Health management task generation method and device, computer device, and storage medium

Abstract

The application relates to the technical field of data analysis, and specifically discloses a health management task generation method and device, computer equipment and a storage medium. The application can further mine the implicit needs of a first object by fusing historical health information and health needs in a health management request to perform knowledge graph reasoning, can realize comprehensive analysis of needs, can improve the comprehensiveness of health management needs, can disassemble health management needs according to an executable task threshold of a second object, can realize alignment of needs and the threshold, can ensure that health management tasks are within the executable range of the second object, can avoid resource mismatching problems, and can improve the executability of health management tasks. The application can be applied to medical and health management business fields such as doctor-to-enterprise, home visits, home-based care for the elderly, and the like, can realize accurate matching of needs and medical services, can generate health management tasks with the ability level of accurately matched medical service personnel, and can improve the executability of health management tasks and user experience.

Description

Technical Field

[0001] This application relates to the field of data analysis technology, and in particular to a method, apparatus, computer equipment, and storage medium for generating health management tasks. It can be applied to healthcare management services such as expert-led outreach to enterprises, in-home consultations, and home-based elderly care. Background Technology

[0002] As companies increasingly prioritize employee health, the enterprise-level health management service market is rapidly developing. Currently, companies typically purchase benefit packages from third-party health service platforms to access services such as lectures by renowned doctors, offline consultations, and health screenings. However, existing service platforms often only perform simple keyword matching or manual interpretation of received health management requests. Furthermore, when generating and assigning tasks based on requests, they usually translate the company's vague needs directly into simple task orders and a simplistic order-assignment model. This frequently leads to resource mismatch issues such as successful assignments but doctor rejections, and mismatches between doctor capabilities and tasks, resulting in low feasibility of health management tasks. Therefore, how to conduct comprehensive needs analysis and accurately match service resources to improve the feasibility of health management tasks has become an urgent problem to be solved. Summary of the Invention

[0003] This application provides a method, apparatus, computer equipment, and storage medium for generating health management tasks, so as to achieve comprehensive demand analysis and precise matching of service resources, thereby improving the feasibility of health management tasks.

[0004] Firstly, this application provides a method for generating health management tasks, the method comprising: Upon receiving a health management request from the first object, the system obtains the first object's historical health information, and parses and performs graph reasoning on the historical health information and the health management request to obtain the first object's health management needs. Based on the health management needs, a second object corresponding to the health management request is determined from a preset second object database; Obtain the executable task threshold of the second object, and decompose the health management requirement based on the executable task threshold to generate the health management task of the health management request.

[0005] Secondly, this application also provides a health management task generation device, the device comprising: The health management needs acquisition module is used to obtain the historical health information of the first object when a health management request is received from the first object, and to parse the historical health information and the health management request and perform graph reasoning to obtain the health management needs of the first object. The second object determination module is used to determine the second object corresponding to the health management request from a preset second object database based on the health management needs. The health management task generation module is used to obtain the executable task threshold of the second object, and decompose the health management requirement based on the executable task threshold to generate the health management task of the health management request.

[0006] Thirdly, this application also provides a computer device, the computer device including a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program and, when executing the computer program, implement the health management task generation method as described above.

[0007] Fourthly, this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to implement the health management task generation method described above.

[0008] This application discloses a method, apparatus, computer device, and storage medium for generating health management tasks. Upon receiving a health management request from a first object, the method acquires the historical health information of the first object, and parses and performs knowledge graph reasoning on the historical health information and the health management request to obtain the health management needs of the first object. Based on the health management needs, a second object corresponding to the health management request is determined from a preset second object database. An executable task threshold for the second object is obtained, and the health management needs are decomposed based on the executable task threshold to generate the health management task for the health management request. This application, by integrating historical health information and health needs from the health management request for knowledge graph reasoning, can further uncover the implicit needs of the first object, achieving comprehensive needs analysis and improving the comprehensiveness of health management needs. Decomposing the health management needs based on the executable task threshold of the second object aligns the needs with the threshold, ensuring that the health management task is within the executable range of the second object, avoiding resource mismatch problems, and improving the executability of the health management task. Attached Figure Description

[0009] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1This is a schematic diagram of the application environment for a health management task generation method provided in an embodiment of this application; Figure 2 This is a schematic flowchart of a health management task generation method provided in the first embodiment of this application; Figure 3 This is a schematic flowchart of a health management task generation method provided in the second embodiment of this application; Figure 4 This is a schematic flowchart of a health management task generation method provided in the third embodiment of this application; Figure 5 A schematic block diagram of a health management task generation device provided for embodiments of this application; Figure 6 A schematic block diagram of the structure of a computer device provided for an embodiment of this application. Detailed Implementation

[0011] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0012] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content and operations / steps, nor does it necessarily have to be performed in the order described. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.

