Method and platform for integrating and sharing health data of old people

Abstract

The invention provides an integrated sharing method and platform for health data of old people, and relates to the technical field of data processing.The method comprises the steps that multi-source heterogeneous health data of the old people are accessed, a rule template in a historical grading preprocessing rule base is matched based on data source capacity and data features, and data cleaning and standardization are completed; further identifying a sensitive field based on a medical data standard and an industry privacy specification, generating a sensitive field vector, carrying out sensitive level scoring, and calling a data desensitization strategy library to execute desensitization operations such as de-identification and masking; an access control function is constructed based on user attributes, access context and field sensitivity levels, authorization levels are determined, differential encryption is performed, and authorization views and data tokens are generated for secure access. Through the method, the problems of heterogeneous sources, complex processing, insufficient privacy protection and lack of flexibility of a sharing mechanism of old-age health data in the prior art can be solved.

Description

Technical Field

[0001] This application relates to the technical field of health data processing, and particularly to a method and platform for integrating and sharing elderly health data. Background Art

[0002] With the intensification of the aging society, the demand for elderly health management is increasing day by day. At present, the sources of health data related to the elderly are extensive, including home monitoring devices, wearable devices, and elderly care service systems, forming a multi-source heterogeneous and structurally inconsistent health data system. In the prior art, there is a lack of an effective mechanism to uniformly access, standardize and semantically align health data from different devices and systems, resulting in difficult integration and utilization of data. At the same time, a large amount of sensitive information is contained in the elderly health data. The prior art has problems such as coarse granularity, weak control, and insufficient security in data desensitization, authorized access, and on-demand sharing, and it is difficult to meet the actual needs of precise protection and controllable sharing of sensitive health data in a multi-institution and multi-role environment. Summary of the Invention

[0003] The purpose of this application is to provide a method and platform for integrating and sharing elderly health data, which is used to solve the problems in the prior art that due to heterogeneous health data sources, inconsistent format structures, lack of effective preprocessing and semantic alignment mechanisms, and imperfect sensitive field identification, desensitization control, and access authorization policies, it is difficult to integrate data between different systems, the sharing efficiency between multiple institutions is low, and the security and compliance guarantee are insufficient. The present invention realizes the standardized integration of multi-source heterogeneous data by constructing a hierarchical preprocessing mechanism oriented to data source capabilities and data characteristics; designs a sensitive field identification and sensitive level scoring method based on medical data standards and industry privacy rules, and combines a data desensitization strategy library to achieve refined de-identification and sensitive information protection; at the same time, integrates visitor attributes, access context, and field sensitive levels to construct a differential access control and data authorization mechanism, so as to realize the safe, on-demand, and efficient sharing of elderly health data in a multi-institution and multi-role environment.

[0004] In view of the above problems, this application provides a method and platform for integrating and sharing elderly health data.

[0005] In the first aspect, this application provides a method for integrating and sharing elderly health data. The method for integrating and sharing elderly health data includes: Collect multi-dimensional health data of the elderly from multiple heterogeneous data sources. The heterogeneous data sources include home health monitoring devices, smart wearable devices, and chain elderly care service platforms. The multi-dimensional health data includes physiological monitoring data, clinical medical data, behavior and environment data, and psychological and social adaptability data of the elderly; Connect the multiple heterogeneous data sources to the information interaction center through a network protocol adaptation component, perform processing capacity evaluation and data feature extraction on the accessed data sources, generate corresponding data source capacity vectors and data feature vectors, and based on the vectors, match and dynamically call corresponding preprocessing rules from the historical hierarchical preprocessing rule library to perform data preprocessing operations; According to the unified data classification standard and semantic mapping rules, perform structure standardization, field alignment and semantic unification on the preprocessed data to generate an elderly health data set in a unified format; Based on medical data standards, industry privacy rules and a preset recognition model, perform sensitivity analysis on the fields in the elderly health data set and output sensitive field vector identifiers; Build a data desensitization policy library in combination with medical data standards and industry privacy rules at the initial stage of system deployment. According to the sensitive field vector identifier, field source and the context environment in the elderly health data set, perform sensitive level scoring on each sensitive field vector to generate a sensitive level vector, and according to the sensitive level vector, call the matching policy template in the data desensitization policy library to perform desensitization operations and output the desensitized elderly health data set; For the elderly health data set after the desensitization operation, combine the user attributes, access context and sensitive field level of the access requestor, input them into a pre-built access control function to obtain the access authorization level, and perform differential encryption processing on the fields allowed to be accessed by the authorization level; Screen the encrypted elderly health data set, extract the fields and their corresponding data contents that the current requestor has access rights to, and generate a data token with an access policy identifier, and transmit it to the access requestor through a communication channel. After receiving the data token, the requestor decrypts the fields and parses the contents according to the access policy identifier contained in the data token to present the corresponding authorized view data.

[0006] In a second aspect, the present application also provides an integrated sharing platform for elderly health data, which is used to execute an integrated sharing method for elderly health data as described in the first aspect. Among them, the integrated sharing platform for elderly health data includes: A data collection module, which is used to collect multi-dimensional health data of the elderly from multiple heterogeneous data sources. The heterogeneous data sources include home health monitoring devices, smart wearable devices, and chain elderly care service platforms. The multi-dimensional health data includes physiological monitoring data, clinical medical data, behavior and environment data, and psychological and social adaptability data of the elderly; A data access and preprocessing module, which is used to access the information interaction center through a network protocol adaptation component for the multiple heterogeneous data sources, perform processing ability evaluation and data feature extraction on the accessed data sources, generate corresponding data source ability vectors and data feature vectors, and based on the vectors, match and dynamically call corresponding preprocessing rules from the historical hierarchical preprocessing rule library to perform data preprocessing operations; A healthy data set generation module, which is used to perform structure standardization, field alignment and semantic unification on the preprocessed data according to unified data classification standards and semantic mapping rules to generate a unified format of the elderly health data set; A sensitive field vector output module, which is used to perform sensitivity analysis on the fields in the elderly health data set based on medical data standards, industry privacy rules and a preset recognition model and output sensitive field vector identifiers; A desensitization processing module, which is used to construct a data desensitization policy library in combination with medical data standards and industry privacy rules at the initial stage of system deployment, perform sensitive level scoring on each sensitive field vector according to the sensitive field vector identifier, field source and the context environment in the elderly health data set to generate a sensitive level vector, and according to the sensitive level vector, call the matching policy template in the data desensitization policy library to perform desensitization operations and output the desensitized elderly health data set; An access authorization level obtaining module, which is used to input the desensitized elderly health data set, in combination with the user attributes, access context and sensitive field levels of the access requestor, into a pre-constructed access control function to obtain the access authorization level, and perform differential encryption processing on the fields allowed to be accessed by the authorization level; An authorized view encapsulation and data token generation module, which is used to screen the encrypted elderly health data set, extract the fields with access permissions of the current requestor and their corresponding data contents, and generate a data token with an access policy identifier, and transmit it to the access requestor through a communication channel. After receiving the data token, the requestor decrypts the fields and parses the contents according to the access policy identifier contained in the data token to present the corresponding authorized view data.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages: Collect multi-dimensional health data of the elderly from multiple heterogeneous data sources, where the heterogeneous data sources include home health monitoring devices, smart wearable devices, and chain-based elderly care service platforms, and the multi-dimensional health data includes physiological monitoring data, clinical medical data, behavior and environment data, and psychological and social adaptability data of the elderly; Connect the multiple heterogeneous data sources to the information interaction center through a network protocol adaptation component, perform processing capacity evaluation and data feature extraction on the connected data sources to generate corresponding data source capacity vectors and data feature vectors, and based on the vectors, match and dynamically call corresponding preprocessing rules from the historical hierarchical preprocessing rule library to perform data preprocessing operations; According to the unified data classification standard and semantic mapping rules, perform structure standardization, field alignment, and semantic unification on the preprocessed data to generate a unified format of the elderly health data set; Based on medical data standards, industry privacy rules, and a preset recognition model, perform sensitivity analysis on the fields in the elderly health data set and output sensitive field vector identifiers; Build a data desensitization policy library in combination with medical data standards and industry privacy rules at the initial stage of system deployment, perform sensitive level scoring on each sensitive field vector according to the sensitive field vector identifier, field source, and context environment in the elderly health data set to generate a sensitive level vector, and according to the sensitive level vector, call the matching policy template in the data desensitization policy library to perform desensitization operations and output the desensitized elderly health data set; For the elderly health data set after the desensitization operation, combine the user attributes, access context, and sensitive field levels of the access requestor, input them into a pre-built access control function to obtain an access authorization level, and perform differential encryption processing on the fields allowed to be accessed by the authorization level; Screen the encrypted elderly health data set, extract the fields with access permissions and their corresponding data contents of the current requestor, and generate a data token with an access policy identifier, and transmit it to the access requestor through a communication channel. After receiving the data token, the requestor decrypts the fields and parses the contents according to the access policy identifier included in the data token to present the corresponding authorized view data. It solves the problems in the prior art such as the difficulty in fusing heterogeneous health data, insufficient sensitivity in identifying and desensitizing sensitive fields, and inflexible and inaccurate data access control, and realizes the standardized processing of the elderly health data, the intelligent identification and hierarchical desensitization of sensitive information, and the refined authorization management and secure sharing based on the attributes and context of the visitors, improving the security and practicality of data sharing.

