a mobile device or server. Detailed Implementation

[0013] The following description, in conjunction with the accompanying drawings, illustrates exemplary embodiments of the present invention, including various details to aid understanding. These details should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the invention. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0014] It should be noted that in the embodiments of this application, certain software, components, models and other existing solutions in the industry may be mentioned. These should be regarded as exemplary and are only intended to illustrate the feasibility of implementing the technical solution of this application. However, it does not mean that the applicant has used or necessarily used the solution.

[0015] Where there is no conflict, the embodiments and features in the embodiments of this invention can be combined with each other. The acquisition, transmission, storage, use, and processing of data in the technical solutions of this invention comply with the relevant provisions of national laws and regulations, are used for legal and reasonable purposes, and are not shared, disclosed, or sold outside of these legal uses, and are subject to supervision and management by regulatory authorities.

[0016] Regarding user information, necessary measures should be taken to prevent unauthorized access to such personal information data, ensure that personnel authorized to access such data comply with relevant laws and regulations, and safeguard the security of user personal information. Once this user personal information data is no longer needed, risks should be minimized by restricting or even prohibiting data collection and / or deleting the data. Where applicable, including in certain relevant applications, user privacy should be protected through data de-identification, such as by removing specific identifiers (e.g., date of birth), controlling the amount or specificity of stored data (e.g., collecting location data at the city level rather than the specific address level), controlling how data is stored, and / or other de-identification methods.

[0017] The technical terms used in this solution are explained as follows: B-end users / C-end users: B-end users refer to users who undergo physical examinations through corporate group check-up channels; C-end users refer to users who undergo physical examinations through individual retail channels.

[0018] Average order value tiering: This refers to classifying users into three tiers—low, medium, and high—based on their historical transaction prices for additional services during health checkups, and using this as the routing basis for recommending additional services.

[0019] Profile snapshot: refers to an immutable set of user features corresponding to a single recommendation request, which is assembled in real time from multiple sources of health data. It is identified by a unique snapshot ID and is used to achieve state isolation among multiple cards.

[0020] Health Intelligent Agent: refers to a server-side component built on a large language model that can make risk assessments and add suggestions based on user profile snapshots.

[0021] Degradation strategy: When a downstream service (such as a health agent or a backup agent) fails, the front end skips non-critical modules (such as the add-item recommendation process), but retains the main process button (such as the order button) as available.

[0022] Community tags: These are dynamic text tags formed by performing TopK statistics on the distribution of additional selections within the same user group based on dimensions such as "gender, age group, enterprise, and average order value".

[0023] The existing recommended additional items for physical examinations mainly include the following: The first type is offline manual recommendation. This involves the user arriving at the medical examination center and the on-site guide verbally recommending several add-on services based on the user's complaints. The user then decides whether to purchase these services and completes the transaction on the spot. This method relies entirely on the personal experience and time of the on-site guide, lacking the intervention of a digital system and data accumulation.

[0024] The second type is the self-service add-on list. This involves displaying all available add-on items in a list format on the health checkup package purchase page or the appointment details page after successful booking. Users can then select items by name and price. The recommendation logic for this method is usually based solely on sales volume, price, or simple category, without providing differentiated recommendations for individual users.

[0025] The third category is recommendations based on a single health questionnaire or historical medical examination report. This means that a separate health questionnaire or a historical medical examination report can be configured at the entry point of the add-item page, and the system will output add-item suggestions in one go based on preset rules (such as recommending cardiovascular examination items if the questionnaire indicates hypertension).

[0026] The fourth category is health-related question-answering agents based on large models. In recent years, some applications have introduced health-related large model dialogue capabilities, providing suggestions for additional items when users actively ask questions. However, their output has not yet been integrated with the actual fulfillment chain of adding items to the physical examination (such as adding to the cart - making an appointment - payment - offline examination), and there is no fallback strategy for abnormal situations such as timeout, failure, or no return in the agent's output.

[0027] The above-mentioned recommended additional medical examination items have the following main technical defects: 1) Health information is limited to a single dimension, making it impossible to construct a complete user profile snapshot. Existing technologies typically rely on a single health questionnaire, a single historical medical examination report, or basic demographic attributes (such as gender and age), failing to structurally integrate multi-source heterogeneous data such as "questionnaire answers, historical medical examination reports, demographic attributes, historical additional orders, and average order value stratification tags." This results in an inability to accurately and stably depict users' potential health risks and consumption preferences.

[0028] 2) The decoupling of recommendation strategies from user willingness to pay leads to low conversion rates. The existing self-service add-on lists and single-rule recommendations are pushed directly on a one-to-one basis according to "risk-item," without considering users' average order value (AOV) tiers. This results in users with different AOV tiers receiving the same recommended health check items, easily causing users with low AOV tiers to abandon the service due to price resistance, while users with high AOV tiers feel the recommendations are too conservative. Overall, the add-on conversion rate (approximately 32.5% for C-end users and approximately 23.6% for B-end users) is far below business expectations.