[0013] It should be understood that the terminology used in this application specification is for the purpose of describing particular embodiments only and is not intended to limit the application. As used in this application specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0014] It should also be further understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0015] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0016] The health management task generation method provided in this embodiment of the invention can be applied to, for example... Figure 1In this application environment, the client communicates with the server via a network. The server can receive health management requests from the client, and upon receiving a health management request from a first object, obtain the historical health information of the first object, and parse and perform knowledge graph reasoning on the historical health information and the health management request to obtain the health management needs of the first object. Based on the health management needs, the server determines the second object corresponding to the health management request from a preset second object database; obtains the executable task threshold of the second object, and decomposes the health management needs based on the executable task threshold to generate the health management task of the health management request. In this invention, by fusing historical health information and health needs in health management requests for knowledge graph reasoning, the implicit needs of the first object can be further mined, achieving comprehensive needs analysis and improving the comprehensiveness of health management needs. Decomposing health management needs based on the executable task threshold of the second object aligns the needs with the threshold, ensuring that the health management task is within the executable range of the second object, avoiding resource mismatch problems, and improving the executability of the health management task. The client can be, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. The server can be implemented using a standalone server or a server cluster consisting of multiple servers. The invention will be described in detail below through specific embodiments.

[0017] like Figure 2 As shown, the health management task generation method specifically includes steps S101 to S103.

[0018] S101. Upon receiving a health management request from the first object, obtain the historical health information of the first object, and parse the historical health information and the health management request and perform graph reasoning to obtain the health management needs of the first object. In one embodiment, the first object refers to the enterprise entity that initiates a health management service request through an online platform, and is the purchaser and demand initiator of the health management service.

[0019] A health management request is a health service request submitted by the first party through an online platform, expressed in natural language or unstructured form (such as voice, video, etc.), and may include elements such as basic company information, expected service type, time, and budget.

[0020] Historical health information refers to the past health service data of the primary individual, including historical health service records, abnormal employee physical examination data, corporate health risk profiles, service satisfaction evaluations, employee health records, etc.

[0021] In one embodiment, historical health information and health management requests are parsed and preprocessed. Preprocessing may include removing noise such as special characters and extra spaces, performing word segmentation on the text and labeling the part of speech of each word, and using named entity recognition technology to extract key entity tags from the text. Entity tags include, but are not limited to, industry attributes (such as internet, manufacturing), population characteristics (such as programmers, factory workers), disease types or body parts (such as cervical spine, lumbar spine), service types (such as lectures, face-to-face consultations, and consultations), budget range (such as around 50,000), time windows (such as next month), and information related to company size.

[0022] Based on the obtained entity tags, path reasoning is performed in the preset health knowledge graph to obtain at least one implicit association tag. Based on the entity tags and implicit association tags, demand analysis is performed to obtain health management needs.

[0023] like Figure 3 As shown, step S101 specifically includes steps S1011 to S1015.

[0024] S1011. The health management request is parsed, the original request text is extracted, and entity recognition is performed on the original request text to obtain at least one entity label; In one embodiment, the health management request is parsed to extract the request statement content input by the first object. The request statement content can be in natural language, or in the form of voice, images, video, etc. Understandably, if the request statement content is in the form of voice, images, video, etc., then it is subjected to text recognition or conversion to obtain the original request text.

[0025] A pre-defined named entity recognition model can be used to perform named entity recognition on the obtained original request text to obtain at least one entity label. This can include multiple dimensions such as industry attributes, demographic characteristics, disease type, service type, budget range, and time frame. For example, for the request text "Our internet company has many programmers, and we'd like to invite an expert to give a lecture on cervical and lumbar spine protection, preferably with on-site consultations of several serious cases. Our budget is around 50,000 yuan, and we hope to arrange this next month," named entity recognition will extract "internet company" as the industry attribute, "programmers" as the demographic characteristic, "cervical spine" and "lumbar spine" as disease types, "lecture on protection" as the health lecture service type, "on-site consultations of cases" as the on-site medical service type, "around 50,000 yuan" as the budget range, and "next month" as the time window.