[0008] The above description is only an overview of the technical solution of this application. In order to be able to more clearly illustrate the technical means of this application, it can be implemented in accordance with the content of the specification. Moreover, in order to make the above and other purposes, features, and advantages of this application more obvious and understandable, the specific embodiments of this application are specifically described below. Brief Description of the Drawings

[0009] To more clearly illustrate the technical solutions of the embodiments of the present disclosure, the accompanying drawings of the embodiments of the present disclosure will be briefly introduced below. Flowcharts are used in this application to illustrate the operations performed by the systems according to the embodiments of the present application. It should be understood that the operations in the front or below do not necessarily need to be executed precisely in sequence. On the contrary, according to needs, various steps can be executed in reverse order or simultaneously. At the same time, other operations can also be added to these processes, or one or several steps can be removed from these processes.

[0010] Figure 1 It is a schematic flowchart of a method for integrating and sharing elderly health data provided by an embodiment of the present application; Figure 2 It is a schematic structural diagram of an elderly health data integration and sharing platform provided by an embodiment of the present application.

[0011] Explanation of reference numerals: data acquisition module 11, data access and preprocessing module 12, health data set generation module 13, sensitive field vector output module 14, desensitization processing module 15, access authorization level acquisition module 16, authorized view encapsulation and data token generation module 17. Detailed implementation manners

[0012] By providing a method and platform for integrating and sharing elderly health data, the present application solves the problems in the prior art such as the difficulty in fusing heterogeneous health data, insufficient identification and desensitization granularity of sensitive fields, and inflexible and inaccurate data access control, realizes the standardized processing of elderly health data, intelligent identification and hierarchical desensitization of sensitive information, and refined authorization management and secure sharing based on the attributes and context of visitors, and improves the security and usability of data sharing.

[0013] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative efforts shall fall within the protection scope of the present application.

[0014] It should be noted that the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products, or devices.

[0015] Embodiment 1, as Figure 1As shown in the figure, this application provides an integrated sharing method for elderly health data. The method includes: Collect multi-dimensional health data of the elderly from multiple heterogeneous data sources, where the heterogeneous data sources include home health monitoring devices, smart wearable devices, and chain-based elderly care service platforms. The multi-dimensional health data includes physiological monitoring data, clinical medical data, behavior and environment data, and psychological and social adaptability data of the elderly.

[0016] Specifically, the multi-dimensional health data of the elderly collected comes from multiple heterogeneous data sources, specifically including home health monitoring devices, smart wearable devices, and chain-based elderly care service platforms. These data sources have differences in aspects such as data collection devices, communication protocols, data structures, and content dimensions, belonging to a typical heterogeneous environment. Due to the involvement of multiple data providers and user privacy protection, before using these data, it is necessary to ensure that the requesting party authorizes and consents, inform the requesting party of the data usage, and the authorization can be traced and revoked. Among them, home health monitoring devices and smart wearable devices mainly collect physiological monitoring data of the elderly, such as heart rate, blood pressure, blood oxygen, respiratory rate, body temperature, and sleep status, etc.; the chain-based community elderly care service platform can obtain behavior and environment data, such as daily activity trajectories, eating habits, environmental temperature, humidity, and air quality, etc., and also includes data related to psychological and social adaptability, such as loneliness assessment, psychological counseling records, social interaction frequency, and community participation. By comprehensively collecting data from the above multiple dimensions, it provides basic data support for the comprehensive perception and dynamic assessment of the health status of the elderly.

[0017] Connect the multiple heterogeneous data sources to the information interaction center through a network protocol adaptation component, perform a processing capacity assessment and data feature extraction on the connected data sources, generate corresponding data source capacity vectors and data feature vectors, and based on the vectors, match and dynamically call corresponding preprocessing rules from the historical hierarchical preprocessing rule library to perform data preprocessing operations.

[0018] Specifically, multiple heterogeneous data sources, including home health monitoring devices, smart wearable devices, and chained elderly care service platforms, have significant heterogeneity due to differences in manufacturers, communication protocols, and data structures. To achieve efficient access and subsequent processing of these data sources, the system sets up a network protocol adaptation component to complete the identification and standardized encapsulation of underlying communication protocols. Specifically, the network protocol adaptation component includes a protocol identification module and an adaptation driver module. The protocol identification module is responsible for automatically identifying the communication protocol types used by each data source (such as MQTT, HTTP, HL7, CoAP), parsing the data packet structure, and extracting key fields. The adaptation driver module then loads the corresponding driver template according to the protocol identification result, realizes the unified encapsulation and structure transformation of the data format, generates an intermediate data object, and pushes it to the information interaction center of the system. After completing the data access, the system constructs a hierarchical preprocessing rule library for different types of data sources to support subsequent data preprocessing tasks. The rule library is hierarchically designed according to the processing capabilities and data characteristics of the data sources, covering data cleaning and format standardization schemes of different complexities, and adapting to the actual situations such as limited computing power of edge devices and large differences in data integrity.

[0019] Furthermore, it includes: Conduct a capacity assessment on the multi-dimensional health data of the elderly from the multiple heterogeneous data sources. The capacity assessment includes indicators such as computing power level, communication frequency, data integrity, supported protocol type, and data update frequency, and outputs the corresponding data source capacity vector; Extract data feature vectors from the multi-dimensional health data of the elderly from multiple heterogeneous data sources. The data feature vectors include structural features, field types, timestamp formats, and unit types; Based on the data source capacity vector and the data feature vector, match the rule template set in the historical hierarchical preprocessing rule library, and select the rule template that matches the data source capacity level and data characteristics; Dynamically call the preprocessing rules corresponding to the rule template. The preprocessing rules include data deduplication, missing value processing, unit standardization, data format conversion, time alignment, outlier correction, and tagging processing.