[0029] 3) The lack of a feedback mechanism for historical check-up status and a closed loop for health education results in a low repurchase rate. The existing solution does not write back the "historically checked items" into the recommendation model (i.e., it lacks the mutual exclusion and decay logic of check-up time intervals), leading to frequent low-quality recommendations such as "repeatedly recommending the same high-cost item in a short period of time." At the same time, there is a lack of periodic health intervention and education strategies for users with weak health awareness, resulting in a consistently low overall annual repurchase rate (only 22.2% in actual C-end annual repurchase rate).

[0030] See Figure 1 The diagram shows the main flowchart of a method for recommending additional items in a physical examination according to an embodiment of the present invention, which includes the following steps: S101: In response to the recommendation request transmitted by the client, collect multi-source information of the user to construct a profile snapshot, and input the profile snapshot into the health intelligent agent to determine the health risk sites; S102: In the preset add-on item library, retrieve multiple physical examination items that correspond to the health risk sites and meet the preset recommendation conditions, and sort the multiple physical examination items based on the customer price stratification tags in the portrait snapshot; S103: Generate recommendation cards corresponding to each of the physical examination items, and send the health risk sites and the associated recommendation cards of each of the physical examination items to the client for sequential display; S104: In response to a transaction instruction transmitted by the client for a target physical examination item in one or more recommended cards, execute the transaction process for the target physical examination item.

[0031] In the above implementation, for step S101, the physical examination business nodes include, but are not limited to: physical examination appointment node, on-site check-in node, in-hospital guidance node, payment success node, report viewing node, and IM (Instant Messaging) intelligent agent interaction node. The report viewing node refers to the node triggered by the user to view reports for some physical examination items after they are issued in batches during the physical examination process. When a user clicks on any physical examination node in the client (App, Mini Program, H5), the client responds by generating a recommendation request to trigger data collection operations at the server's multi-source information collection layer.

[0032] The multi-source information collection layer includes a questionnaire segmentation collector, a historical physical examination report uploader and synchronizer, and a user demographic attribute / order reader. Among them: 1. The questionnaire segmentation data collector is used to divide a pre-set question bank into different segments according to different dimensions (such as basic population, lifestyle, past medical history, and symptom complaints). Based on the "different priorities configured for different segments in advance, and the pre-configured 'node-priority' mapping relationship", the segment corresponding to each physical examination node is determined to achieve the purpose of low-intrusion and gradual data collection. In one optional implementation, the number of priorities can be equal to the number of nodes.

[0033] For example, basic population and lifestyle are prioritized as TOP1, past medical history as TOP2, and symptoms and complaints as TOP3. Different priorities are assigned according to the importance of each examination node, such as appointment node - TOP1 priority segment, on-site check-in node - TOP2 priority segment, and in-hospital guidance node - TOP3 priority segment. Based on the segment corresponding to each examination node, 1-3 questions are extracted from that segment to construct a health questionnaire for each examination node. The final results are shown in Table 1. Table 1:

[0034] In one optional implementation, for the same physical examination node, the system displays a uniformly configured health questionnaire to the user (such as basic questions about smoking status, alcohol consumption, and exercise habits). In another optional implementation, the system can dynamically extract questions from a segmented question bank based on the user's demographic attributes (such as age, gender, marital status, etc.) to generate personalized health questionnaires for different users. Considering that a globally unified configuration approach can effectively reduce system maintenance costs and computational overhead, this invention preferably adopts a globally unified configuration health questionnaire.

[0035] Current technologies generally design health questionnaires as long, independent entry points, requiring users to actively access and complete them in one go. This highly interactive approach results in an extremely low questionnaire completion rate (only 2.2% in our tests (578 / 26761)), making subsequent personalized add-on recommendations lack sufficient data support. To address this issue, this solution, in its practical application phase, uses a questionnaire segmentation collector to identify the user's current physical examination node, determine and display the corresponding health questionnaire for that node (e.g., displaying the corresponding health questionnaire at the physical examination appointment node), and receive the user's answers to that health questionnaire. If the questionnaire is not completed, the user is also shown the questionnaire collection entry point for continued data collection. This approach solves the problems of "overly intrusive health information collection and low data acquisition rate" in existing technologies.

[0036] 2. Historical Medical Examination Report Uploader and Synchronizer: This feature retrieves a user's historical medical examination reports. It should be noted that these reports can be from other organizations where the user has had medical examinations; different organizations may have different report formats. If a user has consistently had medical examinations at their current organization, their historical medical examination reports can be retrieved based on their user identifier (such as their ID number).

[0037] 3. User Demographic Attributes / Order Reader: Used to obtain the user's demographic attributes and historical add-on orders. Demographic attributes can be obtained through age range and gender parsed from the ID number, and marital status can be inferred from the marital status field (e.g., selecting "Married Female Package" for the medical examination). Historical add-on orders differ from historical medical examination orders; they are orders generated from previous purchases of medical examination items. This solution only considers historical add-on orders made by the user within their current workplace, which can be obtained based on the user's identity identifier. The primary focus here is to obtain the transaction price of these historical add-on orders.