[0026] S1012. Based on the preset health knowledge graph and the entity tags, perform graph reasoning to obtain implicit association tags corresponding to the entity tags; In one embodiment, a health knowledge graph is a domain knowledge base centered on the medical and health field. It is a graph constructed with health-related entities as nodes and semantic relationships between entities as edges, where each edge corresponds to a confidence score. Health-related entities include diseases, symptoms, departments, treatment methods, population characteristics, occupational factors, industry risks, etc.

[0027] Implicitly associated tags refer to extended tags that are inferred from health knowledge graphs, are not mentioned in the original request text, but are related to entity tags.

[0028] In one embodiment, entity tags are matched against a health knowledge graph to obtain target graph nodes corresponding to the entity tags. Starting from the target graph node, a depth-first or breadth-first search is performed along the relation edges in the graph nodes to obtain at least one reasoning path. Valid paths are then selected, and entities within these valid paths are identified as implicitly associated tags.

[0029] Further, step S1012 includes: matching the entity tag with graph nodes in the health knowledge graph to determine a target graph node that matches the entity tag; traversing graph nodes along the same path as the target graph node, starting from the target graph node, to obtain at least one inference path; filtering paths according to the type of the endpoint node of each inference path to obtain at least one valid path, and determining at least one implicit association tag based on each valid path.

[0030] In one embodiment, the obtained entity labels are mapped to corresponding nodes in the health knowledge graph. Specifically, the entity labels are matched with entity nodes in the health knowledge graph using semantic similarity matching, synonym matching, and exact matching. When the matching degree is greater than or equal to a preset threshold, the two are determined to be a match. At this point, the entity label is mapped to the corresponding entity node in the health knowledge graph. For example, "cervical spondylosis" is matched to the disease node "cervical spondylosis"; "programmer" is matched to the population characteristic node "software and information technology service personnel"; and "sedentary" is matched to the occupational factor node "long-term seated work".

[0031] It is understandable that if there is an entity label that fails to be mapped (i.e., there is no matching entity node in the health knowledge graph), then that entity label will not participate in graph reasoning.

[0032] In one embodiment, for each successfully mapped entity label, starting from the target graph node corresponding to that entity label, all nodes along the same path are traversed to obtain at least one inference path corresponding to that entity label. A graph traversal algorithm is used, starting from the starting node and performing a depth-first or breadth-first search along the relation edges, recording the path length and path confidence. The path confidence is the product of the confidence of each relation edge along the path; paths below a threshold are pruned. The endpoint node of the path is compared with the health service requirement. If the entity type of the endpoint node matches the health service requirement, the path is considered valid, and all nodes on that path are implicitly associated labels.

[0033] In another embodiment, multiple entity labels can be combined, and a graph traversal algorithm can be used for path reasoning to obtain deep relationships between entities. Specifically, the combination rules can be combinations of entities of the same type (such as multiple disease labels being associated through a co-disease relationship) or combinations of entities of the same type (such as population labels and industry labels being associated through an "industry-population" relationship).

[0034] For example, taking the entity tag combination "programmer, cervical spine, lumbar spine" in the aforementioned example as an example, the reasoning path is: programmer - sedentary - cervical spondylosis - spinal surgery - health lecture. The health lecture is the same as the lecture service type in the health management request. This path is a valid path, so sedentary and spinal surgery are implicitly associated tags.

[0035] In the above embodiments, the knowledge points corresponding to entities are accurately located through graph node matching, potential relationships are mined by traversing the associated paths, and then effective inference links are selected according to node type to efficiently extract implicit association tags. This overcomes the one-sidedness and limitations of relying solely on the original request entity tags to analyze needs, enriches the tag system for the health dimension, and provides comprehensive, accurate, and reliable data support for subsequent accurate analysis of users' current health needs and deduction of the evolution trend of needs, effectively improving the completeness and accuracy of health management needs identification.

[0036] S1013. Perform a requirement analysis on each of the entity tags to obtain the current requirements of the first object; In one embodiment, all entity tags are transmitted to a preset demand analysis model, and the demand analysis model is used to analyze the demand of all entity tags to obtain the current demand.