[0020] Specifically, first, construct the hierarchical preprocessing rule library. This rule library is organized based on rule templates, and each rule template encapsulates a set of preprocessing strategies applicable to specific ability levels and data characteristics, supporting on-demand combination and dynamic expansion. The rule templates cover various data cleaning and standardization operation processes, including data deduplication, missing value processing, unit standardization, data format conversion, time alignment, outlier correction, and tagging operation modules. To achieve the precise adaptation of the rule templates, automatically evaluate the operation and data interaction capabilities of multiple heterogeneous elderly health data sources connected, ensuring that the subsequent preprocessing strategies match the actual capabilities of the devices. Specifically, when a data source is connected, the system completes communication protocol identification and connection initialization through the network protocol adaptation component, and reads the device information and data structure characteristics of the data source. Subsequently, the system continuously monitors and quantitatively analyzes five core ability indicators of the data source during actual operation, including evaluating the computing power level of whether the data source itself has data preprocessing, encryption calculation, and local computing capabilities through system response latency and data processing time, evaluating the communication frequency by counting the number of data reports per unit time to judge the data push frequency and stability of the data source, evaluating the data integrity by detecting whether there are missing fields, format errors, and verification failure ratios in the data packets, evaluating whether it is compatible with the standard communication protocol by identifying and recording the transmission protocol types supported by the data source, and evaluating the data update frequency by calculating the average number of data updates per hour based on the timestamps recorded in the historical data to determine the activity level of the data source. The system normalizes the above evaluation results into a numerical vector, that is, the data source ability vector, and dynamically matches and invokes the policy templates applicable to the data source level in the preprocessing rule library according to the scores of each ability dimension in the ability vector, realizing the selection of differentiated and hierarchical preprocessing solutions, thereby improving the compatibility and processing efficiency of the system for heterogeneous data sources. Then, the system extracts features from the connected multi-dimensional elderly health data to generate a data feature vector, including structural features to determine whether the data is a flat structure, a nested structure, or a multi-layer JSON structure; field types to determine whether each field is of numerical, text, or timestamp type; timestamp formats to determine whether the representation of the time field is consistent; and unit types to confirm whether the numerical values in the data fields have different units. Subsequently, the system performs two-way matching in the rule library based on the data source ability vector and the data feature vector, combines the built-in rule template indexing mechanism and the historical successful preprocessing case-driven mechanism, and selects the rule template that best conforms to the actual situation of the current data source. Among them, the case-driven mechanism can dynamically recommend the optimal rule path based on historical processing records and effect feedback, thereby improving the adaptability and effect stability of the preprocessing. Finally, the system dynamically invokes the preprocessing rules corresponding to the rule template to perform data cleaning and standardization operations on the connected multi-dimensional elderly health data.Each preprocessing includes data deduplication, removing duplicate records by using the primary key field and the overlap of time windows; missing value handling, selecting strategies such as mean filling, interpolation, and default value completion based on field type and historical distribution; unit standardization, unifying the units of indicators such as blood pressure and weight to the system standard; format conversion, including time format standardization and encoding format unification; time alignment, aligning and resampling data with different sampling frequencies or delayed acquisitions; outlier correction and tagging processing, correcting and tagging data that deviate from the range by combining set threshold rules.

[0021] Furthermore, the matching of the data source ability vector and the data feature vector with the rule template set in the historical hierarchical preprocessing rule library, and selecting the rule template that matches the data source ability level and data characteristics includes: Construct and maintain a historical hierarchical preprocessing rule library, which stores the preprocessing strategies and their processing effects adopted by previously connected data sources under different ability levels and data characteristics in a classified, hierarchical, and templated manner; Define the formula for the rule template matching degree function: M i = ; where M i is the matching degree score, X is the data source ability vector, T is the data feature vector, R i is the i-th rule template in the historical hierarchical preprocessing rule library, sim is the similarity calculation function, and λ1 and λ2 are weight parameters that satisfy λ1 + λ2 = 1 and are used to adjust the importance of ability matching and feature matching; Based on the processing effect indicators of the historical data sources recorded in the historical hierarchical preprocessing rule library when using the rule template R i , calculate the historical effect score H i ; The processing effect indicators include: the processing success rate P, processing delay D, and processing accuracy A of the historical data source when executing the rule template R i ; The processing success rate P is obtained by statistically calculating the proportion of the number of data that successfully completed the preprocessing task to the total number of processing times of the rule template R i in previous actual operations; The processing delay D is obtained by calculating the average time consumption from data reception to processing completion of the rule template R i in previous operations; The processing accuracy A is obtained by comparing the data results generated by the rule template R i in previous uses with the manually marked standard results and statistically calculating the proportion of the number of accurate data entries to the total number of entries; Calculate the historical effect score H based on the above metrics i , and its calculation formula is: H i = μ1×P + μ2×(1-D) + μ3×A; where P represents the processing success rate, D represents the processing delay, A represents the processing accuracy, and μ1, μ2, μ3 are weighting coefficients satisfying μ1 + μ2 + μ3 = 1; Perform weighted fusion on the matching degree score Mi and the historical effect score H i to generate a matching score M i ' that integrates historical experience, and its calculation formula is: M i ' = αM i + βH i ; where α and β are real-valued weight coefficients and α + β = 1. They can be configured according to different requirements for the weight between the adaptation degree of the current data source characteristics and the historical application effect in actual applications. By default, α = 0.5, β = 0.5, M i is the matching degree score, H i is the historical effect score, and M i ' is the rule template matching score after integrating historical experience; Sort all rule templates based on the matching score M i ' that integrates historical experience, select the rule template with the highest score as the preprocessing rule template for the current data source, and perform corresponding preprocessing operations on the data uploaded by the data source.

[0022] According to the unified data classification standard and semantic mapping rules, perform structure standardization, field alignment, and semantic unification on the preprocessed data to generate a unified format elderly health data set.

[0023] Furthermore, it includes: Preset a unified data classification standard, and map data items with the same semantics but different field names in heterogeneous data sources to unified field identifiers through a field label mapping module. The mapping relationship of the field label mapping module is stored in the semantic mapping rule library.

[0024] Perform unified conversion on the data structures from different sources, including flattening the data dimensions, disassembling multi-level nested fields, unifying the timestamp format, and performing unified unit conversion to generate a data structure that meets the platform standard.

[0025] Perform field alignment processing on the data after structure standardization to ensure that records from different sources are logically within the same field sequence structure, and complete data value range mapping based on the standard value range set.

[0026] Introduce a semantic conflict detection algorithm to identify data items with semantic conflicts from different data sources, trigger conflict correction strategies, complete automatic screening according to trust weights and time priority rules, and generate a unified format elderly health dataset.

[0027] Specifically, first, the system presets a unified data classification standard as a reference benchmark for the platform to classify various elderly health data. Through the field label mapping module, data items with the same semantics but different field names (such as "heart rate", "HR", "heart_rate") in multiple heterogeneous data sources are uniformly mapped to standard field identifiers. For example, all fields representing "heart rate" can be uniformly identified as standard_heart_rate through the mapping rule library. The above field mapping relationships are stored in the semantic mapping rule library in key-value pair form, supporting continuous updates and expansions.

[0028] Second, the system performs unified conversion operations on the data structures from different data sources, mainly including: flattening the dimensions of complex structures (such as expanding JSON multi-level nesting into a flat table structure), deconstructing and separating multi-level nested fields, unifying time fields from different sources into the ISO8601 format (such as yyyy-MM-dd HH:mm:ss), and converting and normalizing metric units (such as unifying "kg" and "lbs" into "kg"). The above conversion operations ensure that all data conforms to the structure specifications required by the platform, facilitating subsequent processing.

[0029] Third, after completing the structure unification, the system performs field alignment processing on the data from different sources, that is, aligns the records from different sources according to the unified field arrangement order, so that the data logically belonging to the same indicator is in the same field position. Further, the system performs mapping conversion on the field values based on the standard value range set (for example, mapping "high", "medium", "low" risk ratings to "3", "2", "1" respectively) to ensure the consistency of data value ranges between different data sources, facilitating statistical analysis and modeling applications.