[0038] After acquiring multi-source information about the user, a user profile snapshot is constructed based on this information. This operation is performed by the server-side health profile fusion layer. The health profile fusion layer receives multi-source information collected by the multi-source information acquisition layer and performs a normalization operation on the multi-source information through a multi-source signal normalization module. This solves the problem of existing systems where "health information has a single dimension, making it impossible to construct a complete user profile snapshot." Specifically: 1. Normalize the questionnaire answers into standard features according to data type, such as enumeration, numerical, or range. For example, normalize the answer to "Have you ever smoked?" into an enumeration feature (such as "Yes", "No", "Quit smoking"), normalize the answer to "Daily smoking amount" into a numerical feature (such as "10 cigarettes / day"), and normalize the answer to "Age" into a range feature (such as "30-40 years old").

[0039] 2. Using methods such as OCR (Optical Character Recognition) and structured parsing, extract positive items and their corresponding historical examination dates from historical medical examination reports. For example, from unstructured historical medical examination report files, identify and extract "thyroid nodule (TI-RADS 3)" as a positive item, and simultaneously extract its corresponding historical examination date (e.g., 2023-05-12).

[0040] 3. Determine the grouping buckets for demographic attributes, such as cross-grouping by age range (18-35 years old, 35-50 years old, 50 years and older) and gender. For example, users can be divided into different grouping buckets such as "males aged 18-35", "males aged 35-50", and "males aged 50 and older".

[0041] 4. Using the average order value (AOV) tiering tagger, determine the target range based on the transaction prices of historical add-on orders, and then determine the corresponding AOV tiering tag for that range. For example, high AOV (>500 yuan), medium AOV (200~500 yuan), and low AOV (<200 yuan). The target range is determined based on AOV. The AOV can be determined by: determining the transaction time of each historical add-on order, selecting the N (e.g., 10) historical add-on orders closest to the current time, summing the transaction prices of these N historical add-on orders, and averaging the sum to obtain the average AOV.

[0042] Alternatively, the highest transaction price among all historical add-on orders can be used as the average order value. Or, the transaction price of the user's current purchase of a health checkup package can be used as the average order value. Another option is to input these historical add-on orders and transaction prices into a consumer behavior model (which can be built using techniques known to those skilled in the art) to predict the average order value. These methods of calculating the average order value essentially all incorporate the user's willingness to pay dimension and are interchangeable in effect.

[0043] In one alternative implementation, questionnaire answers, historical medical examination reports, demographic attributes, etc., can be directly encoded into latent vectors using a single vector model (such as an Embedding model finely tuned for the medical field), without explicitly constructing structured features, and can achieve the same effect. There is no limitation on this.

[0044] Finally, the aforementioned multi-dimensional features are written into the user profile snapshot, generating a unique profile snapshot ID. This user profile snapshot is then sent to the health agent, requesting it to return a health risk list and an additional recommendation list within a timeout threshold T. If the health agent's return result includes a health risk list, the user's health risk sites, such as the thyroid and liver, can be obtained from it. Here, the health agent can pre-set prompts, similar to a fill-in-the-blank method, filling the profile snapshot into the prompts to generate structured text. This structured text, combined with preset constraints (such as word count constraints and judgment constraints), is then input into the health agent.

[0045] For step S102, the results returned by the health agent may vary. Different branches are executed based on the different results returned by the health agent, thereby solving the problem of "decoupling between recommendation strategies and user willingness to pay, resulting in low conversion rates" in existing technologies. See the overall process below. Figure 2 As shown, this process illustrates the complete decision-making path from when the user triggers the health check node to when the additional recommendation card is rendered and an order is placed.

[0046] Branch 1: The health agent outputs a list of health risks and a list of recommended additional items. The health risk list and additional recommendation list output by the health intelligence agent, along with the average order value (AOP) tiered tags from the profile snapshot, are input into the AOP-aware recommendation engine. The AOP-aware recommendation engine extracts the health risk sites (e.g., liver) from the health risk list and the corresponding medical examination items (e.g., 5 liver function tests) from the additional recommendation list.

[0047] The "Dual-Price Candidate Generator for Same Risk" retrieves at least two health checkup items from a pre-defined add-on item library that target the specific health risk area and meet pre-defined recommendation criteria, such as comparing a 3-item liver function test with a 5-item liver function test, or comparing DR (Digital Radiography) with CT (Computed Tomography). Pre-defined recommendation criteria include, but are not limited to, significant price differences and high order frequency (to avoid recommending less popular checkup items to users). Alternatively, based on the checkup items in the add-on item recommendation list corresponding to the specific health risk area, the pre-defined add-on item library retrieves other health checkup items targeting the specific health risk area but with significantly different prices and high order frequency.

[0048] Besides considering dual-priced health checkup items, in practical applications, tri-priced or multi-priced health checkup items can also be considered, or comparisons of two body parts within the same price range can be performed. For example, for the same health risk area, three different price levels of health checkup items—basic, advanced, and flagship—can be provided for users to choose from. Then, duplicate health checkup items corresponding to that health risk area in the recommended list and the retrieved health checkup items are deduplicated to obtain the final recommended health checkup items, such as comparing 3 liver function tests with 5 liver function tests, or comparing DR with CT scans.