[0037] S1014. Perform demand prediction on the historical health information, the entity tags, and each of the implicit association tags to obtain the predicted demand; Further, step S1014 includes: analyzing the historical health information in chronological order to construct a health service timeline; mapping and aligning each entity label and the implicit association label with each record node in the health service timeline to obtain an alignment result; generating a multimodal temporal feature vector based on the alignment result, and processing the multimodal temporal feature vector based on a preset demand evolution model to obtain the predicted demand.

[0038] In one embodiment, the service event records in the historical health information are sorted chronologically to construct a health service timeline. Each record node in the timeline may include time, service event type, event content, effect evaluation, etc.

[0039] The entity labels and implicitly associated labels are mapped and aligned with corresponding records in the historical timeline to obtain the alignment results. Specifically, a strategy of prioritizing exact matching and supplementing with fuzzy matching can be adopted for mapping and alignment. All record nodes in the timeline are traversed to find records that perfectly match the labels in the corresponding dimensions. A perfect match means identical encoding or identical text. Labels that successfully achieve an exact match are marked as precisely aligned. For labels that fail to achieve an exact match, fuzzy matching is performed using semantic similarity calculation. Semantic similarity calculation uses a pre-trained language model to encode the label text and record text, and calculates the cosine similarity. Similarity scores exceeding a threshold are marked as fuzzy aligned.

[0040] In another embodiment, for tags for which neither exact matching nor fuzzy matching is successful, association inference can be performed through knowledge graph reasoning. For example, the current tag "dry eye syndrome" has no direct correspondence in the history, but a knowledge graph query reveals that "blurred vision" and "dry eyes" are typical symptoms of dry eye syndrome, leading to inference that it is an indirect alignment.

[0041] In one embodiment, feature encoding is performed based on the alignment results to generate a multimodal temporal feature vector. Specifically, each alignment point is traversed, and the entity label of each alignment point is encoded and combined to obtain a sub-feature vector. The sub-feature vectors of different modalities under each node are concatenated sequentially according to the chronological order, preserving the temporal logical relationship. The sub-feature vectors corresponding to all temporal nodes are integrated to obtain the multimodal temporal feature vector.

[0042] In one embodiment, the demand evolution model can employ a temporal fusion deep learning model, trained using labeled historical health time-series samples. Each sample is bound to a temporal feature vector and a manually labeled, accurate predicted demand label.

[0043] Multimodal time-series feature vectors are transmitted to the demand evolution model to predict demand trends and obtain predicted demand.

[0044] In the above embodiments, a health service timeline is constructed by performing time-series analysis on historical health information, and explicit entity labels and implicit association labels are mapped and aligned to each record node. This solves the problem of relatively isolated entity labels and implicit association labels, clearly presenting the state change process. Multimodal time-series feature vectors are obtained through encoding, achieving deep fusion and unified representation of multi-source heterogeneous health information. Finally, demand evolution trend prediction is performed based on these multimodal time-series feature vectors, enabling accurate prediction of demand and effectively improving the comprehensiveness of health management needs.

[0045] S1015. Based on the predicted demand and the current demand, generate the health management demand.

[0046] In one embodiment, predicted and current demands are deduplicated and merged to generate health management demands.

[0047] In the above embodiments, by parsing the request and identifying entity tags, the current needs are obtained. By combining the reasoning of the health knowledge graph to mine implicit associations, potential needs are obtained. The current needs and potential needs are integrated to form complete health management needs, realizing a comprehensive analysis of needs and providing a reliable basis for the generation of subsequent health management tasks.

[0048] S102. Based on the health management needs, determine the second object corresponding to the health management request from the preset second object database; In one embodiment, the second object refers to the service personnel registered in the platform database (second object database), such as doctors, nurses, medical teams, etc., who are the implementers of health management services.

[0049] The second object database stores the qualification information, future schedule occupancy status, and historical task execution records for each second object.

[0050] The health management needs are matched against the second object database to identify the second object.

[0051] Further, step S102 includes: extracting and encoding features from the health management needs to obtain a needs feature vector; traversing the second object database to extract relevant information for each of the second objects, and extracting and encoding features from the relevant information to obtain a capability feature vector corresponding to the second object; performing similarity matching between the needs feature vector and the capability feature vector to obtain a similarity score; and comparing the similarity score with a preset score threshold to determine the second object.

[0052] In one embodiment, health management needs are feature-extracted, including key matching dimensions such as required specialty, disease intervention type, expert qualification level requirements, service type requests, service geographical scope, expected service time period, service population characteristics, health risk level, industry occupational disease suitability tags, and task urgency and priority. The extracted features are then encoded to obtain a demand feature vector.