[0030] Finally, to avoid semantic ambiguity and incorrect data fusion, the system introduces a semantic conflict detection algorithm to identify possible semantic conflicts in similar fields from different data sources. For example, if the unit of the "body temperature" field in two data sources is inconsistent or the meaning is different (one represents "oral temperature" and the other represents "axillary temperature"), the system will trigger the semantic conflict correction mechanism. Based on the preset conflict handling rules, this mechanism automatically screens or fuses the optimal values by considering factors such as the trust weight of the field source and the priority of data update time, thus finally generating a semantically unified and structurally standardized elderly health dataset.

[0031] Based on medical data standards, industry privacy rules, and a preset recognition model, perform sensitivity analysis on the fields in the elderly health dataset and output sensitive field vector identifiers; Further, it includes: Load medical data standards and industry privacy rules, construct a sensitive field basic dictionary for field recognition, the sensitive field basic dictionary is used to support the extraction of sensitive fields, and identify and mark potential sensitive fields in the unified format elderly health dataset; Extract, from the unified format elderly health dataset, field feature vectors including field name, data type, historical access frequency, context module attribution information of the field in the data structure, and semantic similarity with known sensitive fields; Input the field feature vectors into a preset sensitivity recognition model. The sensitivity recognition model includes matching the field name with the sensitive field basic dictionary based on a rule engine, calculating the sensitivity score of each field based on word vectors and structure embedding algorithms, and outputting the field sensitivity score result; According to predefined level mapping rules, map the field sensitivity scores to sensitive levels, generate a sensitive field level table, and encode the sensitive field level table into a fixed-length vector in the order of the original data fields, and output sensitive field vector identifiers.

[0032] Specifically, the system first loads medical data standards that comply with national industry regulations, such as the "Electronic Medical Record Standard" and the "PHI Definition Guide", and privacy protection rules, such as the "Personal Information Protection Law" and the "GDPR", extracts typical sensitive fields defined therein, such as ID numbers, medical records, disease diagnoses, contact information, and financial information, and constructs a basic dictionary of sensitive fields for field identification. This dictionary contains not only the standard names of sensitive fields but also common synonymous expressions, abbreviated forms, and classification labels. Analyze each field in the elderly health dataset in the unified format, and extract the field name and semantic label, field data type, access frequency of the field in the historical data access record, context module attribution information of the field in the data structure, and semantic similarity with the fields in the basic dictionary of sensitive fields to form a field feature vector. Input the above field feature vector into a preset sensitivity recognition model. The sensitivity recognition model includes three types of recognition methods, including dictionary matching of field names based on a rule engine to quickly identify sensitive fields with directly matching names; introducing an incremental learning mechanism to dynamically adjust model parameters according to user feedback, scenario changes, and data drift to achieve adaptive optimization of the model in different deployment environments. Based on semantic calculation methods and structure embedding algorithms, judge the similarity with standard sensitive fields to identify sensitive fields under fuzzy expressions and new representation methods; the system determines the sensitivity level of each field according to predefined sensitivity level mapping rules, combined with the model judgment results and industry classification standards, to form a complete sensitivity level table of sensitive fields. Finally, according to the arrangement order of fields in the unified format dataset, standardize and encode the sensitivity level of each field to generate a fixed-length sensitive field vector identifier.

[0033] At the initial stage of system deployment, construct a data desensitization strategy library in combination with medical data standards and industry privacy rules. According to the sensitive field vector identifier, field source, and context environment in the elderly health dataset, perform a sensitivity level score on each sensitive field vector to generate a sensitivity level vector, and according to this sensitivity level vector, call the matching strategy template in the data desensitization strategy library to perform desensitization operations and output the desensitized elderly health dataset; Furthermore, it includes: At the initial stage of system deployment, construct a data desensitization strategy library based on general medical data standards and industry privacy rules. The data desensitization strategy library is managed by sensitivity level grading, and each sensitivity level corresponds to a desensitization strategy template. Each desensitization strategy template clearly stipulates the desensitization processing method and specific applicable conditions for the corresponding sensitivity level; Collect the context environment information of the sensitive field vector in the elderly health dataset, including field source, usage scenario, data call frequency and access behavior trajectory, field historical desensitization record, and feedback result; Based on a preset sensitive level scoring function, form a sensitive level vector S in the order of fields; According to the field level values in the sensitive level vector S, match the corresponding policy templates in the data desensitization policy library. The policy templates include performing de-identification on highly sensitive fields, performing masking on moderately sensitive fields, performing generalization on low-sensitivity fields, and performing pseudo-value replacement on fields in special scenario fields; Perform the data desensitization operations corresponding to the policy templates on the sensitive fields, generate a set of data sub-views containing processing results of different sensitive levels, and bind specific access authorization rules to each data sub-view to adapt to the sharing needs of users with different permission roles.

[0034] Specifically, in the initial stage of system deployment, a data desensitization policy library is constructed based on general medical data standards and industry privacy protection rules. The policy library organizes multiple preset policy templates according to field sensitive levels. Each level of policy template defines the corresponding desensitization processing methods and their applicable conditions. For example, highly sensitive level fields are suitable for de-identification (such as hash substitution, ID removal), medium level fields are suitable for masking (such as partial covering), low level fields can use generalization processing (such as fuzzy description, interval expression), and fields in specific sensitive scenarios can adopt methods such as pseudo-value replacement. To achieve accurate assessment of sensitive levels, the system collects the context information of each sensitive field in the dataset, including field sources (such as medical institutions, home devices, wearable devices), usage scenarios (such as assisted diagnosis and treatment, remote monitoring, behavior analysis), data call frequency and access behavior trajectories, field historical desensitization records and user feedback results. The system inputs each sensitive field vector into a preset sensitive level scoring function, which comprehensively scores in dimensions such as the inherent sensitivity of the field itself (such as the essential attributes of name, ID number), the risk level of the field appearance environment (such as whether it is used in public sharing scenarios), the field usage frequency and exposure risk (access frequency and cross-scenario reference degree), and the legal supervision intensity area where it is located (the impact of different regional regulations on desensitization requirements). The scoring results are used to assign sensitive levels to each field in the form of labels, and a sensitive level vector S with a fixed structure is formed in the original order of the fields. Subsequently, the system matches the corresponding policy templates from the data desensitization policy library according to the level values of each field in the sensitive level vector S, and dynamically performs the corresponding data desensitization operations. Specifically, it includes: performing de-identification processing on highly sensitive fields; performing masking on moderately sensitive fields; performing generalization on low-sensitivity fields; performing pseudo-value replacement and other methods on fields in special protection scenarios. Finally, the system generates a set of data sub-views containing processing effects of different sensitive levels based on the above desensitization processing results, and specific access authorization rules can be bound to each sub-view.

[0035] Further, the preset sensitive level scoring function includes: Obtain the sensitive fields in the health dataset of the elderly, and construct a sensitive level scoring function formula based on the inherent sensitivity, usage scenario risk level, environmental exposure risk, and legal supervision intensity of these fields to score each sensitive field and generate a sensitive level vector S; The sensitive level scoring function formula is: ; Among them, S k represents the sensitive level score of the kth field; S k1 represents the inherent sensitivity score of this field; S k2 represents the usage scenario risk level score of this field; S k3 represents the environmental exposure risk score of this field; S k4 represents the legal supervision intensity score of this field; a i is the weight coefficient of the scoring factor, satisfying: , a i ∈[0,1]. Among them, a1 is the weight coefficient of inherent sensitivity, a2 is the usage scenario risk, a3 is the environmental exposure risk, and a4 is the weight coefficient of legal supervision intensity. The value of each weight coefficient is preset according to the actual situation.

