Medical care and nursing combined data intercommunication and service collaboration system oriented to community old-age nursing

Abstract

The invention relates to the field of community old-age care, and discloses a community old-age care-oriented medical-care-and-care-combined data intercommunication and service coordination system, which comprises a sensing module used for acquiring physiological data of old people, old-age care service data, environment data and service demand information to form multi-modal information; the network module is used for carrying out multi-modal information transmission, protocol conversion and edge calculation preprocessing to obtain processed data; and the data module is used for constructing a standardized data lake according to the processed data by adopting a unified data standard and a main data management technology, and carrying out fusion storage, treatment and safety protection to obtain multi-source information. Multi-modal information is collected to provide a data basis for cross-mechanism cooperation, data transmission and preprocessing are performed, and a standardized data lake is constructed through a unified data standard and main data management, so that medical institutions, pension institutions and community grids can share global data such as health states, service records and the like of old people in real time.

Description

Technical Field

[0001] The present invention relates to the field of community elderly care technology, and specifically to a medical and elderly care integrated data intercommunication and service collaboration system for community elderly care. Background Art

[0002] With the acceleration of the aging process, community-based elderly care has become the core model for meeting the demand for home-based elderly care services. As a key path to improving the quality of elderly care services, the integration of medical care and elderly care urgently needs to build a technical system that deeply integrates medical resources and elderly care services.

[0003] Traditional community-based integrated elderly care and medical care systems mainly use a manually coordinated resource scheduling mechanism. When faced with complex service scenarios, management personnel need to communicate by phone or hold offline meetings to coordinate the connection and scheduling of medical and elderly care services. For example, when a nursing home discovers an abnormal heart rate in an elderly person, it is necessary to manually contact the community hospital to confirm the availability of emergency resources, and then coordinate ambulance dispatch and subsequent care arrangements, resulting in inefficient cross-institutional service collaboration. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present invention provides a medical and nursing integrated data interoperability and service collaboration system for community-based elderly care, which solves the problem of low efficiency of cross-institutional service collaboration in traditional community-based elderly care medical and nursing integrated systems.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: a medical and elderly care integrated data intercommunication and service collaboration system for community elderly care, including: Perception module: used to collect elderly people's physiological data, elderly care service data, environmental data and service demand information to form multimodal information; Network module: used for multimodal information transmission, protocol conversion and edge computing preprocessing to obtain processed data; Data module: Used to adopt unified data standards and master data management technology, build a standardized data lake based on processed data, perform integrated storage, governance, and security protection, and obtain multi-source information; Model module: used to generate service decision instructions based on multi-source information through cross-modal data fusion, intelligent decision-making algorithms and multi-party differential game models; Application module: Used to convert service decision instructions into service functions including intelligent dispatching, risk warning, and cross-institutional collaborative work orders through the business logic engine, and drive collaborative operations among medical institutions, elderly care institutions, and community grids; Interaction module: used for two-way transmission of human-computer information through an aging-friendly interaction interface.

[0006] By adopting the above technical solution, the perception module collects multimodal information to provide a data foundation for cross-institutional collaboration, the network module performs data transmission and preprocessing, and the data module builds a standardized data lake through unified data standards and master data management, enabling medical institutions, nursing homes, and community grids to share global data such as the health status and service records of the elderly in real time. The model module extracts features through cross-modal data fusion based on multi-source information, optimizes dispatch routes using a multi-resource vehicle routing model with a time window, and quantifies the service utility and cost of medical institutions, nursing homes, and communities through a multi-party differential game model. The Nash equilibrium strategy is solved and collaborative decision instructions that take into account the interests of all parties are generated. The application module converts the decision instructions into cross-institutional collaborative work orders, driving multiple parties to work according to preset processes. At the same time, feedback data from collaborative operations is collected and fed back to the model module to iteratively optimize the dispatch routes and game strategies, achieving dynamic scheduling of medical and nursing service resources and balancing the interests of multiple parties. This solves the problem of low efficiency of cross-institutional service collaboration in traditional community-based integrated medical and nursing systems, and improves service response speed and resource utilization efficiency.

[0007] Preferably, the perception module includes a physiological data acquisition unit, an elderly care service data acquisition unit, an environmental data acquisition unit, a demand information acquisition unit and a multimodal information integration unit. The physiological data acquisition unit is used to collect physiological data through wearable devices and bedside monitoring devices, and output original physiological data. The physiological data includes heart rate, blood pressure, sleep quality and gait stability. The elderly care service data acquisition unit is used to record the type, duration and effect of elderly care services through service terminal equipment, and output original elderly care service data. The elderly care services include bathing assistance and rehabilitation training. The environmental data acquisition unit is used to collect temperature and humidity, living environment safety level and aging-friendly renovation item status through environmental sensors, and output original environmental data. The environmental sensors include temperature and humidity sensors, door magnetic sensors and smoke alarms. The demand information acquisition unit is used to obtain service demand instructions and SOS requests through voice interaction devices and emergency call devices, and output original demand data. The multimodal information integration unit is used to align timestamps and associate device identifiers with the original physiological data, original elderly care service data, original environmental data and original demand data to form multimodal information.

[0008] Preferably, the network module includes a communication protocol conversion unit, an edge computing preprocessing unit and a network transmission control unit. The communication protocol conversion unit is used to convert the DICOM and private protocol data in the multimodal information into a unified JSON format and output the protocol conversion data. The edge computing preprocessing unit is used to perform Gaussian filtering denoising, image compression and fall risk threshold detection on the protocol conversion data and output the preprocessed data. The network transmission control unit is used to transmit the preprocessed data through the 5G medical dedicated network and the LoRaWAN Internet of Things, synchronously record the data transmission timestamp and the device geographic location, and output the processed data.

[0009] Preferably, the data module includes a data standard unification unit, a master data management unit, a data lake storage unit and a data security protection unit. The data standard unification unit is used to execute the HL7 FHIR medical data standard and the extended ISO 20752 pension data standard mapping on the processed data and output standardized data. The master data management unit is used to establish the association relationship between the elderly files, institutional information and personnel skill labels, and generate the master data index. The data lake storage unit is used to perform spatiotemporal index encoding and distributed storage on the standardized data and master data index based on the Delta Lake architecture to form multi-source information. The data security protection unit is used to perform national secret SM4 encryption and RBAC-based access control on the multi-source information, and synchronously store the data operation log on the chain.