[0053] In one embodiment, each second object in the database is traversed, and its qualification information, future availability, historical task execution records, and other feature information are extracted, encoded, and used to generate a capability feature vector.

[0054] The similarity between the demand feature vector and the capability feature vector is calculated to obtain a similarity score. The second object with a similarity score higher than the preset similarity threshold is selected as the candidate second object and sorted in descending order of similarity score.

[0055] In one embodiment, the obtained candidate second object can be displayed to the first object, and the target second object determined by the first object can be received as the second object corresponding to the current health management request.

[0056] S103. Obtain the executable task threshold of the second object, and decompose the health management requirement based on the executable task threshold to generate the health management task of the health management request.

[0057] In one embodiment, the executable task threshold refers to a set of dynamic capability boundary parameters characterizing the serviceability of the second object, including quantitative constraints in dimensions such as time, space, expertise, and physical strength. This threshold can be continuously calibrated and updated based on the second object's historical service data, pending tasks, self-reported information, and actual execution feedback. It can be stored in the second object's database.

[0058] In one embodiment, the executable task threshold is parsed and converted into constraints in the task generation process. Based on preset conditions, the health management needs are broken down into several sub-tasks. The task volume of each sub-task does not exceed the corresponding executable task threshold. The obtained sub-tasks are sorted and packaged according to preset rules to obtain the final health management task.

[0059] Furthermore, the step of decomposing the health management requirement based on the executable task threshold to generate the health management task for the health management request includes: parsing the executable task threshold to obtain task constraints; decomposing the health management requirement based on the task constraints to obtain at least one sub-task; and sorting and encapsulating each sub-task according to a preset rule to obtain the health management task corresponding to the health management request.

[0060] In one embodiment, the executable task threshold may include time-dimensional thresholds (such as the maximum continuous duration of a single service), spatial-dimensional thresholds (such as the maximum distance of a single trip, the maximum number of cross-city services per day), and physical-dimensional thresholds (such as the unit time consumption coefficient for different service types, the maximum cumulative consumption per day), etc.

[0061] The executable task thresholds are structured and parsed, and constraint parameters are extracted by dimension. Using these constraints as task boundaries, health management needs are decomposed. Specifically, health management needs are broken down into multiple independent executable sub-tasks based on constraints such as service type, service population size, professional subject, service duration, and time. Attribute information for each sub-task is generated based on information from each dimension, including service type, service population, service duration, professional subject, expected time period, and execution location.

[0062] For example, if a health management requirement necessitates in-person consultations for 30 cases, while the threshold for a single consultation for the second individual is 15 cases, the system will break down the requirement into two in-person consultation sub-tasks, which will be scheduled to be executed on different dates or at different times on the same date.

[0063] In one embodiment, the preset rules may include logical sequence rules, risk reduction rules, etc., and each rule has a corresponding priority, such as logical sequence rules as the first priority and risk reduction rules as the second priority. First, the basic sequence is determined according to the logical sequence rules, and then the order of uncertain sub-tasks is fine-tuned according to the risk reduction rules.

[0064] Among these, the logical sequence rule is used to ensure the inherent logical order of task execution. For example, information gathering precedes analysis and judgment. The risk reduction rule is used to arrange sub-tasks with high uncertainty earlier and sub-tasks with high certainty later, facilitating adjustments later. For example, the preliminary review of medical records carries the risk of incomplete data; arranging it earlier allows for the early identification and supplementation of problems.

[0065] In one embodiment, after sorting, the subtask sequence and the attribute information of each subtask are encapsulated into a final health management task, and a unique identifier is assigned to the health management task.

[0066] In the above embodiments, by parsing the executable task threshold of the second object to extract constraints, and decomposing health management needs according to the constraints, the task division conforms to the actual execution situation, making the generated health management tasks compliant, adaptable, and logically clear, avoiding the situation where the task exceeds the capability threshold of the second object, and improving the executability of the task.

[0067] like Figure 4 As shown, after step S103, steps S201 to S203 are also included.

[0068] S201. Add the health management task to the task list to be executed of the second object; S202. When the task generation stop condition corresponding to the second object is met, obtain the task execution condition corresponding to each health management task in the list of tasks to be executed; S203. Based on the execution conditions of each task, construct multi-dimensional constraints, and perform multi-task planning for each health management task according to the preset multi-objective optimization scheduling algorithm and the multi-dimensional constraints to obtain the task execution strategy, so as to guide the second object to execute each health management task.