[0049] Branch 2: The health agent only outputs a list of health risks: The server adopts a fallback strategy, which can also rely on a price-per-order-value-based recommendation engine. It extracts the health risk area (e.g., liver) from the health risk list and checks if the user's currently purchased health check package contains any checkup items corresponding to that health risk area, thus detecting any coverage gaps. If the currently purchased health check package does not contain any checkup items corresponding to that health risk area, it retrieves at least two checkup items from a pre-set add-on library that target that health risk area and meet the pre-set recommendation criteria to avoid an empty page.

[0050] Branch 3: The health agent call times out or fails, meaning no health risk list and / or additional recommendation list are returned within the timeout threshold T. In this case, a fallback strategy is adopted, assuming that the user does not currently have any health risk areas, directly terminating the additional recommendation process while retaining the main order entry point (such as a button) to ensure that the core ordering link is not disrupted; simultaneously, an alarm event is reported to the backend monitoring system.

[0051] In existing solutions that incorporate large health-related models or agents, when the agent returns a timeout, fails to return a recommendation, or only returns a risk assessment without additional suggestions, it can cause the front-end page to become blank or loading to be blocked, leading to user churn. Branch 3 of this solution effectively solves this problem, preventing business process interruptions when the health agent malfunctions.

[0052] In one optional implementation, a retry mechanism can also be implemented (for example, setting the maximum number of retries to 5) or a backup agent can be switched to call the healthy agent; if the retry result or the execution result of the backup agent is still a call timeout or an abnormal output, the item recommendation process is terminated. It should be noted that the training data of the backup agent can be consistent with that of the healthy agent, but the underlying model architecture or deployment environment can be different to improve disaster recovery and backup capabilities.

[0053] For branches 1 and 2, the recommended health checkup items also need to be sorted based on the average order value (AOV) tier tag. For example, when recommending two health checkup items (e.g., comparing cost-effectiveness and completeness), if the user's AOV tier tag is "low," the lower-priced item will be ranked higher; if the user's AOV tier tag is "high," the more complete item will be ranked higher. Furthermore, based on health risk areas and information associated with those areas in the user profile snapshot, the copywriting AI can be invoked to generate recommendation text for each health checkup item (or just the first item). For example, a "cost-effectiveness" reason can be added to the lower-priced item, and a tag like "for a more thorough checkup" can be added to the higher-priced item.

[0054] For branch 1 and branch 2, filtering and / or expansion operations can also be performed: 1. Filtering Operation: Through the "Recently Tested Negative Filter", based on the user's historical physical examination reports, the latest (i.e. most recent) examination date of the physical examination items corresponding to the health risk areas is determined. The time difference between the latest examination date and the current date is calculated, and physical examination items with the same name that have a time difference less than the preset time threshold D (such as 90 days) are filtered out, so as to avoid low-quality duplicate recommendations such as "recommending CT when a CT was just done 3 months ago".

[0055] 2. Extended Operations: Using the "Community Group Tag Generator," based on dimensions such as group buckets and average order value stratification tags in the user profile snapshot (and possibly the user's company ID), determine the target user group to which the user belongs. In the offline statistical distribution table of group-added items for this target user group, retrieve the historical order volume of each health checkup item, and filter the Top K health checkup items in descending order of historical order volume as recommended health checkup items. For example, add C to the existing health checkup items A and B.

[0056] The distribution table of items added in the same group can be changed from offline T+1 statistics to streaming real-time statistics; the tag dimensions can also be expanded from "gender / age / company ID" to "occupation / work schedule / historical positive clusters". Furthermore, based on historical order data, dynamic tags such as "80% of male users have selected", "70% of users aged 35-45 have selected", and "60% of your colleagues have added this item" can be generated and attached to the physical examination items.

[0057] For step S103, existing recommendation cards typically only display the "item name and price," lacking core decision-making information that users care about when seeking medical treatment offline, such as "why it should be done, how to perform the examination, what the benefits are, and the basis for its credibility." This makes it difficult for users to understand the value of the item and develop trust online, forcing the addition of items to be postponed offline and resulting in a lost opportunity for online conversion.

[0058] To address this issue, this solution provides a five-element card for recommendation. The five-element card rendering layer assembles structured card data according to five fixed slots: "What / Why / How to check / What are the benefits / Is it trustworthy", and establishes a binding relationship between the recommendation card and the user profile snapshot ID.

[0059] The "What is it?" field is for filling in basic information such as the name and description of the physical examination items. The "Why" slot: Fill in the health risk list output by the health agent with the corresponding health risk location and cause for this physical examination item. If no risk cause exists, a preset catch-all text can be used; The "How to check" field should be filled with information about the examination process for this physical examination item. The "What are the advantages?" section should be filled with the advantages of this health checkup item (such as painless, non-invasive, advanced equipment, etc.). The "Trustworthy" slot is filled with the doctor's qualification information for the physical examination and feedback from other users on the physical examination (such as community tags).