[0036] Specifically, the inherent sensitivity score S k1 is statically configured by the system referring to the data classification guidelines published by the state (such as the "Personal Information Protection Law" and the "Basic Specifications for Electronic Medical Records"). The scenario risk level score sk2 is used to evaluate the potential sensitivity risk of the sensitive field in the current access scenario. The scoring value range is [0,1], and the higher the value, the greater the usage scenario risk. The system dynamically calculates s according to the access request context information k2, The specific evaluation method is as follows: Construct the following weighted scoring function sk2 = β1*u + β2*i + β3*p + β4*d, where u represents the data usage risk factor, which is determined according to the usage field marked in the access request. For example, the value assigned for scientific research desensitization analysis is 0.3, and the value assigned for third-party platform data sharing is 0.9; i represents the interface type risk factor, which is evaluated according to the API type of the call field. For example, the value assigned for an internal controlled interface is 0.3, and the value assigned for an external open interface is 0.8; p represents the docking platform trust level factor, which is determined by identifying the registration information and authentication qualifications of the access platform. For example, the value assigned for the self-owned platform is 0.3, and the value assigned for a non-authenticated platform is 1.0; d represents the data flow openness factor, which is used to indicate whether there is a risk of chain transmission of data. For example, the value assigned for only local caching is 0.2, and the value assigned for supporting multi-level sharing is 1.0. The four weight coefficients β1, β2, β3, β4 ∈ [0, 1], and their values are preset according to the system policy and satisfy the sum of 1. The above factor data are all automatically extracted by the system from the access request header information, platform access registry, interface metadata, and access behavior logs, and finally generate the value of sk2 for the sensitive level scoring function. The environmental exposure risk score (s k3 ) is used to quantify the risk degree of a certain field being illegally leaked or unauthorizedly propagated in the current data access environment. Its scoring value range is [0, 1], and the higher the value, the greater the risk. The system comprehensively evaluates the following multi-dimensional environmental factors by constructing a weighted scoring function. The specific weighted scoring function is: , where n is the network security level factor, which is assigned according to the network environment detected by the system for the access request. For example, the intranet of a hospital and a dedicated VPN network are assigned 0.2, the HTTPS-encrypted public network is assigned 0.5, and the ordinary public network and insecure network (HTTP) are assigned 0.9; e is the device credibility factor, which is judged according to the registered identity, authentication information and historical behavior of the access device. For example, a hospital-specific device is assigned 0.2, a registered mobile device is assigned 0.5, and an unknown and abnormal device is assigned 0.9; t is the access time compliance factor. The system checks whether the request occurs within the specified business time window: for example, the working period is assigned 0.3, and the night and holiday periods are assigned 0.8; f is the access frequency abnormality factor. The system conducts behavioral analysis on the access frequency of the visitor to this field within a unit time: for example, the normal frequency is assigned 0.2, the slightly higher frequency is assigned 0.6, and the abnormally high-frequency access is assigned 0.9. The weight coefficients γ1, γ2, γ3, γ4 ∈ [0, 1] are used to adjust the influence degree of each environmental factor, and the sum satisfies γ1 + γ2 + γ3 + γ4 = 1. Their values are preset by the system administrator according to specific industry supervision requirements and business risk models. The original data of the above factors are automatically collected through the system access control gateway, device fingerprint recognition module, time management module and access behavior audit log, and sk3 is dynamically calculated by the environmental risk assessment module, which is an important input in the sensitive level scoring function and is used to finally evaluate the sensitive level of the field. The legal supervision intensity score (s k4 ) is used to evaluate the supervision intensity of data fields under different laws, regulations and privacy policies. The system first integrates multiple legal and regulatory databases, including national laws (such as the Personal Information Protection Law), industry regulations (such as the Medical Information Protection Management Measures) and international laws (such as GDPR). Through natural language processing technology, the system can automatically identify sensitive data (such as health information, personal identity information, etc.) in the field content, and match these data with relevant clauses in the regulatory database to determine whether the field belongs to the scope of sensitive data protected by law. Secondly, the system will evaluate the applicable legal supervision intensity according to the geographical location of data storage and processing. The system assigns a legal supervision intensity score to each field according to the intensity of regulatory matching and geographical applicability, using a preset scoring standard. The specific scoring standard is as follows: For fields involving highly sensitive data (such as health data), they will be rated as high supervision intensity (score 0.9 to 1.0); for fields protected by industry regulations, they will be medium supervision intensity (score 0.6 to 0.8); and for fields with less control, they will be low supervision intensity (score 0.3 to 0.5). Some fields may not be subject to specific legal control, so they will be given a low score (0.1 to 0.3). Finally, the system will generate the corresponding legal supervision intensity score s for each field according to these evaluation results k4。

[0037] For the dataset of the health conditions of the elderly after the desensitization operation, taking into account the user attributes, access context, and sensitive field levels of the access requester, input them into a pre-built access control function to obtain an access authorization level, and perform differential encryption processing on the fields allowed to be accessed at that authorization level.

[0038] Specifically, the system performs access control processing on the dataset of the health conditions of the elderly that has completed desensitization. First, encode the user attributes of the access requester (including the identity category of the requester, the affiliated institution, historical access behavior records, permission level, and data processing qualifications, etc.) into a user attribute vector; at the same time, extract the context information at the time of the access behavior (including access time, device type, network environment, access frequency, geographical location, and session behavior pattern, etc.) to form a context environment vector; then combine the generated sensitive level vector, which records the sensitivity score values of each field. The system takes the above three types of vectors as inputs and sends them into a pre-built access control function, generates an access comprehensive score through weighted calculation, and determines the authorization level of the current access request according to the authorization level interval to which the comprehensive score belongs (such as full access, partial access, restricted access, or denied access). Subsequently, the system determines the range of accessible fields according to this access authorization level, and adopts differential encryption strategies to process the content of fields whose sensitive levels are higher than the current authorization level threshold, such as encrypting and controlling high-sensitivity fields based on attribute-based encryption (ABE) technology, and implementing protection strategies such as masking or generalization for medium-sensitivity fields, so as to ensure controllable sharing of sensitive data based on user roles and access environments while meeting the requirements of data security and privacy protection.

[0039] Furthermore, the access control function includes: The access control function f(U, C, S) obtains a comprehensive score Score by performing weighted summation on three types of factors. The three types of factors include the user attribute vector U = {u1, u2, u3, u4, u5}, where u1 is the identity category of the requesting user, u2 is the affiliated institution, u3 is the historical access behavior, u4 is the permission level, u5 is the data processing qualification, the context environment vector C = {c1, c2, c3, c4, c5, c6}, where c1 is the access time, c2 is the device type, c3 is the network environment, c4 is the access frequency, c5 is the geographical location, c6 is the session behavior pattern; the sensitive level vector S = {s1, s2, …, s n}, and the sensitivity level is expressed in numerical form and is divided into high-sensitivity fields, medium-sensitivity fields, low-sensitivity fields, and non-sensitivity fields according to the numerical values; The formula for the comprehensive score Score is: ; where, 、 、 They are the weight coefficients of the user attribute vector, the weight coefficient of the context environment vector, and the weight coefficient of the sensitive level vector, respectively, satisfying = 1, = 1, = 1 normalization conditions. The user attribute information vector is the i-th attribute score in the user attribute vector , and the value of has been normalized to the interval [0, 1]. The higher the value, the more credible the attribute; is the j-th environmental factor score in the context environment vector , and is the k-th sensitive field level score in the sensitive level vector ; According to the value range of the score Score, determine the access authorization level L. The specific rules are as follows: When Score ≥ T full , the access level L = 3, indicating full access. The system will return the desensitized elderly health data set; When T partial ≤ Score < T full , the access level L = 2, indicating partial field access. The system will perform attribute-based encryption (ABE) processing on the high-sensitive field content of the desensitized elderly health data set and return the desensitized elderly health data set for medium and low-sensitive fields and non-sensitive fields; When T mask ≤ Score < T partial , the access level L = 1, indicating desensitized access. The system hides high-sensitive fields, performs masking processing on medium-sensitive fields, and performs generalization processing on low-sensitive fields; When Score < T mask , the access level L = 0, indicating access rejection. The system will only return a structure null value; Among them, T full , T partial , T mask are preset thresholds, satisfying T full > T partial > T mask , and the default value of T full is 0.75, the default value of T partial is 0.50, and the default value of T mask is 0.30.