[0010] Preferably, the model module includes a cross-modal feature extraction unit, a transmission alignment unit, a graph attention fusion unit, an intelligent decision-making calculation unit and a collaborative optimization processing unit. The cross-modal feature extraction unit is used to extract the 2048-dimensional feature vector of medical images in multi-source information through a convolutional neural network, and extract the 128-dimensional feature vector of the elderly time series data in multi-source information through a long short-term memory network, and output a single-modal feature. The transmission alignment unit is used to solve the transmission matrix based on the single-modal feature through the Wasserstein distance and output the alignment feature matrix. The graph attention fusion unit is used to map the alignment feature matrix to 256 through the graph attention network. dimensional fusion feature vector, constructs a knowledge graph containing the entity relationships between the elderly and medical institutions, nursing homes, and community grid resources, and performs cross-modal data fusion. The intelligent decision-making calculation unit is used to solve the dispatch path based on the fusion feature vector using an intelligent decision-making algorithm through a multi-resource vehicle path model with a time window, and construct a three-dimensional state space to output the risk prediction result. The collaborative optimization processing unit is used to quantify the time constraint of the dispatch path into a service utility item based on the resource status of the dispatch path and medical institutions, nursing homes and community grids, and quantify the resource status into a resource cost item to construct a multi-party differential game model, solve the collaborative strategy, and output the service decision instruction.

[0011] Preferably, the transmission matrix ,satisfy ,in, is the Euclidean distance matrix between unimodal features, and They are the marginal distributions of medical imaging features and pension time series features respectively; The objective function of the intelligent decision-making algorithm is: ,in, It is a collection of medical institutions, nursing homes, community grids and the elderly. A collection of resource types. For resource types From the service point arrive The driving cost, is the timeout penalty item, For early arrival penalty, For service points The dispatch path is obtained by solving the objective function in the time window.

[0012] Preferably, the three-dimensional state space includes physiological indicators , including blood pressure, blood oxygen and heart rate variability, functional status , representing self-care, semi-disability, and disability, respectively, and environmental variables Including temperature, humidity and living environment safety score, the coupled evolution process of the three-dimensional state space is modeled by stochastic differential equations. The stochastic differential equation is: ,in, is the drift term describing the health trend, is the diffusion term describing the random disturbance, The system is a Wiener process that obtains the dynamic evolution trajectory of the three-dimensional state space by numerically solving the stochastic differential equation. Based on the transfer law of the functional state in the trajectory, the disability risk prediction results are generated and integrated into the service decision instructions. The payoff function of the multi-party differential game model is ,in, , For service utility, For service cost, is the discount rate, and the collaborative strategy is determined by solving the Nash equilibrium condition to satisfy the strategy of any subject i. ,in, To balance the strategy combination, the collaborative strategy is decomposed into task execution rules for medical institutions, nursing homes and community grids, and mapped into structured service decision instructions.

[0013] Preferably, the application module includes a business logic parsing unit, an intelligent dispatching execution unit, a risk warning issuing unit and a cross-institutional collaborative work order unit. The business logic parsing unit is used to parse the service decision instructions into dispatching parameters, risk levels and collaborative processes, and output structured business instructions. The intelligent dispatching execution unit is used to generate a resource execution path containing time nodes and service content based on the dispatching parameters. The risk warning issuing unit is used to send warning signals through voice interaction devices and family-end mini-programs according to the risk level, and synchronously trigger the emergency plan. The emergency plan includes emergency resource scheduling and family notification processes. The cross-institutional collaborative work order unit is used to generate a collaborative work order containing multi-party tasks through a workflow engine, drive medical institutions, nursing homes and community grids to operate according to preset processes, output collaborative work feedback data, and feed it back to the multi-resource vehicle path model with time window and the multi-party differential game model for parameter iterative optimization and strategy update.

[0014] Preferably, the interaction module includes an aging-friendly interface generation unit, a voice interaction processing unit and a multimodal feedback unit. The aging-friendly interface generation unit is used to generate a visual interface with large fonts and high contrast, and perform triple interaction of voice, touch screen and physical buttons. The voice interaction processing unit is used to perform dialect recognition through the DeepSpeech 2 model. The multimodal feedback unit is used to output prompt information through the voice synthesis module, dynamically adjust the tone and rhythm according to the risk prediction results, and synchronously feedback user operation behavior data to the perception module.

[0015] The method for data intercommunication and service collaboration for integrated medical and nursing care in community-based elderly care is applied to the aforementioned data intercommunication and service collaboration system for integrated medical and nursing care in community-based elderly care, and includes the following steps: Perception: Collecting elderly people’s physiological data, elderly care service data, environmental data, and service demand information to form multimodal information; Network: Performs multimodal information transmission, protocol conversion, and edge computing preprocessing to obtain processed data; Data: Adopt unified data standards and master data management technologies, build a standardized data lake based on processed data, perform integrated storage, governance, and security protection, and obtain multi-source information; Model: Generates service decision instructions based on multi-source information through cross-modal data fusion, intelligent decision-making algorithms, and multi-party differential game models; Application: The business logic engine transforms service decision instructions into service functions including intelligent dispatching, risk warning, and cross-institutional collaborative work orders, driving collaborative operations among medical institutions, elderly care institutions, and community grids. Interaction: Bidirectional transmission of human-computer information through an aging-friendly interactive interface.

[0016] The present invention provides a data communication and service collaboration system for integrated medical and elderly care services in community-based elderly care. It has the following beneficial effects: 1. The present invention provides a data basis for cross-institutional collaboration by collecting multimodal information, performs data transmission and preprocessing, and builds a standardized data lake through unified data standards and master data management, so that medical institutions, nursing homes and community grids can share global data such as the health status and service records of the elderly in real time. Based on multi-source information, features are extracted through cross-modal data fusion, and a multi-resource vehicle path model with a time window is used to optimize the dispatch path. The service utility and cost of medical institutions, nursing homes and communities are quantified through a multi-party differential game model to generate collaborative decision-making instructions that take into account the interests of all parties. The application module converts the decision-making instructions into cross-institutional collaborative work orders, drives multiple parties to work according to preset processes, and collects feedback data from collaborative operations to feed back the model module to iteratively optimize the dispatch path and game strategy, thereby realizing dynamic scheduling of medical and nursing service resources and balancing the interests of multiple subjects, thereby improving service response speed and resource utilization efficiency.

[0017] 2. The present invention collects multimodal information covering physiological, service, environmental and demand dimensions, performs heterogeneous protocol conversion and edge computing preprocessing, builds a standardized data lake based on unified data standards and performs master data management, so that medical institutions, nursing homes and community grids can share global data such as the health status of the elderly, service records and environmental safety in real time, providing a complete and standardized data foundation for cross-institutional collaborative decision-making, and improving the situation where there is insufficient basis for service connection due to data fragmentation.

[0018] 3. The present invention maps heterogeneous data such as medical images and elderly care time series to a unified feature space, uses Wasserstein distance to solve modal distribution differences, and combines a multi-resource scheduling model with a time window with a three-dimensional state space constructed by stochastic differential equations to achieve service path optimization and dynamic prediction of health risks. At the same time, the multi-party differential game model quantifies the conflicts of interest among multiple institutions and solves the Nash equilibrium strategy, so that service decisions are upgraded from relying on manual experience to data-driven intelligent calculations, effectively responding to resource scheduling and risk warning needs in complex scenarios, avoiding the lag and one-sidedness of traditional decision-making models, and improving the accuracy and foresight of medical and nursing service decisions through cross-modal data fusion and intelligent decision-making algorithms.