[0060] For example, assuming a user identifies liver risk, the average order value tier is labeled as low, and the five elements are filled in as follows: What it is: Liver function tests (ALT (Alanine Aminotransferase) / AST (Aspartate Aminotransferase) / GGT (Gamma-Glutamyl Transferase)). Why: Your questionnaire indicates a long-term drinking habit, and combined with elevated ALT levels in previous reports, we recommend that you have a liver function test. How it is done: Blood test, no fasting required, takes about 5 minutes to complete; Advantages: Non-invasive, painless, and allows for early detection of fatty liver and liver fibrosis; Is it credible? Recommended doctor: Wang XX (Chief Physician of Hepatology Department, Top-Tier Hospital); 80% of men in the same age group have already chosen this option.

[0061] The above method solves the problems of "lack of decision support information for recommendation cards and high user decision-making costs" in existing technologies. Even if a user triggers different health check nodes multiple times (e.g., entering the appointment, check-in, and guidance nodes at different times, with each node triggering a recommendation), the different multi-source information collected at each check node leads to different profile snapshots and subsequent process execution results, resulting in different generated recommendation cards.

[0062] Furthermore, even if a user is pushed multiple recommendation cards in a short period of time, since each recommendation card is bound to an independent profile snapshot ID, when the user triggers a new physical examination node, the answers to the health questionnaire will not be written back to overwrite the recommendation cards of the previous physical examination node. This achieves multi-card state isolation and avoids the situation where "in the scenario where multiple questionnaire cards are triggered in parallel, the answer data of the latest card is easily mistakenly synchronized to the historical card, resulting in the corruption of the historical snapshot data".

[0063] In this solution, the server sends recommendation cards for the health risk sites and associated physical examination items to the client for display to the user. These recommendation cards are displayed directly in a pop-up window. Alternatively, only the health risk sites and associated add-on entry points are displayed. In response to the user's triggering of the add-on entry point, the client then requests the server to display recommendation cards for the physical examination items associated with that health risk site. For example, taking the liver as a health risk site, in Figure 3(a), the pop-up page directly displays the recommendation cards corresponding to liver function items 3 and 5; while in Figure 3(b), the pop-up page only displays the recommendation cards corresponding to liver function items 3 and 5 after the user triggers the add-on entry point, as shown in Figure 3(c).

[0064] For step S104, users can click on one or more target health checkup items in the recommended cards to perform transaction operations such as adding to cart, placing an order, and making payment (e.g., clicking the "Add to Items" button). After successful payment, the server will update the user's personal health record with the target health checkup items selected by the user, the corresponding health risk information (such as the health risk sites and causes displayed in the recommended cards), and the profile snapshot ID, as input for "Recently Tested" data in subsequent recommendations.

[0065] It should be noted that in this solution, users perform interactive operations on the client side, such as triggering the physical examination node, entering questionnaire answers, uploading historical physical examination reports, triggering the additional items entry, and selecting recommended cards. All other data processing and personal health record storage are deployed on the server side. The client and server communicate via HTTPS (Hypertext Transfer Protocol Secure). The server asynchronously consumes order events through a message queue to update personal health records. See also... Figure 4 This is a schematic diagram of the overall architecture of the present invention, showing the connection relationship between the multi-source information acquisition layer, the health profile fusion layer, the average order value perception recommendation engine, the intelligent agent reasoning layer, the anomaly degradation module, the five-element card rendering layer, and the physical examination fulfillment link.

[0066] It is understood that the specific values ​​listed in the above hypothetical description of the recommended additional items for physical examination are only illustrative examples and not restrictive. In practice, other values ​​can be configured according to needs. Similarly, the specific values ​​listed in the subsequent embodiments are only examples and are not the only limitation.

[0067] The method provided in the above embodiments integrates multi-source information to construct a user profile snapshot, introduces a health intelligence agent to accurately locate the user's health risk areas, and introduces a tiered tag based on average order value to intelligently sort recommended physical examination items. This not only ensures that the recommended add-ons are highly aligned with the user's actual health needs and consumption intentions, but also significantly reduces the user's online decision-making costs through structured card display and seamless transaction links. It achieves an efficient business closed loop from "risk discovery" to "add-on conversion," significantly improving the personalized recommendation conversion rate of physical examination add-ons and user satisfaction.

[0068] To address the problems in existing technologies, such as "low matching degree between recommendation results and users' actual health risks and consumption preferences, lack of structured information display carriers, and failure to form an effective closed loop with the add-on transaction link," this invention provides a method for recommending add-on items for physical examinations. This method effectively improves the add-on conversion rate, questionnaire completion rate, and user repurchase rate by integrating multi-source information, perceiving user average order value stratification tags, collecting questionnaire answers in a low-intrusive manner, and providing business fallback for intelligent agent anomalies. Simultaneously, it ensures that the add-on fulfillment link is not interrupted under any abnormal circumstances. This invention has the following beneficial effects: 1. This invention enables low-intrusion, gradual collection of user information from multiple sources, significantly improving questionnaire completion rates. The original independent health questionnaire is broken down into several segments according to different dimensions. Based on priority, the correspondence between each questionnaire segment and a physical examination node is determined, and these segments are embedded into the corresponding physical examination nodes for display and data collection. This method effectively replaces the traditional independent, long questionnaire entry point, reducing the user's operational burden and resistance, achieving low-intrusion, gradual data collection, and providing a sufficient data foundation for subsequent accurate recommendations.