[0040] The encrypted elderly health data set is screened to extract the fields and their corresponding data content to which the current requester has access rights, and a data token with an access policy identifier is generated and transmitted to the access requester through a communication channel. After receiving the data token, the requester decrypts the fields and parses the content according to the access policy identifier contained in the data token, and presents the corresponding authorized view data.

[0041] Further, the data token includes: A view identifier field used to uniquely identify the authorization view number; A policy summary field used to identify the access control policy applicable to this data view, and the policy summary field is generated based on the access authorization level obtained by the access control function f(U, C, S); A key index field generated based on the attribute vector U of the requester for verifying the encryption and decryption of sensitive fields by the access subject; A timeliness identifier field that limits the access time range of this token; An encrypted check field used to protect the integrity of the above token fields to prevent man-in-the-middle attacks and illegal tampering.

[0042] Specifically, the system organizes the screened fields and their corresponding data content in a unified format to construct an authorized view. This authorized view is a subset of health data that the current requester is allowed to access in the current access scenario, and its content has been desensitized and encrypted according to the authorization policy to ensure data privacy and compliance in use. To support the permission verification and secure transmission during the data access process, the system generates a data token with an access policy identifier for the authorized view, which is used for secure transmission and permission verification, and embeds multiple structured fields, including a view identifier field used to uniquely identify the authorization view number to achieve view isolation and access record tracking for different access requests; a policy summary field used to describe the access control policy, and its value is generated based on the access authorization level output by the access control function f(U, C, S), corresponding to the access control policy summary hash applicable to the current access, and is used for the recipient to verify the policy source and access boundary; a key index field for key matching when the access subject decrypts the data, which is calculated by the attribute vector U of the access requester to ensure that the requester can only obtain the decryption key of the sensitive fields within its permission range; a timeliness identifier field used to set the valid time range of the data token, which clearly stipulates the token's effective time and expiration time to prevent expired access; and an integrity protection field for anti-tampering and integrity verification, which is hashed and encrypted by the system using a preset key for the above token fields to ensure that the data token has not been attacked by a man-in-the-middle and illegally tampered with during transmission. After generation, the system transmits the data token and the corresponding authorized view through the communication channel Figure 1and transmit it to the access requester. After receiving the response content, the access requester can complete the access permission verification and field decryption based on the data token, and finally display the field content in the authorization view on its interface, realizing the legal, secure, and accurate shared access to the health data of the elderly.

[0043] In summary, the embodiments of the present application have at least the following technical effects: Collect multi-dimensional health data of the elderly from multiple heterogeneous data sources, where the heterogeneous data sources include home health monitoring devices, smart wearable devices, and elderly care service platforms, and the multi-dimensional health data includes physiological monitoring data, clinical medical data, behavior and environment data, and psychological and social adaptability data of the elderly; connect the multiple heterogeneous data sources to the information interaction center through a network protocol adaptation component, perform processing ability evaluation and data feature extraction on the connected data sources, generate corresponding data source ability vectors and data feature vectors, and based on the vectors, match and dynamically call corresponding preprocessing rules from the historical hierarchical preprocessing rule library to perform data preprocessing operations; perform structure standardization, field alignment, and semantic unification on the preprocessed data according to a unified data classification standard and semantic mapping rule to generate a unified format of the elderly health data set; perform sensitivity analysis on the fields in the elderly health data set based on medical data standards, industry privacy rules, and a preset recognition model and output sensitive field vector identifiers; construct a data desensitization policy library in combination with medical data standards and industry privacy rules at the initial stage of system deployment, perform sensitive level scoring on each sensitive field vector according to the sensitive field vector identifier, field source, and the context environment in the elderly health data set to generate a sensitive level vector, and according to the sensitive level vector, call the matching policy template in the data desensitization policy library to perform desensitization operations and output the desensitized elderly health data set; For the elderly health data set after the desensitization operation, combine the user attributes, access context, and sensitive field level of the access requester, input them into a pre-constructed access control function to obtain the access authorization level, and perform differential encryption processing on the fields allowed to be accessed by the authorization level; screen out the fields and their corresponding data content that the current requester has access rights from the encrypted elderly health data set, generate a data token with an access policy identifier, and transmit it to the access requester through a communication channel, and the requester views the field content in the authorization view.

[0044] Embodiment 2 Based on the same inventive concept as the integrated sharing method of a kind of elderly health data in the foregoing embodiment, as Figure 2 shown, the present application provides an integrated sharing platform for elderly health data. The platform and method embodiments in the present application are based on the same inventive concept. Among them, the platform includes: Data acquisition module 11, which is used to collect multi-dimensional health data of the elderly from multiple heterogeneous data sources. The heterogeneous data sources include home health monitoring devices, smart wearable devices, and elderly care service platforms. The multi-dimensional health data includes physiological monitoring data, clinical medical data, behavior and environment data, and psychological and social adaptability data of the elderly; Data access and preprocessing module 12, which is used to connect the multiple heterogeneous data sources to the information interaction center through a network protocol adaptation component, perform processing ability evaluation and data feature extraction on the accessed data sources, generate corresponding data source ability vectors and data feature vectors, and based on the vectors, match and dynamically call corresponding preprocessing rules from the historical hierarchical preprocessing rule library to perform data preprocessing operations; Health data set generation module 13, which is used to perform structure standardization, field alignment, and semantic unification on the preprocessed data according to unified data classification standards and semantic mapping rules, and generate a unified format of the elderly health data set; Sensitive field vector output module 14, which is used to perform sensitivity analysis on the fields in the elderly health data set based on medical data standards, industry privacy rules, and a preset recognition model, and output sensitive field vector identifiers; Desensitization processing module 15, which is used to construct a data desensitization strategy library in combination with medical data standards and industry privacy rules at the initial stage of system deployment, perform sensitive level scoring on each sensitive field vector according to the sensitive field vector identifier, field source, and the context environment in the elderly health data set, generate a sensitive level vector, and according to the sensitive level vector, call the matching policy template in the data desensitization strategy library to perform desensitization operations, and output the desensitized elderly health data set; Access authorization level acquisition module 16, which is used to input the desensitized elderly health data set, combined with the user attributes, access context, and sensitive field level of the access requestor, into a pre-built access control function, obtain the access authorization level, and perform differential encryption processing on the fields allowed to be accessed by the authorization level; Authorized view encapsulation and data token generation module 17, which is used to screen the encrypted elderly health data set, extract the fields and their corresponding data contents that the current requestor has access rights to, and generate a data token with an access policy identifier, and transmit it to the access requestor through a communication channel. After receiving the data token, the requestor decrypts the fields and parses the contents according to the access policy identifier included in the data token to present the corresponding authorized view data.

[0045] It should be noted that the above order of the embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. Moreover, the above specific embodiments of this specification have been described. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in a different order than in the embodiments and still achieve the desired results. Additionally, the processes depicted in the drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0046] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included within the protection scope of the present application.

[0047] This specification and the drawings are only exemplary descriptions of the present application and are considered to have covered any and all modifications, variations, combinations, or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications.