[0019] 4. The present invention parses the service decision instructions generated by the model into cross-institutional collaborative work orders, uses the workflow engine to clarify the task responsibilities and execution sequence of each party, and drives medical, elderly care, and community resources to collaborate efficiently according to preset processes. At the same time, the collaborative work feedback data is reversely input into the model module for iterative optimization of multi-resource scheduling model parameters and differential game strategies, forming a dynamic closed loop, reducing resource conflicts and process faults between institutions, and enabling service responses and resource allocation to be dynamically adjusted according to real-time data, thereby improving the coordination and flexibility of integrated medical and elderly care services. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is a system architecture diagram of the medical and nursing integrated data intercommunication and service collaboration system for community-based elderly care proposed by the present invention; Figure 2 This is a method flow chart of the medical and nursing integrated data intercommunication and service collaboration method for community-based elderly care proposed by the present invention. DETAILED DESCRIPTION

[0021] The following will clearly and completely describe the technical solution of the present invention in conjunction with the accompanying drawings. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0022] Example Hundreds of elderly people live in a certain urban community, most of whom suffer from chronic diseases and some have mobility problems. Previously, the community's medical services and elderly care operated independently. The elderly's medical data, such as physical examination reports and medication records, could not be communicated with the care logs of the elderly care institutions, resulting in untimely care due to information lag. To improve this situation, the community introduced the present invention's medical and elderly care data intercommunication and service collaboration system for community elderly care: Please see the attached Figure 1 -Attached Figure 2 The embodiment of the present invention provides a medical and elderly care integrated data intercommunication and service collaboration system for community elderly care, including: Perception module: used to collect elderly people's physiological data, elderly care service data, environmental data and service demand information to form multimodal information; Furthermore, the perception module includes a physiological data acquisition unit, an elderly care service data acquisition unit, an environmental data acquisition unit, a demand information acquisition unit and a multimodal information integration unit. The physiological data acquisition unit is used to collect physiological data through wearable devices and bedside monitoring devices, and output original physiological data. The physiological data includes heart rate, blood pressure, sleep quality and gait stability. The elderly care service data acquisition unit is used to record the type, duration and effect of elderly care services through service terminal equipment, and output original elderly care service data. Elderly care services include bathing assistance and rehabilitation training. The environmental data acquisition unit is used to collect temperature and humidity, living environment safety level and aging-friendly renovation item status through environmental sensors, and output original environmental data. Environmental sensors include temperature and humidity sensors, door magnetic sensors and smoke alarms. The demand information acquisition unit is used to obtain service demand instructions and SOS requests through voice interaction devices and emergency call devices, and output original demand data. The multimodal information integration unit is used to align timestamps and associate device identifiers with the original physiological data, original elderly care service data, original environmental data and original demand data to form multimodal information.

[0023] Specifically, in this embodiment, the perception module is used to collect and integrate multi-dimensional data to form multimodal information to support subsequent processing.

[0024] Specifically, the physiological data acquisition unit uses wearable devices and bedside monitoring equipment to achieve real-time acquisition of physiological indicators. Typically, wearable devices are smart bracelets or patches that use photoplethysmography technology to collect data such as heart rate and blood oxygen levels. Bedside monitoring equipment includes smart mattresses, which monitor sleep quality through a matrix of pressure sensors, and millimeter-wave radars, which analyze gait stability using the Doppler effect. The raw physiological data output includes a timestamp and device identification information.

[0025] The elderly care service data collection unit uses service terminal devices to record information about the elderly care service process. Optionally, the service terminal device can be a bathing robot or a rehabilitation training terminal. For example, during bathing services, the device automatically records the start and end time and service type. Sensors also collect data on the elderly person's status during the service process, generating raw elderly care service data for subsequent service effectiveness evaluation and process optimization.

[0026] The environmental data collection unit monitors the living environment and safety status through various environmental sensors. In some embodiments, these environmental sensors include temperature and humidity sensors, door magnetic sensors, and smoke alarms. The temperature and humidity sensors are deployed in the elderly person's living space to collect real-time temperature and humidity data; the door magnetic sensors are installed on the door to monitor entry and exit status to assess the elderly person's activity patterns; and the smoke alarms provide early warning of fires. The raw environmental data output also includes information on the safety level of the living environment and the status of aging-friendly renovations.

[0027] The demand information collection unit acquires service demands and emergency requests through a voice interaction device and an emergency call device. In one possible implementation, the voice interaction device is a smart interactive screen with a camera, which uses voice recognition technology to convert the elderly person's service demand instructions into text data. The emergency call device is a waterproof button deployed in accident-prone areas such as bathrooms. When pressed, it immediately generates an SOS request signal. The output raw demand data includes the instruction content and trigger location information.

[0028] The multimodal information integration unit fuses and processes the data output by each acquisition unit. First, the timestamps of the original physiological data, original elderly care service data, original environmental data, and original demand data are aligned, ensuring consistency in the time base of each data through a global clock synchronization mechanism. Then, data from different sources is linked based on device identification. For example, the physiological data from a smart bracelet is linked to the temperature and humidity data of the living environment through the elderly person's unique ID. Ultimately, multimodal information containing multi-dimensional features is formed, providing standardized input for data transmission in subsequent network modules and analysis in model modules. Generally speaking, the integrated multimodal information can reduce data lag and improve the accuracy of subsequent service decisions.

[0029] The perception module deploys millimeter-wave radars, environmental sensors, and other devices in senior housing, as well as rehabilitation training equipment and other terminals in community day care centers. This enables multi-dimensional data collection on physiological indicators, the elderly care service process, residential safety, and service needs. The raw data collected by these devices is timestamped and associated with device identifiers to form multimodal information. This provides the data foundation for the community to resolve the disconnect between elderly medical data and care logs. This integrated information enables the community to fully understand the elderly's condition, improving the situation where data fragmentation previously led to delayed care.

[0030] Network module: used for multimodal information transmission, protocol conversion and edge computing preprocessing to obtain processed data; Furthermore, the network module includes a communication protocol conversion unit, an edge computing preprocessing unit and a network transmission control unit. The communication protocol conversion unit is used to convert the DICOM and private protocol data in the multimodal information into a unified JSON format, and output the protocol conversion data. The edge computing preprocessing unit is used to perform Gaussian filtering denoising, image compression and fall risk threshold detection on the protocol conversion data, and output the preprocessed data. The network transmission control unit is used to transmit the preprocessed data through the 5G medical private network and the LoRaWAN Internet of Things, synchronously record the data transmission timestamp and device geographic location, and output the processed data.

[0031] Specifically, in this embodiment, the network module is used to implement transmission processing and communication control of multimodal information to ensure reliable interaction of data between the perception layer and the back-end system.