[0069] 2. Constructing a dual-axis user profile snapshot based on health and consumption to improve the sufficiency and accuracy of recommendation data. This invention constructs a user profile snapshot based on multi-source information fusion, and incorporates the transaction prices of historical add-on orders during the construction process to determine the average order value stratification tags. Through this dual-axis profile snapshot construction method, the user's health risk characteristics and consumption preference characteristics are normalized and expressed in the same profile snapshot, ensuring that each subsequent add-on recommendation has comprehensive and objective factual basis, effectively solving the problem of decoupling recommendation strategies from user willingness to pay.

[0070] 3. Provide differentiated recommendations and a high-availability fallback mechanism based on tiered tags for average order value to improve overall conversion rate and process stability. This invention introduces a health-focused intelligent agent to process user profile snapshots. When the agent outputs a health risk list normally, it generates a "dual-price comparison" recommendation strategy for the same health risk area and intelligently sorts the recommendations based on tiered tags for average order value (for example, low-average-order-value users are prioritized for cost-effective items, while high-average-order-value users are prioritized for highly complete items). This breaks the limitations of the traditional one-to-one "risk-item" recommendation and accurately matches the psychological expectations of users at different consumption levels. When the agent call times out or malfunctions, a fallback strategy is adopted to retain the main order entry point, ensuring that the core transaction process is not interrupted and guaranteeing the high availability of the system.

[0071] 4. Construct a multi-dimensional filtering and rich information display mechanism, and establish a data write-back closed loop to effectively improve repurchase rates in the long term. This invention introduces a recently tested negative filter to effectively avoid the low-quality situation of repeatedly recommending the same test item in a short period of time; at the same time, it introduces a community group tag generator to expand recommendations by utilizing the high-frequency order items of users in the same group. In addition, each recommendation card is configured with structured information (five elements) containing core decision-making elements, and dynamic community tags are superimposed, so that the information density and decision support capability of online add-on cards are aligned with offline doctor consultations, greatly reducing the user's decision-making cost. Finally, after the transaction is completed, the add-on data, health risk information, and profile ID are written back to the individual's health record, forming a data closed loop with the next round of recommendations, and establishing a long-term repurchase enhancement mechanism.

[0072] See Figure 5The diagram shows the main modules of a physical examination supplement recommendation device 500 provided in an embodiment of the present invention, including: The risk assessment module 501 is used to respond to the recommendation request transmitted by the client, collect multi-source information of the user to construct a profile snapshot, and input the profile snapshot into the health intelligent agent to determine the health risk sites; The add-item recommendation module 502 is used to retrieve multiple physical examination items that correspond to the health risk sites and meet the preset recommendation conditions from the preset add-item library, and to sort the multiple physical examination items based on the customer price stratification tags in the profile snapshot; The recommendation card module 503 is used to generate recommendation cards corresponding to each of the physical examination items, and send the health risk sites and the associated recommendation cards of each of the physical examination items to the client for sequential display. The add-item transaction module 504 is used to execute the transaction process for the target physical examination item in response to the transaction instruction transmitted by the client for one or more recommended cards.

[0073] In the device of this invention, the health risk site is determined based on the health risk list output by the health intelligence agent, and the additional item recommendation module 502 is further used for: In response to the health intelligence agent outputting a health risk list and an additional item recommendation list, the physical examination items corresponding to the health risk sites are extracted from the additional item recommendation list, and deduplication is performed with the retrieved multiple physical examination items to obtain the final recommended physical examination items; In response to the health intelligence agent only outputting a list of health risks, if the user's current physical examination order does not include physical examination items corresponding to the health risk sites, the retrieved multiple physical examination items will be determined as the final recommended physical examination items.

[0074] The apparatus for implementing this invention further includes an exception handling module, used for: In response to a timeout or output error in the health agent call, a retry operation is performed and / or a backup agent of the health agent is called. If the execution result is still a timeout or output error, the item recommendation process is terminated.

[0075] In the apparatus of this invention, the multi-source information includes the following: questionnaire answers, historical medical examination reports, demographic attributes, and historical additional order information; the profile snapshot is constructed based on the following: The questionnaire answers were normalized into standard features according to data type. Extract positive items and their corresponding historical examination dates from the historical medical examination reports; Determine the grouping bucket to which the demographic attribute belongs; The average order value tier labels are determined based on the transaction prices of the historical add-on orders.