Claims

1. An integrated sharing method for elderly health data, characterized in that, Including: Collecting multi-dimensional health data of the elderly from multiple heterogeneous data sources, where the heterogeneous data sources include home health monitoring devices, smart wearable devices, and chain elderly care service platforms, and the multi-dimensional health data includes physiological monitoring data, clinical medical data, behavior and environment data, and psychological and social adaptability data of the elderly; Connecting the multiple heterogeneous data sources to the information interaction center through a network protocol adaptation component, performing processing capacity evaluation and data feature extraction on the connected data sources, generating corresponding data source capacity vectors and data feature vectors, and based on the vectors, matching and dynamically invoking corresponding preprocessing rules from the historical hierarchical preprocessing rule library to perform data preprocessing operations; According to the unified data classification standard and semantic mapping rules, performing structure standardization, field alignment, and semantic unification on the preprocessed data to generate a unified format elderly health data set; Based on medical data standards, industry privacy rules, and a preset recognition model, performing sensitivity analysis on the fields in the elderly health data set and outputting sensitive field vector identifiers; Constructing a data desensitization policy library in combination with medical data standards and industry privacy rules at the initial stage of system deployment, performing sensitive level scoring on each sensitive field vector according to the sensitive field vector identifier, field source, and context environment in the elderly health data set to generate a sensitive level vector, and according to the sensitive level vector, invoking the matching policy template in the data desensitization policy library to perform desensitization operations and outputting the desensitized elderly health data set; For the elderly health data set after the desensitization operation, combining the user attributes, access context, and sensitive field level of the access requestor, inputting them into a pre-built access control function to obtain an access authorization level, and performing differential encryption processing on the fields allowed to be accessed by the authorization level; Screening the encrypted elderly health data set, extracting the fields and their corresponding data contents that the current requestor has access rights to, and generating a data token with an access policy identifier, and transmitting it to the access requestor through a communication channel. After receiving the data token, the requestor decrypts the fields and parses the contents according to the access policy identifier included in the data token to present the corresponding authorized view data.

2. The integrated sharing method of elderly health data according to claim 1, characterized in that, The operation of connecting the multiple heterogeneous data sources to the information interaction center through a network protocol adaptation component, performing processing capacity evaluation and data feature extraction on the connected data sources, generating corresponding data source capacity vectors and data feature vectors, and based on the vectors, matching and dynamically invoking corresponding preprocessing rules from the historical hierarchical preprocessing rule library to perform data preprocessing operations includes: Performing capacity evaluation on the multi-dimensional health data of the elderly from the multiple heterogeneous data sources, where the capacity evaluation includes computing power level, communication frequency, data integrity, supported protocol type, and data update frequency metrics, and outputting corresponding data source capacity vectors; Extracting data feature vectors from the multi-dimensional health data of the elderly from multiple heterogeneous data sources, where the data feature vectors include structure features, field types, timestamp formats, and unit types; Based on the matching between the data source capability vector and the data feature vector with the rule template set in the historical hierarchical preprocessing rule library, select the rule template that matches the data source capability level and data features; Dynamically call the preprocessing rules corresponding to the rule template, and the preprocessing rules include data deduplication, missing value processing, unit standardization, data format conversion, time alignment, outlier correction, and tagging.

3. The integrated sharing method of elderly health data according to claim 2, wherein The step of selecting the rule template that matches the data source capability level and data features based on the matching between the data source capability vector and the data feature vector with the rule template set in the historical hierarchical preprocessing rule library includes: Construct and maintain a historical hierarchical preprocessing rule library, which stores the preprocessing strategies and their processing effects adopted by previously accessed data sources under different capability levels and data features in a classified, hierarchical, and templated manner; Define the rule template matching degree function formula: M i = ； Among them, M i is the matching degree score, X is the data source capability vector, T is the data feature vector, R i is the i-th rule template in the historical hierarchical preprocessing rule library, sim is the similarity calculation function, and λ1 and λ2 are weight parameters satisfying λ1 + λ2 = 1, which are used to adjust the importance of ability matching and feature matching; Based on the processing effect indicators of the historical data sources recorded in the historical hierarchical preprocessing rule library when using the rule template R i calculate the historical effect score H i ; The processing effect indicators include: the historical data source when executing the rule template R i with a processing success rate P, a processing delay D, and a processing accuracy A; The processing success rate P is obtained by counting the data times of successfully completing the preprocessing tasks in the previous actual operation of the rule template R i as the proportion of the number of times of successfully completing the preprocessing tasks in the previous actual operation to the total number of processing times; The processing delay D is obtained by calculating the average time taken from data reception to processing completion in previous runs for the rule template R i in previous runs; The processing accuracy A is obtained by comparing the data results generated in past uses with the manually marked standard results for the rule template R i and counting the proportion of the number of accurate data entries to the total number of entries Calculate the historical effect score H based on the above indicators i , and its calculation formula is as follows: H i = μ1×P + μ2×(1 - D) + μ3×A; Where P represents the processing success rate, D represents the processing delay, A represents the processing accuracy, and μ1, μ2, μ3 are weighting coefficients, satisfying μ1 + μ2 + μ3 = 1; Compare the matching degree score Mi with the historical effect score H i Perform weighted fusion to generate a matching score M i ′, and its calculation formula is: M i ′ = αM i + βH i ; Among them, α and β are weight coefficients that are real numbers and α + β = 1. They can be configured according to different requirements for the weight between the adaptation degree of the current data source characteristics and the historical application effect in actual applications. By default, α = 0.5 and β = 0.5, M i is the matching degree score, H i is the historical effect score, M i ′ is the rule template matching score after integrating historical experience; The matching score M based on the integrated historical experience i Sort all the rule templates, select the rule template with the highest score as the preprocessing rule template for the current data source, and perform the corresponding preprocessing operations on the data uploaded by this data source.

4. The integrated sharing method of elderly health data according to claim 1, characterized in that, According to the unified data classification standard and semantic mapping rules, perform structure standardization, field alignment, and semantic unification on the preprocessed data to generate a unified format elderly health dataset, including: Preset a unified data classification standard, and map the data items with the same semantics but different field names in heterogeneous data sources to unified field identifiers through the field label mapping module, and the mapping relationship of the field label mapping module is stored in the semantic mapping rule library; Uniformly transform the data structures from different sources, including flattening the data dimensions, disassembling multi-level nested fields, unifying the timestamp format, and performing unified unit conversion to generate a data structure that conforms to the platform standard; Perform field alignment processing on the data after structure standardization to ensure that records from different sources are logically within the same field sequence structure, and complete the data value range mapping based on the standard value range set; Introduce a semantic conflict detection algorithm to identify data items with semantic conflicts from different data sources, trigger conflict correction strategies, and complete automatic screening according to the trust weight and time priority rules to produce a unified format elderly health dataset.

5. The integrated sharing method of elderly health data according to claim 1, characterized in that, Based on medical data standards, industry privacy rules, and a preset recognition model, perform sensitivity analysis on the fields in the elderly health dataset and output sensitive field vector identifiers, including: Load medical data standards and industry privacy rules, construct a sensitive field basic dictionary for field recognition, which is used to support the extraction of sensitive fields, and identify and mark potential sensitive fields in the unified format elderly health dataset; Extract the field name, data type, historical access frequency, context module attribution information of the field in the data structure, and semantic similarity with known sensitive fields from the unified format elderly health dataset as the field feature vector; Input the field feature vector into a preset sensitivity recognition model, which includes matching the field name with the basic dictionary of sensitive fields based on a rule engine, calculating the sensitivity score of each field based on word vectors and structure embedding algorithms, and outputting the field sensitivity score result; According to predefined rank mapping rules, map the field sensitivity score to a sensitivity level, generate a sensitive field level table, and encode the sensitive field level table into a fixed-length vector in the order of the original data fields, and output the sensitive field vector identifier.