[0032] Specifically, the communication protocol conversion unit is responsible for the format unification of heterogeneous data. Generally, multimodal information includes DICOM format image data generated by medical equipment and private protocol data of various sensors. These data need to be converted into a unified JSON format for subsequent processing. In one possible implementation method, the tag information of the DICOM file is parsed into JSON key-value pairs through a preset protocol mapping table, and the frame structure of the private protocol data is parsed. For example, the binary data stream of a certain model of millimeter-wave radar is converted into a JSON object containing timestamps and detection parameters according to the protocol specification. The output protocol conversion data has a standardized data structure.

[0033] The edge computing preprocessing unit performs optimization processing on the protocol conversion data. As an option, the physiological data is first denoised using a Gaussian filtering algorithm, and random noise interference is eliminated by setting appropriate Gaussian kernel parameters; for medical imaging data, lossy compression is performed using the JPEG compression algorithm to reduce the amount of data while ensuring the clarity required for diagnosis. In some embodiments, a fall risk threshold detection function is also integrated. By analyzing the gait data of the millimeter wave radar, characteristic parameters such as stride length and center of gravity offset are calculated. When the characteristic value exceeds the preset threshold, a fall warning mark is generated. The output preprocessed data includes denoising, compression and warning information.

[0034] The network transmission control unit is responsible for the reliable transmission of data and metadata recording. Specifically, data with high real-time requirements is transmitted through the 5G medical dedicated network, and its low latency characteristics are used to ensure that urgent data is transmitted first. For non-real-time data such as environmental monitoring, it is transmitted through the LoRaWAN Internet of Things to reduce power consumption. During the data transmission process, the data transmission timestamp and device geographic location information are synchronously recorded. This information is encapsulated as transmission metadata and output as processed data together with the pre-processed data. Under normal circumstances, there is also a network disconnection cache mechanism. When the network is interrupted, the data is temporarily stored in the storage medium of the edge node. After the network is restored, it is reissued in chronological order to ensure data integrity.

[0035] The network module deploys an edge computing gateway in the homes of the elderly in the community, builds a transmission link through the 5G medical dedicated network and the LoRaWAN Internet of Things, converts the multimodal information collected by the perception module into a unified JSON format and performs pre-processing such as denoising and compression, realizing standardized transmission and edge computing of heterogeneous data within the community, reducing the amount of data transmission and the pressure on cloud computing power, ensuring that the health data and environmental data of the elderly can be transmitted to the community data center in real time and reliably, laying the transmission foundation for subsequent cross-institutional data sharing.

[0036] Data module: Used to adopt unified data standards and master data management technology, build a standardized data lake based on processed data, perform integrated storage, governance, and security protection, and obtain multi-source information; Furthermore, the data module includes a data standard unification unit, a master data management unit, a data lake storage unit and a data security protection unit. The data standard unification unit is used to execute the HL7 FHIR medical data standard and the extended ISO 20752 pension data standard mapping on the processed data and output standardized data. The master data management unit is used to establish the association relationship between the elderly files, institutional information and personnel skill labels, and generate the master data index. The data lake storage unit is used to perform spatiotemporal index encoding and distributed storage of standardized data and master data indexes based on the Delta Lake architecture to form multi-source information. The data security protection unit is used to perform national secret SM4 encryption and RBAC-based access control on multi-source information, and synchronously store data operation logs on the chain.

[0037] Specifically, in this embodiment, the data module is used to achieve standardized governance, integrated storage and security protection of medical and nursing data, providing reliable data support for subsequent model analysis.

[0038] Specifically, the data standard unification unit is responsible for the standardized mapping of heterogeneous data. In general, the processed data includes multi-source data such as medical images, physiological indicators, and care records output by the network module, which are converted into a unified standard format through preset mapping rules. In one possible implementation method, for medical data, the basic patient information in the electronic medical record is mapped to the Patient resource of FHIR according to the HL7 FHIR standard, and the diagnostic information is mapped to the Condition resource; for elderly care data, the bathing service record is mapped to the CareRecord entity according to the extended ISO 20752 standard, and the rehabilitation training data is mapped to the RehabilitationActivity entity. The output standardized data has a unified data model and field definition.

[0039] The master data management unit is used to construct core data entities and relationships in the medical and nursing field. As an option, with the unique identification of the elderly as the core, the relationship between the elderly archives, medical institution information, and service personnel skill tags is established. For example, the electronic medical records, health monitoring data, and care logs are associated through the elderly ID, and the doctor qualifications of the community health service center and the skills of the nursing staff in the nursing home are structured and stored. In some embodiments, graph database technology is used to construct a knowledge graph to generate a master data index containing the relationship between entities, supporting efficient cross-domain data queries, such as quickly retrieving the elderly’s medical records, care history, and adapted service resources through the elderly ID.

[0040] The data lake storage unit implements distributed storage and management of data based on the Delta Lake architecture. Specifically, spatiotemporal index encoding is performed on standardized data and master data indexes. For elderly care service data containing geographic location information, the Z-Order curve is used to divide the spatial dimension. For time-series physiological indicator data, partition storage is performed according to time windows such as daily and weekly to achieve efficient data organization and retrieval. In one possible implementation, the ACID transaction characteristics of Delta Lake are used to ensure the consistency of data updates, support data version management, and trace the status of historical data, forming a multi-source information set containing multi-dimensional and multi-modal information, providing a complete data foundation for model training.

[0041] The data security protection unit implements data security protection throughout the entire life cycle. In general, the national secret SM4 algorithm is used to encrypt multi-source information. During data transmission, the transport layer security protocol TLS is used in combination with the SM4 algorithm to ensure data confidentiality. Sensitive fields such as ID numbers and medical records are encrypted at the storage layer at the field level. As an option, based on the RBAC role-based access control model, fine-grained access rights are assigned to different roles such as the elderly, family members, medical staff, and managers. For example, family members can only view the basic health data of the elderly, and medical staff can access complete medical records. In some embodiments, data operation logs are stored on the chain through blockchain technology such as Hyperledger Fabric to record operations such as data creation, modification, and access, to ensure the traceability of data operations and improve data security and credibility.

[0042] The data module builds a standardized data lake based on the Delta Lake architecture in the community data center, maps the processed data output by the network module according to the HL7 FHIR and extended ISO 20752 standards, and establishes an associated index of elderly files and institutional information through master data management, realizing the integrated storage and security governance of medical diagnosis and treatment data and elderly care data in the community, enabling community hospitals, nursing homes and other institutions to access the entire domain data of the elderly in real time, providing complete data support for cross-institutional collaborative decision-making.