[0076] In the device of this invention, the recommendation request is generated by the client in response to the user triggering the physical examination node; The process of obtaining the questionnaire answers includes: determining the health questionnaire corresponding to the physical examination node and sending it to the client for display, so as to receive the questionnaire answers input by the user transmitted by the client; The process of constructing the health questionnaire includes: The questionnaire segmentation data collector splits the pre-set question bank into different segments according to different pre-configured dimensions. Determine the priority of each shard in advance, and determine the shard corresponding to each shard based on the pre-configured mapping relationship between the shard and the priority. A predetermined number of questions are extracted from the segments corresponding to each of the physical examination nodes to construct a health questionnaire for each of the physical examination nodes.

[0077] In the apparatus of the present invention, the item recommendation module 502 further includes at least one of the following: Based on the historical physical examination reports, determine the latest examination date of the health risk area, calculate the time difference between the latest examination date and the current date, and filter out physical examination items with the same name whose time difference is less than a preset time threshold from the physical examination items to be recommended; Based on the grouping bucket and the average order value stratification label, the target user group is determined. In the additional item distribution table of the target user group, multiple physical examination items are selected as recommended physical examination items according to the historical order volume.

[0078] In the apparatus of this invention, the recommended card module 503 is used for: According to multiple preset slots, structured card data is assembled for the physical examination items to generate recommendation cards; wherein, the multiple slots are used to fill the following items: basic information of the items, risk association information, examination process information, item advantage tags, and credibility information; The basic information of the project, the examination process information, and the project advantage tags are extracted from the information pre-configured for the physical examination project. The risk association information is generated based on the health risk sites and risk causes corresponding to the physical examination items in the health risk list output by the health agent; when the risk cause does not exist, it is generated based on the health risk sites and preset fallback text. The credibility information is determined based on the doctor qualification information pre-configured for the physical examination items and / or feedback information from other users regarding the physical examination items.

[0079] The apparatus for implementing this invention further includes a document module, used for: Based on the health risk sites and the information associated with the health risk sites in the portrait snapshot, the copywriting AI is invoked to generate recommendation text for each of the physical examination items.

[0080] Furthermore, the specific implementation details of the device described in the embodiments of the present invention have been described in detail in the above-described method, so the details will not be repeated here.

[0081] Figure 6 An exemplary system architecture 600 to which embodiments of the present invention can be applied is shown, including terminal devices 601, 602, 603, network 604, and server 605 (only an example).

[0082] Terminal devices 601, 602, and 603 can be various electronic devices with displays and support for web browsing, and have various communication client applications installed. Users can use terminal devices 601, 602, and 603 to interact with server 605 through network 604 to receive or send messages, etc.

[0083] Network 604 is a medium used to provide a communication link between terminal devices 601, 602, 603 and server 605. Network 604 may include various connection types, such as wired or wireless communication links or fiber optic cables, etc.

[0084] Server 605 can be a server providing various services, such as a backend management server supporting shopping websites browsed by users using terminal devices 601, 602, and 603 (this is just an example). The backend management server can analyze and process received data such as product information query requests, and feed back the processing results (e.g., target push information, product information—this is just an example) to the terminal devices. It should be noted that the method provided in this embodiment of the invention is generally executed by server 605, and correspondingly, the apparatus is generally set in server 605.

[0085] It should be understood that Figure 6 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0086] The following is for reference. Figure 7 It shows a schematic diagram of the structure of a computer system 700 suitable for implementing a terminal device of the present invention. Figure 7 The terminal device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the present invention.

[0087] like Figure 7As shown, the computer system 700 includes a central processing unit (CPU) 701, which can perform various appropriate actions and processes based on programs stored in read-only memory (ROM) 702 or programs loaded from storage section 708 into random access memory (RAM) 703. The RAM 703 also stores various programs and data required for the operation of the system 700. The CPU 701, ROM 702, and RAM 703 are interconnected via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0088] The following components are connected to the I / O interface 705: an input section 706 including a keyboard, mouse, etc.; an output section 707 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 708 including a hard disk, etc.; and a communication section 709 including a network interface card such as a LAN card, modem, etc. The communication section 709 performs communication processing via a network such as the Internet. A drive 710 is also connected to the I / O interface 705 as needed. A removable medium 711, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 710 as needed so that computer programs read from it can be installed into the storage section 708 as needed.

[0089] In particular, according to the embodiments disclosed in this invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 709, and / or installed from removable medium 711. When the computer program is executed by central processing unit (CPU) 701, it performs the functions defined above in the system of this invention.

[0090] It should be noted that the computer-readable medium shown in this invention can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this invention, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0091] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0092] The modules described in the embodiments of the present invention can be implemented in software or hardware. The described modules can also be housed in a processor; for example, a processor may be described as including a risk assessment module, an add-on recommendation module, a recommendation card module, and an add-on transaction module. The names of these modules do not necessarily limit the module itself; for example, the risk assessment module may also be described as a "profile snapshot module."

[0093] In another aspect, the present invention also provides a computer-readable medium, which may be included in the device described in the above embodiments; or it may exist independently and not assembled into the device. The computer-readable medium carries one or more programs that, when executed by the device, cause the device to perform any of the above-described methods for recommending additional health check items.

[0094] The computer program product of the present invention includes a computer program that, when executed by a processor, implements the physical examination additional item recommendation method in the embodiments of the present invention.