6. The integrated sharing method of elderly health data according to claim 1, characterized in that In the initial stage of system deployment, construct a data desensitization policy library in combination with medical data standards and industry privacy rules. According to the sensitive field vector identifier, field source, and context environment in the elderly health dataset, perform a sensitivity level score on each sensitive field vector to generate a sensitivity level vector, and based on this sensitivity level vector, call the matching policy template in the data desensitization policy library to perform desensitization operations, and output the desensitized elderly health dataset, including: In the initial stage of system deployment, construct a data desensitization policy library based on general medical data standards and industry privacy rules. The data desensitization policy library is managed by sensitivity level grading, and each sensitivity level corresponds to a desensitization policy template. Each desensitization policy template clearly stipulates the desensitization processing method and specific applicable conditions for the corresponding sensitivity level; Collect the context environment information of the sensitive field vector in the elderly health dataset, including field source, usage scenario, data call frequency and access behavior trajectory, field historical desensitization record and feedback result; Based on a preset sensitivity level scoring function, form a sensitivity level vector S in the order of fields; According to the field level value in the sensitivity level vector S, match the corresponding policy template in the data desensitization policy library. The policy template includes de-identification for highly sensitive fields, masking for medium-sensitive fields, generalization for low-sensitive fields, and pseudo-value replacement for special scenario fields; Perform the data desensitization operation corresponding to the policy template on the sensitive field, generate a data sub-view set containing processing results of different sensitivity levels, and bind specific access authorization rules to each data sub-view to adapt to the sharing needs of users with different permission roles.

7. The integrated sharing method for elderly health data according to claim 6, characterized in that, The preset sensitivity level scoring function includes: Obtain the sensitive fields in the elderly health dataset, and construct a sensitive level scoring function formula based on the inherent sensitivity, usage scenario risk level, environmental exposure risk, and legal supervision intensity of these fields to score each sensitive field and generate a sensitivity level vector S; The formula for the sensitivity level scoring function is as follows: ; Among them, S k represents the sensitivity level score of the k-th field; S k1 represents the inherent sensitivity score of the field; S k2 represents the usage scenario risk level score of the field; S k3 represents the environmental exposure risk score of the field; S k4 represents the legal supervision intensity score of the field; a i is the weight coefficient of the scoring factor, satisfying: , a i ∈[0,1]. Among them, a1 is the weight coefficient of inherent sensitivity, a2 is the usage scenario risk, a3 is the environmental exposure risk, and a4 is the weight coefficient of legal supervision intensity. The value of each weight coefficient is preset according to the actual situation.

8. The integrated sharing method of elderly health data according to claim 1, characterized in that The data token includes: A view identifier field used to uniquely identify the authorized view number; A policy summary field used to identify the access control policy applicable to this data view. The policy summary field is generated based on the access authorization level obtained by the access control function f(U, C, S); A key index field for verifying the encryption and decryption of sensitive fields by the access subject based on the attribute vector U of the requestor; A timeliness identification field that limits the access time range of this token; An encrypted check field for integrity protection of the above token field to prevent man-in-the-middle attacks and illegal tampering; The access control function includes: The access control function f(U, C, S) obtains a comprehensive score Score by weighted summation of three types of factors. The three types of factors include the user attribute vector U = {u1, u2, u3, u4, u5}, where u1 is the identity category of the requesting user, u2 is the affiliated institution, u3 is the historical access behavior, u4 is the permission level, u5 is the data processing qualification, the context environment vector C = {c1, c2, c3, c4, c5, c6}, where c1 is the access time, c2 is the device type, c3 is the network environment, c4 is the access frequency, c5 is the geographical location, c6 is the session behavior pattern; the sensitivity level vector S = {s1, s2, …, s n}, and the sensitivity level is expressed in numerical form and is divided into high-sensitivity fields, medium-sensitivity fields, low-sensitivity fields, and non-sensitivity fields according to the numerical value; The formula for the comprehensive score Score is: ； Among them, , , are the weight coefficients of the user attribute vector, the weight coefficient of the context environment vector, and the weight coefficient of the sensitivity level vector respectively, satisfying = 1, = 1, = 1 for the normalization condition. The user attribute information vector is the i-th attribute score in the user attribute vector , and the value of has been normalized to the interval [0, 1], and the higher the value, the more credible the attribute; is the j-th environmental factor score in the context environment vector , and is the k-th sensitive field level score in the sensitivity level vector . According to the value range of the score Score, determine the access authorization level L. The specific rules are: When Score ≥ T full the access level L = 3, indicating full access, and the system will return the elderly health data set after the desensitization operation; When T partial ≤ Score < T full When this occurs, the access level L = 2, indicating partial field access. The system will perform attribute-based encryption (ABE) processing on the highly sensitive field content of the desensitized elderly health data set, and return the desensitized elderly health data set for medium and low sensitive fields and non-sensitive fields; When T mask ≤Score < T partial the access level L = 1, indicating de-sensitized access, where the system hides highly sensitive fields, masks moderately sensitive fields, and generalizes low-sensitivity fields; When Score < T mask the access level L = 0, indicating a denied access, and the system will only return a null structure; Among them, T full , T partial , T mask are preset thresholds, satisfying T full > T partial > T mask , T full has a default value of 0.75, T partial has a default value of 0.50, T mask has a default value of 0.

30.

9. An integrated sharing platform for elderly health data, characterized in that, For implementing the steps of the integrated sharing method of an elderly health data according to any one of claims 1 to 8, the system includes: A data acquisition module, which is used to acquire multi-dimensional health data of the elderly from multiple heterogeneous data sources. The heterogeneous data sources include home health monitoring devices, smart wearable devices, and chain elderly care service platforms. The multi-dimensional health data includes physiological monitoring data, clinical medical data, behavior and environment data, and psychological and social adaptability data of the elderly; A data access and preprocessing module, which is used to access the multiple heterogeneous data sources to the information interaction center through a network protocol adaptation component, perform processing ability evaluation and data feature extraction on the accessed data sources, generate corresponding data source ability vectors and data feature vectors, and based on the vectors, match and dynamically call corresponding preprocessing rules from the historical hierarchical preprocessing rule library to perform data preprocessing operations; A health data set generation module, which is used to perform structural standardization, field alignment, and semantic unification on the preprocessed data according to unified data classification standards and semantic mapping rules to generate a unified format of the elderly health data set; A sensitive field vector output module, which is used to perform sensitivity analysis on the fields in the elderly health data set based on medical data standards, industry privacy rules, and a preset recognition model and output sensitive field vector identifiers; A desensitization processing module, which is used to construct a data desensitization policy library in combination with medical data standards and industry privacy rules at the initial stage of system deployment. According to the sensitive field vector identifier, field source, and context environment in the elderly health data set, perform a sensitive level score on each sensitive field vector to generate a sensitive level vector, and according to the sensitive level vector, call the matching policy template in the data desensitization policy library to perform desensitization operations and output the desensitized elderly health data set; An access authorization level obtaining module, which is used to input the desensitized elderly health data set, in combination with the user attributes, access context, and sensitive field level of the access requestor, into a pre-constructed access control function to obtain the access authorization level, and perform differential encryption processing on the fields allowed to be accessed by the authorization level; Authorization view encapsulation and data token generation module, which is used to screen the encrypted elderly health data set, extract the fields and their corresponding data contents that the current requester has access rights to, generate a data token with an access policy identifier, and transmit it to the access requester through a communication channel. After receiving the data token, the requester decrypts the fields and parses the contents according to the access policy identifier contained in the data token, and presents the corresponding authorized view data.

Images