[0043] Model module: used to generate service decision instructions based on multi-source information through cross-modal data fusion, intelligent decision-making algorithms and multi-party differential game models; Furthermore, the model module includes a cross-modal feature extraction unit, a transmission alignment unit, a graph attention fusion unit, an intelligent decision-making calculation unit and a collaborative optimization processing unit. The cross-modal feature extraction unit is used to extract the 2048-dimensional feature vector of medical images in multi-source information through a convolutional neural network, and to extract the 128-dimensional feature vector of elderly care time series data in multi-source information through a long short-term memory network, and output single-modal features. The transmission alignment unit is used to solve the transmission matrix based on the single-modal features through the Wasserstein distance and output the alignment feature matrix. The graph attention fusion unit is used to map the alignment feature matrix to a 256-dimensional matrix through the graph attention network. The fusion feature vector of the data is used to construct a knowledge graph containing the entity relationships between the elderly and medical institutions, nursing homes, and community grid resources, and perform cross-modal data fusion. The intelligent decision-making calculation unit is used to solve the dispatch path based on the fusion feature vector using the intelligent decision-making algorithm through a multi-resource vehicle path model with a time window, and to construct a three-dimensional state space to output the risk prediction result. The collaborative optimization processing unit is used to quantify the time constraint of the dispatch path into a service utility item based on the resource status of medical institutions, nursing homes, and community grids, and to quantify the resource status into a resource cost item to construct a multi-party differential game model, solve the collaborative strategy, and output the service decision instructions.

[0044] Furthermore, the transmission matrix ,satisfy ,in, is the Euclidean distance matrix between unimodal features, and They are the marginal distributions of medical imaging features and pension time series features respectively; The objective function of the intelligent decision-making algorithm is ,in, It is a collection of medical institutions, nursing homes, community grids and the elderly. A collection of resource types. For resource types From the service point arrive The driving cost, is the timeout penalty item, For early arrival penalty, For service points The dispatch path is obtained by solving the objective function in the time window.

[0045] Furthermore, the three-dimensional state space includes physiological indicators , including blood pressure, blood oxygen and heart rate variability, functional status , representing self-care, semi-disability, and disability, respectively, and environmental variables Including temperature, humidity and living environment safety score, the coupled evolution process of the three-dimensional state space is modeled by stochastic differential equations. The stochastic differential equation is: ,in, is the drift term describing the health trend, is the diffusion term describing the random disturbance, The system is a Wiener process that obtains the dynamic evolution trajectory of the three-dimensional state space by numerically solving the stochastic differential equation. Based on the transfer law of the functional state in the trajectory, the disability risk prediction results are generated and integrated into the service decision instructions. The payoff function of the multi-party differential game model is ,in, , For service utility, For service cost, is the discount rate, and the collaborative strategy is determined by solving the Nash equilibrium condition to satisfy the strategy of any subject i. ,in, To balance the strategy combination, the collaborative strategy is decomposed into task execution rules for medical institutions, nursing homes and community grids, and mapped into structured service decision instructions.

[0046] Specifically, in this embodiment, the model module realizes the generation of intelligent decision-making instructions based on multi-source information, and solves the resource scheduling and risk prediction problems in complex scenarios through cross-modal data fusion, state space modeling and game theory optimization.

[0047] Specifically, the cross-modal feature extraction unit performs feature engineering on multi-source information. Typically, a pre-trained ResNet-50 convolutional neural network is used to extract 2048-dimensional feature vectors from medical images, and a global average pooling layer is used to obtain an abstract representation of the image. For elderly care time series data, a long short-term memory (LSTM) network is used to capture dynamic changes in the time series and output a 128-dimensional feature vector. In one possible implementation, the BERT model is used to extract semantic features from text information in multimodal data, ultimately outputting a unimodal feature set encompassing multiple modalities.

[0048] The transmission alignment unit addresses the distribution discrepancies between modalities. Alternatively, an optimal transmission problem is constructed based on Wasserstein distance theory to solve the transmission matrix. Using the Sinkhorn algorithm for iterative optimization, the marginal distributions of medical image features and elderly care time series features are aligned to generate a transmission matrix that satisfies the constraints. In some embodiments, to improve computational efficiency, entropy regularization is employed to smooth the transmission matrix and ensure the stability of the alignment process.

[0049] In this embodiment, the transmission alignment unit receives the unimodal features output by the cross-modal feature extraction unit, including the feature vector set of medical images and the feature vector set of elderly care time series data. Specifically, the Euclidean distance matrix C between the two types of features is first calculated, and the optimal transmission problem is constructed based on the distance matrix to solve the transmission matrix ,make The marginal distribution of medical image characteristics is satisfied , columns and are marginal distributions of pension time series characteristics constraints, while minimizing The inner product of C. Through iterative optimization of the Sinkhorn algorithm, the aligned feature matrix is ​​output, and the medical and elderly care features are mapped to a unified distribution space, providing a basis for subsequent cross-modal data fusion.

[0050] The graph attention fusion unit achieves deep fusion of multimodal data. It constructs a knowledge graph encompassing elderly individuals, institutions, and resource entities. The graph attention network (GAT) performs a nonlinear transformation on the aligned feature matrix. Through multiple layers of network iteration, the aligned features are mapped into a fused feature vector that simultaneously captures entity attributes and interactions, forming a cross-modal fused knowledge representation.

[0051] The intelligent decision-making calculation unit optimizes service paths and predicts risks based on fusion features. For multi-resource scheduling problems, a multi-resource vehicle path model with time windows is constructed. Service points V include medical institutions, nursing homes, community grids, and the elderly. The resource type set K corresponds to emergency equipment, nursing staff, etc. This unit extracts information related to service costs and time windows from the fusion features to determine the travel cost of resource type k from service point i to j. , and the time window of service point j , corresponding to the diagnosis and treatment time of medical institutions, the care time of nursing homes or the time required by the elderly. Based on the above parameters, a multi-resource vehicle routing model with time windows is constructed. Its objective function is to minimize the sum of resource travel cost and time penalty term, and the time penalty term includes overtime penalty. and early arrival penalties , which are calculated by the deviation between the resource arrival time and the time window. This objective function is solved using a simulated annealing algorithm to output the optimal dispatch path, clarify the service order and time schedule of each resource, and achieve efficient scheduling of multiple resources under time constraints.

[0052] When the intelligent decision-making computing unit receives the multi-source information output by the data module, it extracts physiological indicators , functional status and environment variables As input. Based on stochastic differential equation , describing the evolution of health status, where Determined by fitting physiological trends, functional status and environmental influences, Characterize the influence of external random disturbances, Simulate uncertainty fluctuations. By numerically solving this equation, the system outputs disability risk prediction results, providing a dynamic evolution basis for health status in service decision-making instructions and supporting proactive adjustments to resource scheduling.

[0053] The collaborative optimization processing unit solves the resource allocation problem of multiple stakeholders. Based on the dispatch path with time window constraints, service point association relationships, and resource status such as medical institution resource load, nursing home caregiver availability, and community grid material reserves fed back by the data module, the time constraints of the dispatch path are quantified as service utility items. , such as the timely response of medical institutions to the diagnosis and treatment value of dispatched orders, quantifying resource status into resource cost items For example, when a nursing home allocates caregiver hours, a multi-party differential game model is constructed. By solving Nash equilibrium conditions, the optimal coordination strategy for each entity is determined, ensuring maximum resource utilization while meeting service needs. One possible implementation uses gradient descent to iteratively solve the equilibrium strategy, ensuring both convergence and effectiveness.