[0095] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can occur depending on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for recommending additional items in a physical examination, characterized in that, include: In response to the recommendation request transmitted by the client, multi-source information of the user is collected to construct a profile snapshot, and the profile snapshot is input into the health intelligent agent to determine the health risk sites; In the preset add-on item library, multiple physical examination items that correspond to the health risk sites and meet the preset recommendation conditions are retrieved, and the multiple physical examination items are sorted based on the customer price stratification tags in the profile snapshot; Generate recommendation cards corresponding to each of the aforementioned physical examination items, and send the health risk sites and the associated recommendation cards for each of the aforementioned physical examination items to the client for sequential display; In response to a transaction instruction transmitted by the client for a target health checkup item in one or more recommended cards, the transaction process for the target health checkup item is executed.

2. The method according to claim 1, characterized in that, The health risk site is determined based on the health risk list output by the health agent, and the method further includes: In response to the health intelligence agent outputting a health risk list and an additional item recommendation list, the physical examination items corresponding to the health risk sites are extracted from the additional item recommendation list, and deduplication is performed with the retrieved multiple physical examination items to obtain the final recommended physical examination items; In response to the health intelligence agent only outputting a list of health risks, if the user's current physical examination order does not include physical examination items corresponding to the health risk sites, the retrieved multiple physical examination items will be determined as the final recommended physical examination items.

3. The method according to claim 1, characterized in that, The method further includes: In response to a timeout or output error in the health agent call, a retry operation is performed and / or a backup agent of the health agent is called. If the execution result is still a timeout or output error, the item recommendation process is terminated.

4. The method according to any one of claims 1-3, characterized in that, The multi-source information includes the following: questionnaire answers, historical medical examination reports, demographic attributes, and historical additional order information; the profile snapshot is constructed based on the following: The questionnaire answers were normalized into standard features according to data type. Extract positive items and their corresponding historical examination dates from the historical medical examination reports; Determine the grouping bucket to which the demographic attribute belongs; The average order value tier labels are determined based on the transaction prices of the historical add-on orders.

5. The method according to claim 4, characterized in that, The recommendation request is generated by the client in response to the user triggering the physical examination node; The process of obtaining the questionnaire answers includes: determining the health questionnaire corresponding to the physical examination node and sending it to the client for display, so as to receive the questionnaire answers input by the user transmitted by the client; The process of constructing the health questionnaire includes: The questionnaire segmentation data collector splits the pre-set question bank into different segments according to different pre-configured dimensions. Determine the priority of each shard in advance, and determine the shard corresponding to each shard based on the pre-configured mapping relationship between the shard and the priority. A predetermined number of questions are extracted from the segments corresponding to each of the physical examination nodes to construct a health questionnaire for each of the physical examination nodes.

6. The method according to claim 4, characterized in that, The method further includes at least one of the following: Based on the historical physical examination reports, determine the latest examination date of the health risk area, calculate the time difference between the latest examination date and the current date, and filter out physical examination items with the same name whose time difference is less than a preset time threshold from the physical examination items to be recommended; Based on the grouping bucket and the average order value stratification label, the target user group is determined. In the additional item distribution table of the target user group, multiple physical examination items are selected as recommended physical examination items according to the historical order volume.

7. The method according to claim 1 or 6, characterized in that, The generation of recommendation cards corresponding to each of the physical examination items includes: According to multiple preset slots, structured card data is assembled for the physical examination items to generate recommendation cards; wherein, the multiple slots are used to fill the following items: basic information of the items, risk association information, examination process information, item advantage tags, and credibility information; The basic information of the project, the examination process information, and the project advantage tags are extracted from the information pre-configured for the physical examination project. The risk association information is generated based on the health risk sites and risk causes corresponding to the physical examination items in the health risk list output by the health agent; when the risk cause does not exist, it is generated based on the health risk sites and preset fallback text. The credibility information is determined based on the doctor qualification information pre-configured for the physical examination items and / or feedback information from other users regarding the physical examination items.

8. The method according to claim 1, characterized in that, The method further includes: Based on the health risk sites and the information associated with the health risk sites in the portrait snapshot, the copywriting AI is invoked to generate recommendation text for each of the physical examination items.

9. A device for recommending additional items in a physical examination, characterized in that, include: The risk assessment module is used to respond to recommendation requests transmitted by the client, collect multi-source information of users to construct a profile snapshot, and input the profile snapshot into the health intelligent agent to determine the health risk sites; The add-on recommendation module is used to retrieve multiple physical examination items that correspond to the health risk sites and meet the preset recommendation conditions from the preset add-on item library, and to sort the multiple physical examination items based on the customer price stratification tags in the profile snapshot; The recommendation card module is used to generate recommendation cards corresponding to each of the physical examination items, and send the health risk sites and the associated recommendation cards of each of the physical examination items to the client for sequential display. The add-item transaction module is used to execute the transaction process for the target physical examination item in response to the transaction instruction transmitted by the client for one or more recommended cards.

10. An electronic device, characterized in that, include: One or more processors; Storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-8.

11. A computer-readable medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-8.

12. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-8.

Images