[0069] In one embodiment, the list of tasks to be executed is the set of all health management tasks assigned to the second object but not yet executed. Executed tasks are removed from the list, and newly generated health management tasks are added to the list.

[0070] In one embodiment, the task stop generation condition refers to the condition that stops matching requests and assigning tasks to the second object. This could be that the number of tasks in the pending task list reaches a certain threshold, the second object is in a paused task assignment state, or other pre-configured conditions that prevent the second object from receiving health management tasks.

[0071] Once the task stop condition is triggered, all tasks in the pending task list are traversed, and their execution conditions are parsed and extracted one by one. Specifically, the subtask attribute information of each pending task is extracted, and the execution conditions of the pending task are determined by combining the attribute information of each subtask. These conditions may include the expected execution time period, service location geolocation, service duration, service user scale, service requirements, etc.

[0072] The extracted task execution conditions are encoded into structured constraint parameters, and a multi-dimensional constraint space is constructed based on these parameters. This space includes time constraints, spatial constraints, and priority constraints.

[0073] A pre-defined multi-objective optimization scheduling algorithm is adopted, with the optimization objectives of minimizing commuting time, maximizing task completion time, balancing resource load, prioritizing the fulfillment of high-priority tasks, and aggregating tasks within the same region. Under multi-dimensional constraints, the algorithm performs global planning, time-series scheduling, and path planning for all health management tasks in the list.

[0074] After the multi-objective optimization scheduling algorithm converges, it outputs a complete task execution strategy, including: a task execution time sequence plan, the order of enterprise visits, the optimal commuting route, the standard start and end times for each task, task allocation configuration, contingency plans for abnormal tasks, and on-site service execution specifications. The task execution strategy is then pushed to the second-party terminal and the backend management terminal, guiding the second-party to execute each health management task sequentially according to the planned time, route, and sequence.

[0075] In the above embodiments, health management tasks are aggregated into a unified list of tasks to be executed by a second object for centralized management. A multi-dimensional constraint system is built based on the execution conditions of each task, and a multi-objective optimization scheduling algorithm is used to complete the overall planning of multiple tasks. This approach not only systematically summarizes various health management tasks, avoiding disorganized tasks, but also balances the execution demands of multiple tasks based on multiple constraints, outputting reasonable and feasible task execution strategies. This effectively improves the rationality and efficiency of health management task arrangement, ensuring the standardized and efficient execution of health management work.

[0076] Please see Figure 5 , Figure 5 This is a schematic block diagram of a health management task generation device provided in an embodiment of this application. This device is used to execute the aforementioned health management task generation method. The health management task generation device can be configured on a server.

[0077] like Figure 5 As shown, the health management task generation device 300 includes: The health management needs acquisition module 301 is used to obtain the historical health information of the first object when a health management request of the first object is received, and to parse the historical health information and the health management request and perform graph reasoning to obtain the health management needs of the first object. The second object determination module 302 is used to determine the second object corresponding to the health management request from a preset second object database based on the health management needs. The health management task generation module 303 is used to obtain the executable task threshold of the second object, and to decompose the health management requirement based on the executable task threshold to generate the health management task of the health management request.

[0078] In one embodiment, the health management needs acquisition module 301 includes: An entity recognition unit is used to parse the health management request, extract the original request text, and perform entity recognition on the original request text to obtain at least one entity label. The graph reasoning unit is used to perform graph reasoning based on a preset health knowledge graph and the entity tags to obtain implicit association tags corresponding to the entity tags. The current requirement acquisition unit is used to perform requirement analysis on each of the entity tags to obtain the current requirement of the first object; The demand prediction unit is used to predict the demand based on the historical health information, the entity tags, and each of the implicit association tags. A health management needs generation unit is used to generate the health management needs based on the predicted needs and the current needs.

[0079] In one embodiment, the demand forecasting unit includes: The timeline construction subunit is used to analyze the historical health information in chronological order and construct a health service timeline. The alignment result acquisition sub-unit is used to map and align each entity label and the implicit association label with each record node in the health service timeline to obtain the alignment result; The predicted demand acquisition subunit is used to generate a multimodal temporal feature vector based on the alignment result, and to process the multimodal temporal feature vector based on a preset demand evolution model to obtain the predicted demand.