[0054] For the multi-party differential game model, the collaborative optimization processing unit obtains the service utility of medical institutions, nursing homes and community grids , service costs and discount rate As input, based on the profit function Construct a game model and determine the optimal strategy combination of each subject by solving the Nash equilibrium conditions The output collaborative strategy clarifies the resource allocation and service division of each party, such as the scheduling rhythm of emergency resources in medical institutions and the dispatching time periods of caregivers in nursing homes, to achieve a balance of interests among multiple entities in service provision and ensure the collaborative execution efficiency of service decision-making instructions.

[0055] Generally speaking, the output results of the model module are issued in the form of service decision instructions, including service paths, risk warning levels, resource allocation plans, etc., thereby improving the intelligence level and collaborative efficiency of integrated medical and nursing services, and achieving increased service response speed, optimized resource utilization efficiency, and improved risk prediction accuracy.

[0056] The model module is based on multi-source information from the community data lake. Through cross-modal feature fusion, a multi-resource scheduling model with a time window, and a multi-party differential game model, it generates service decision instructions that take into account the interests of both medical and nursing institutions. It quantifies the health risk prediction and resource scheduling strategies of the elderly in the community into data-driven intelligent decisions, replacing traditional manual experience judgments. When faced with sudden health problems of the elderly, the community can automatically generate collaborative strategies that include emergency resource scheduling and care connection, thereby improving the accuracy and foresight of decision-making.

[0057] Application module: Used to convert service decision instructions into service functions including intelligent dispatching, risk warning, and cross-institutional collaborative work orders through the business logic engine, and drive collaborative operations among medical institutions, elderly care institutions, and community grids; Furthermore, the application module includes a business logic parsing unit, an intelligent dispatching execution unit, a risk warning issuing unit and a cross-institutional collaborative work order unit. The business logic parsing unit is used to parse service decision instructions into dispatching parameters, risk levels and collaborative processes, and output structured business instructions. The intelligent dispatching execution unit is used to generate a resource execution path containing time nodes and service content based on the dispatching parameters. The risk warning issuing unit is used to send warning signals through voice interaction devices and family-side mini-programs based on risk levels, and simultaneously trigger emergency plans. The emergency plan includes emergency resource scheduling and family notification processes. The cross-institutional collaborative work order unit is used to generate collaborative work orders containing multi-party tasks through a workflow engine, drive medical institutions, nursing homes and community grids to operate according to preset processes, output collaborative work feedback data, and feed it back to the multi-resource vehicle path model with time windows and the multi-party differential game model for parameter iterative optimization and strategy updates.

[0058] Specifically, in this embodiment, the application module realizes the implementation and coordination of service functions through multi-unit collaboration based on the service decision instructions output by the model module. Specifically, the business logic parsing unit receives the service decision instruction, which contains the resource scheduling, risk prediction and collaborative strategy information generated by the model module through cross-modal fusion, intelligent decision-making and game optimization. As an option, according to the preset business rules, the service decision instruction is parsed into dispatch parameters, risk levels and collaborative processes, where the dispatch parameters correspond to the multi-resource scheduling results with time windows output by the intelligent decision-making calculation unit, the risk level is associated with the three-dimensional state space prediction conclusion constructed by the intelligent decision-making calculation unit, and the collaborative process is consistent with the multi-party game strategy of the collaborative optimization processing unit, and structured business instructions are output to provide an execution basis for subsequent units.

[0059] The intelligent dispatch execution unit receives dispatch parameters from structured business instructions, including resource type, service point time constraints, and task priority. In one possible implementation, the unit invokes a path planning algorithm, combined with the real-time status of resources, to generate a resource execution path containing time nodes and service content. Time nodes are determined based on the time constraints of a multi-resource vehicle path model with time windows, and service content is associated with the type of elderly care services and medical treatment needs. Furthermore, the unit supports manual intervention and adjustment. When sudden resource conflicts or changes in the needs of the elderly occur, the resource execution path can be re-optimized to ensure flexible service response.

[0060] The risk warning issuance unit receives the risk level in the structured business instruction, which is derived from the disability risk prediction result generated by the intelligent decision-making calculation unit based on the stochastic differential equation. Based on the severity of the risk level, the warning signal is sent through the voice interaction device and the family-side mini-program. The higher the risk level, the more prominent the warning method. The emergency plan is triggered simultaneously. The emergency plan includes the emergency resource scheduling process and the family notification process. The emergency resource scheduling is linked to the resource execution path of the intelligent dispatch execution unit, prioritizing the allocation of nearby emergency resources. The family notification process reaches the elderly person's family through preset contact information to achieve a rapid response to risk events.

[0061] The cross-institutional collaborative work order unit receives the collaborative process in the structured business instructions, which is built based on the multi-party differential game strategy of the collaborative optimization processing unit. Through the workflow engine, the collaborative process is broken down into specific tasks for medical institutions, nursing homes and community grids, and a collaborative work order containing multi-party tasks is generated. The task content clearly defines the service responsibilities, execution time and connection requirements of each institution. During the task execution process, the unit collects collaborative operation feedback data, including service execution time, task completion status and service effect evaluation, and feeds the feedback data back to the intelligent decision-making calculation unit and collaborative optimization processing unit of the model module. The intelligent decision-making calculation unit iteratively optimizes the parameters of the multi-resource vehicle path model with time window based on the feedback data. The collaborative optimization processing unit updates the strategy of the multi-party differential game model based on the feedback data, realizes dynamic optimization of service decisions, and improves the efficiency and accuracy of medical and nursing collaborative operations.

[0062] The application module parses the decision-making instructions generated by the model module into intelligent dispatching parameters, risk warning levels and cross-institutional collaborative processes. Through the workflow engine, it drives community hospitals, nursing homes and community grids to operate according to preset processes, and collects feedback data to feed back model optimization, realizing closed-loop collaboration of medical and elderly care services in the community, enabling efficient connection of the entire process of services for the elderly from health monitoring to emergency response, and ensuring the consistency and flexibility of service execution.

[0063] Interaction module: used for two-way transmission of human-computer information through an aging-friendly interaction interface.

[0064] Furthermore, the interaction module includes an aging-friendly interface generation unit, a voice interaction processing unit and a multimodal feedback unit. The aging-friendly interface generation unit is used to generate a visual interface with large fonts and high contrast, and perform triple interaction of voice, touch screen and physical buttons. The voice interaction processing unit is used to perform dialect recognition through the DeepSpeech 2 model. The multimodal feedback unit is used to output prompt information through the voice synthesis module, dynamically adjust the tone and rhythm according to the risk prediction results, and synchronously feedback user operation behavior data to the perception module.

[0065] Specifically, in this embodiment, the interactive module connects the data collection of the perception module and the service feedback of the application module, and realizes aging-friendly human-computer interaction through multi-unit collaboration. Specifically, the aging-friendly interface generation unit receives the service status information output by the application module, including intelligent dispatch results, risk warning prompts, etc., and generates a large-font, high-contrast visual interface based on the vision and operation characteristics of the elderly. The interface layout follows the principle of simplicity, highlighting the core service entrance and the warning information display area. As an option, the unit supports three interaction modes: voice, touch screen, and physical buttons. Voice interaction is used to wake up service instructions, touch screen interaction adapts to gesture operations, and physical buttons meet the quick triggering of emergency scenarios, such as the one-button call function, so that the elderly can conveniently initiate services and query status.

[0066] The voice interaction processing unit collects elderly residents' voice commands, which include service requests, status feedback, and more. In one possible implementation, this unit deploys the DeepSpeech 2 model, which uses a pre-trained dialect dataset to optimize model recognition capabilities based on common dialect characteristics of community residents. The unit then converts voice signals into text commands, which are then output to the perception or application modules to accurately capture and respond to service requests.

[0067] The multimodal feedback unit receives the risk prediction results output by the model module and the prompt information of the application module. Specifically, the unit calls the speech synthesis module to dynamically adjust the intonation and rhythm according to the severity of the risk prediction results. The risk level is associated with the three-dimensional state space prediction conclusion of the intelligent decision-making calculation unit. The higher the level, the more rapid the intonation and rhythm to enhance the prompt effect. At the same time, the user's behavioral data in interface operation and voice interaction, such as command trigger frequency, operation path, etc., are recorded and synchronously fed back to the perception module to provide interactive side data supplement for the multimodal information collection of the perception module, realize a two-way closed loop of human-computer information, and improve the aging-friendly experience of service interaction and the integrity of data feedback.

[0068] The interactive module deploys a large-font, high-contrast smart interactive screen in the elderly's home, supporting triple interaction through voice, touch screen, and physical buttons. It optimizes dialect recognition through the DeepSpeech 2 model and dynamically adjusts the voice feedback rhythm according to the risk level to suit the operating habits and physiological characteristics of the elderly in the community, allowing the elderly to easily initiate service requests or obtain health warnings. At the same time, the interactive behavior data is fed back to the perception module, forming a two-way closed loop of human-computer information, thereby improving the reach and user experience of community elderly care services.

[0069] By building a collaborative architecture of perception modules, network modules, data modules, model modules, application modules and interaction modules, we can achieve intelligent management of the entire process of integrated medical and nursing services. The perception module collects multimodal information on the elderly's physiology, services, environment, and needs. The network module performs data transmission, protocol conversion, and edge computing preprocessing. The data module uses unified standards to build a standardized data lake and implements cross-institutional data fusion and storage through master data management technology. The model module extracts features through cross-modal data fusion based on multi-source information. It optimizes service paths using a multi-resource scheduling model with time windows and quantifies the service utility and cost of medical institutions, elderly care institutions, and communities with the help of a multi-party differential game model, generating collaborative decision-making instructions that balance the interests of multiple parties. The application module converts decision-making instructions into intelligent dispatching, risk warnings, and cross-institutional collaborative work orders, driving multiple parties to operate according to preset processes. It also collects feedback data to feed back into iterative model optimization. The interaction module enables two-way information transmission between humans and machines through an aging-friendly interface, ensuring convenient service access. Through full-link data interoperability, quantitative optimization of decision models, and closed-loop management of collaborative processes, it achieves dynamic scheduling of medical and nursing service resources and balances the interests of multiple parties. This solves the problem of inefficient cross-institutional service collaboration in traditional community-based elderly care systems and improves service response speed and resource utilization efficiency.

[0070] The method for data intercommunication and service collaboration for integrated medical and nursing care in community-based elderly care is applied to the aforementioned data intercommunication and service collaboration system for integrated medical and nursing care in community-based elderly care, and includes the following steps: Perception: Collecting elderly people’s physiological data, elderly care service data, environmental data, and service demand information to form multimodal information; Network: Performs multimodal information transmission, protocol conversion, and edge computing preprocessing to obtain processed data; Data: Adopt unified data standards and master data management technologies, build a standardized data lake based on processed data, perform integrated storage, governance, and security protection, and obtain multi-source information; Model: Generates service decision instructions based on multi-source information through cross-modal data fusion, intelligent decision-making algorithms, and multi-party differential game models; Application: The business logic engine transforms service decision instructions into service functions including intelligent dispatching, risk warning, and cross-institutional collaborative work orders, driving collaborative operations among medical institutions, elderly care institutions, and community grids. Interaction: Bidirectional transmission of human-computer information through an aging-friendly interactive interface.

[0071] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A data intercommunication and service collaboration system for integrated medical and elderly care services in the community, characterized by: include: Perception module: used to collect elderly people's physiological data, elderly care service data, environmental data and service demand information to form multimodal information; Network module: used for multimodal information transmission, protocol conversion and edge computing preprocessing to obtain processed data; Data module: Used to adopt unified data standards and master data management technology, build a standardized data lake based on processed data, perform integrated storage, governance, and security protection, and obtain multi-source information; Model module: used to generate service decision instructions based on multi-source information through cross-modal data fusion, intelligent decision-making algorithms and multi-party differential game models; Application module: Used to convert service decision instructions into service functions including intelligent dispatching, risk warning, and cross-institutional collaborative work orders through the business logic engine, and drive collaborative operations among medical institutions, elderly care institutions, and community grids; Interaction module: used for two-way transmission of human-computer information through an aging-friendly interaction interface.

2. The medical and elderly care integrated data intercommunication and service collaboration system for community elderly care according to claim 1 is characterized by: The perception module includes a physiological data acquisition unit, an elderly care service data acquisition unit, an environmental data acquisition unit, a demand information acquisition unit and a multimodal information integration unit. The physiological data acquisition unit is used to collect physiological data through wearable devices and bedside monitoring devices and output original physiological data. The physiological data includes heart rate, blood pressure, sleep quality and gait stability. The elderly care service data acquisition unit is used to record the type, duration and effect of elderly care services through service terminal equipment and output original elderly care service data. The elderly care services include bathing assistance and rehabilitation training. The environmental data acquisition unit is used to collect temperature and humidity, living environment safety level and aging-friendly renovation item status through environmental sensors and output original environmental data. The environmental sensors include temperature and humidity sensors, door magnetic sensors and smoke alarms. The demand information acquisition unit is used to obtain service demand instructions and SOS requests through voice interaction devices and emergency call devices, and output original demand data. The multimodal information integration unit is used to align timestamps and associate device identifiers of the original physiological data, original elderly care service data, original environmental data and original demand data to form multimodal information.

3. The data communication and service collaboration system for integrated medical and elderly care services for community elderly care according to claim 1 is characterized by: The network module includes a communication protocol conversion unit, an edge computing preprocessing unit and a network transmission control unit. The communication protocol conversion unit is used to convert the DICOM and private protocol data in the multimodal information into a unified JSON format and output the protocol conversion data. The edge computing preprocessing unit is used to perform Gaussian filtering denoising, image compression and fall risk threshold detection on the protocol conversion data and output the preprocessed data. The network transmission control unit is used to transmit the preprocessed data through the 5G medical private network and the LoRaWAN Internet of Things, synchronously record the data transmission timestamp and the device geographic location, and output the processed data.