[0080] In one embodiment, the graph inference unit includes: The target graph node determination subunit is used to match the entity label with the graph nodes in the health knowledge graph to determine the target graph node that matches the entity label. The reasoning path acquisition subunit is used to traverse the graph nodes with the same path as the target graph node, starting from the target graph node, to obtain at least one reasoning path. The implicit association label obtaining subunit is used to perform path filtering based on the type of the endpoint node of each inference path, obtain at least one valid path, and determine at least one implicit association label based on each valid path.

[0081] In one embodiment, the health management task generation module 303 includes: The constraint condition acquisition unit is used to parse the executable task threshold and obtain the task constraint conditions; The subtask acquisition unit is used to decompose the health management requirements based on the task constraints to obtain at least one subtask. The health management task acquisition unit is used to sort and encapsulate each of the sub-tasks according to preset rules to obtain the health management task corresponding to the health management request.

[0082] In one embodiment, the health management task generation device 300 further includes a task planning module, the task planning module comprising: The task transmission unit is used to add the health management task to the task list to be executed of the second object; The task execution condition acquisition unit is used to acquire the task execution conditions corresponding to each health management task in the list of tasks to be executed when the task generation stop condition corresponding to the second object is met. The task planning unit is used to construct multi-dimensional constraints based on the execution conditions of each task, and to perform multi-task planning for each health management task according to a preset multi-objective optimization scheduling algorithm and the multi-dimensional constraints, so as to obtain the task execution strategy to guide the second object to execute each health management task.

[0083] In one embodiment, the second object determination module 302 includes: The demand feature vector acquisition unit is used to extract and encode the health management demand to obtain the demand feature vector; The capability feature vector acquisition unit is used to traverse the second object database, extract relevant information of each second object, and perform feature extraction and encoding on the relevant information to obtain the capability feature vector corresponding to the second object. The similarity score acquisition unit is used to perform similarity matching between the demand feature vector and the capability feature vector to obtain a similarity score; The second object determination unit is used to compare the similarity score with a preset scoring threshold to determine the second object.

[0084] It should be noted that those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the above-described apparatus and modules can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0085] The aforementioned device can be implemented as a computer program, which can be used in, for example... Figure 6 It runs on the computer device shown.

[0086] Please see Figure 6 , Figure 6 This is a schematic block diagram illustrating the structure of a computer device according to an embodiment of this application. The computer device may be a server.

[0087] See Figure 6 The computer device includes a processor, memory, and network interface connected via a system bus, wherein the memory may include non-volatile storage media and internal memory.

[0088] Non-volatile storage media can store operating systems and computer programs. These computer programs include program instructions that, when executed, cause the processor to perform any method for generating health management tasks.

[0089] The processor provides computing and control capabilities, supporting the operation of the entire computer device.

[0090] Internal memory provides an environment for the execution of computer programs in non-volatile storage media. When these computer programs are executed by the processor, the processor can perform any health management task generation method.

[0091] This network interface is used for network communication, such as sending assigned tasks. Those skilled in the art will understand that... Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0092] It should be understood that the processor can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among these, a general-purpose processor can be a microprocessor or any conventional processor.

[0093] In one embodiment, the processor is configured to run a computer program stored in memory to perform the following steps: Upon receiving a health management request from the first object, the system obtains the first object's historical health information, and parses and performs graph reasoning on the historical health information and the health management request to obtain the first object's health management needs. Based on the health management needs, a second object corresponding to the health management request is determined from a preset second object database; Obtain the executable task threshold of the second object, and decompose the health management requirement based on the executable task threshold to generate the health management task of the health management request.

[0094] The embodiments of this application also provide a computer-readable storage medium storing a computer program, the computer program including program instructions, and the processor executing the program instructions to implement any of the health management task generation methods provided in the embodiments of this application.

[0095] The computer-readable storage medium may be an internal storage unit of the computer device described in the foregoing embodiments, such as the hard disk or memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, SmartMedia Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the computer device.

[0096] It should be noted that any AI models, software tools, or components not belonging to this company appearing in the embodiments of this application are merely illustrative examples and do not represent actual use. All user personal information involved in the embodiments of this application has been authorized (with the knowledge and consent) by the relevant parties or has been fully authorized by all parties, and the executing entity may obtain it through various legal and compliant means. The collection, storage, use, processing, transmission, provision, and disclosure of the information, data, and signals involved all comply with relevant laws and regulations and do not violate public order and good morals.