4. The data communication and service collaboration system for integrated medical and elderly care services for community elderly care according to claim 1 is characterized by: The data module includes a data standard unification unit, a master data management unit, a data lake storage unit and a data security protection unit. The data standard unification unit is used to execute the HL7 FHIR medical data standard and the extended ISO20752 pension data standard mapping on the processed data and output standardized data. The master data management unit is used to establish the association relationship between the elderly archives, institutional information and personnel skill labels, and generate the master data index. The data lake storage unit is used to perform spatiotemporal index encoding and distributed storage on the standardized data and master data index based on the DeltaLake architecture to form multi-source information. The data security protection unit is used to perform national secret SM4 encryption and RBAC-based access control on the multi-source information, and synchronously store data operation logs on the chain.

5. The data communication and service collaboration system for integrated medical and elderly care services for community elderly care according to claim 1 is characterized by: The model module includes a cross-modal feature extraction unit, a transmission alignment unit, a graph attention fusion unit, an intelligent decision-making calculation unit and a collaborative optimization processing unit. The cross-modal feature extraction unit is used to extract the 2048-dimensional feature vector of medical images in multi-source information through a convolutional neural network, and extract the 128-dimensional feature vector of elderly care time series data in multi-source information through a long short-term memory network, and output a single-modal feature. The transmission alignment unit is used to solve the transmission matrix based on the single-modal feature through the Wasserstein distance and output the alignment feature matrix. The graph attention fusion unit is used to map the alignment feature matrix to a 256-dimensional matrix through the graph attention network. The feature vectors are integrated to construct a knowledge graph containing the entity relationships between the elderly and medical institutions, nursing homes, and community grid resources, and cross-modal data fusion is performed. The intelligent decision-making calculation unit is used to solve the dispatch path based on the fused feature vectors using an intelligent decision-making algorithm through a multi-resource vehicle path model with a time window, and to construct a three-dimensional state space to output the risk prediction results. The collaborative optimization processing unit is used to quantify the time constraints of the dispatch path into service utility terms and the resource status of medical institutions, nursing homes, and community grids based on the dispatch path and the resource status of medical institutions, nursing homes, and community grids, and to quantify the resource status into resource cost terms to construct a multi-party differential game model, solve the collaborative strategy, and output service decision instructions.

6. The data communication and service collaboration system for integrated medical and elderly care services for community elderly care according to claim 5 is characterized by: The transmission matrix ,satisfy ,in, is the Euclidean distance matrix between unimodal features, and They are the marginal distributions of medical imaging features and pension time series features respectively; The objective function of the intelligent decision-making algorithm is: ,in, It is a collection of medical institutions, nursing homes, community grids and the elderly. A collection of resource types. For resource types From the service point arrive The driving cost, is the timeout penalty item, For early arrival penalty, For service points The dispatch path is obtained by solving the objective function in the time window.

7. The medical and elderly care integrated data intercommunication and service collaboration system for community elderly care according to claim 5 is characterized by: The three-dimensional state space includes physiological indicators , including blood pressure, blood oxygen and heart rate variability, functional status , representing self-care, semi-disability, and disability, respectively, and environmental variables Including temperature, humidity and living environment safety score, the coupled evolution process of the three-dimensional state space is modeled by stochastic differential equations. The stochastic differential equation is: ,in, is the drift term describing the health trend, is the diffusion term describing the random disturbance, The system is a Wiener process that obtains the dynamic evolution trajectory of the three-dimensional state space by numerically solving the stochastic differential equation. Based on the transfer law of the functional state in the trajectory, the disability risk prediction results are generated and integrated into the service decision instructions. The payoff function of the multi-party differential game model is ,in, , For service utility, For service cost, is the discount rate, and the collaborative strategy is determined by solving the Nash equilibrium condition to satisfy the strategy of any subject i. ,in, To balance the strategy combination, the collaborative strategy is decomposed into task execution rules for medical institutions, nursing homes and community grids, and mapped into structured service decision instructions.

8. The medical and elderly care integrated data intercommunication and service collaboration system for community elderly care according to claim 1 is characterized by: The application module includes a business logic parsing unit, an intelligent dispatching execution unit, a risk warning issuing unit and a cross-institutional collaborative work order unit. The business logic parsing unit is used to parse service decision instructions into dispatching parameters, risk levels and collaborative processes, and output structured business instructions. The intelligent dispatching execution unit is used to generate a resource execution path containing time nodes and service content based on the dispatching parameters. The risk warning issuing unit is used to send warning signals through voice interaction devices and family-side mini-programs based on risk levels, and synchronously trigger emergency plans. The emergency plans include emergency resource scheduling and family notification processes. The cross-institutional collaborative work order unit is used to generate collaborative work orders containing multi-party tasks through a workflow engine, drive medical institutions, nursing homes and community grids to operate according to preset processes, output collaborative work feedback data, and feed it back to the multi-resource vehicle path model with time windows and the multi-party differential game model for parameter iterative optimization and strategy update.

9. The medical and elderly care integrated data intercommunication and service collaboration system for community elderly care according to claim 1 is characterized by: The interaction module includes an aging-friendly interface generation unit, a voice interaction processing unit, and a multimodal feedback unit. The aging-friendly interface generation unit is used to generate a visual interface with large fonts and high contrast, and perform triple interaction through voice, touch screen, and physical buttons. The voice interaction processing unit is used to perform dialect recognition through the DeepSpeech 2 model. The multimodal feedback unit is used to output prompt information through the speech synthesis module, dynamically adjust the tone and rhythm according to the risk prediction results, and synchronously feedback user operation behavior data to the perception module.

10. A method for data intercommunication and service collaboration for integrated medical and elderly care services in community-based elderly care, characterized by: The medical and elderly care integrated data intercommunication and service collaboration system for community elderly care as described in any one of claims 1 to 9 comprises the following steps: Perception: Collecting elderly people’s physiological data, elderly care service data, environmental data, and service demand information to form multimodal information; Network: Performs multimodal information transmission, protocol conversion, and edge computing preprocessing to obtain processed data; Data: Adopt unified data standards and master data management technologies, build a standardized data lake based on processed data, perform integrated storage, governance, and security protection, and obtain multi-source information; Model: Generates service decision instructions based on multi-source information through cross-modal data fusion, intelligent decision-making algorithms, and multi-party differential game models; Application: The business logic engine transforms service decision instructions into service functions including intelligent dispatching, risk warning, and cross-institutional collaborative work orders, driving collaborative operations among medical institutions, elderly care institutions, and community grids. Interaction: Bidirectional transmission of human-computer information through an aging-friendly interactive interface.

Images