[0097] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for generating health management tasks, characterized in that, include: Upon receiving a health management request from the first object, the system obtains the first object's historical health information, and parses and performs graph reasoning on the historical health information and the health management request to obtain the first object's health management needs. Based on the health management needs, a second object corresponding to the health management request is determined from a preset second object database; Obtain the executable task threshold of the second object, and decompose the health management requirement based on the executable task threshold to generate the health management task of the health management request.

2. The health management task generation method according to claim 1, characterized in that, The step of parsing the historical health information and the health management request, and performing graph reasoning to obtain the health management needs of the first object, includes: The health management request is parsed to extract the original request text, and entity recognition is performed on the original request text to obtain at least one entity label; Based on the preset health knowledge graph and the entity tags, graph reasoning is performed to obtain implicit association tags corresponding to the entity tags; Perform a requirements analysis on each of the entity tags to obtain the current requirements of the first object; Demand prediction is performed on the historical health information, the entity tags, and each of the implicit association tags to obtain predicted demand; Based on the predicted demand and the current demand, the health management demand is generated.

3. The health management task generation method according to claim 2, characterized in that, The step of predicting demand based on the historical health information, the entity tags, and each of the implicit association tags to obtain predicted demand includes: The historical health information is analyzed in chronological order to construct a health service timeline; The entity tags and implicit association tags are mapped and aligned with each record node in the health service timeline to obtain the alignment result; A multimodal temporal feature vector is generated based on the alignment result, and the multimodal temporal feature vector is processed based on a preset demand evolution model to obtain the predicted demand.

4. The health management task generation method according to claim 2, characterized in that, The step of performing graph reasoning based on a preset health knowledge graph and the entity tags to obtain implicit association tags corresponding to the entity tags includes: The entity label is matched with the graph nodes in the health knowledge graph to determine the target graph node that matches the entity label; Starting from the target graph node, traverse the graph nodes that share the same path as the target graph node to obtain at least one reasoning path; Based on the type of the endpoint node of each inference path, path filtering is performed to obtain at least one valid path, and at least one implicit association label is determined based on each valid path.

5. The health management task generation method according to claim 1, characterized in that, The step of decomposing the health management requirement based on the executable task threshold to generate the health management task for the health management request includes: The executable task threshold is parsed to obtain the task constraints; Based on the aforementioned task constraints, the health management requirements are broken down into tasks to obtain at least one sub-task. The subtasks are sorted and encapsulated according to preset rules to obtain the health management task corresponding to the health management request.

6. The health management task generation method according to claim 1, characterized in that, After obtaining the executable task threshold of the second object, and decomposing the health management requirement based on the executable task threshold to generate the health management task of the health management request, the method further includes: Add the health management task to the task list to be executed for the second object; When the task generation stop condition corresponding to the second object is met, obtain the task execution condition corresponding to each health management task in the list of tasks to be executed; Based on the execution conditions of each task, multi-dimensional constraints are constructed, and multi-task planning is performed on each health management task according to the preset multi-objective optimization scheduling algorithm and the multi-dimensional constraints to obtain the task execution strategy, so as to guide the second object to execute each health management task.

7. The health management task generation method according to any one of claims 1 to 6, characterized in that, Based on the health management needs, determining the second object corresponding to the health management request from a preset second object database includes: The health management needs are feature extracted and encoded to obtain a demand feature vector; Traverse the second object database, extract relevant information for each second object, and perform feature extraction and encoding on the relevant information to obtain the capability feature vector corresponding to the second object; The demand feature vector and the capability feature vector are matched for similarity to obtain a similarity score; The similarity score is compared with a preset scoring threshold to determine the second object.

8. A health management task generation device, characterized in that, include: The health management needs acquisition module is used to obtain the historical health information of the first object when a health management request is received from the first object, and to parse the historical health information and the health management request and perform graph reasoning to obtain the health management needs of the first object. The second object determination module is used to determine the second object corresponding to the health management request from a preset second object database based on the health management needs. The health management task generation module is used to obtain the executable task threshold of the second object, and decompose the health management requirement based on the executable task threshold to generate the health management task of the health management request.

9. A computer device, characterized in that, The computer device includes a memory and a processor; The memory is used to store computer programs; The processor is configured to execute the computer program and, in executing the computer program, implement the health management task generation method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, causes the processor to implement the health management task generation method as described in any one of claims 1 to 7.

Images