Smart old-age care service construction method based on Fabric block chain and big data AI

Abstract

The invention discloses an intelligent old-age care service construction method based on a Fabric block chain and big data AI, and the method comprises the following steps: S1, collecting multi-source heterogeneous data of old people, and carrying out the standardization processing of the data; s2, performing classified storage on the standardized data processed in the step S1 on the basis of a Hyperledger Fabric block chain and an IPFS (Internet Protocol File System) distributed network; s3, constructing a multi-modal AI health early warning and decision model architecture, analyzing the standardized data stored in the step S2, and generating a health risk early warning and personalized nursing scheme; s4, cross-institution data authorization sharing is carried out through the block chain smart contract, an automatic process triggering rule of the smart contract is preset, and cross-institution real-time data sharing and business collaboration are carried out; and S5, configuring an aging-suitable interaction terminal, receiving health risk early warning, a personalized nursing scheme and a business cooperation event process in real time, capturing user interaction operation, and returning operation data to the step S3 to perform AI model optimization on the multi-mode AI health early warning and decision model architecture.

Description

Technical Field

[0001] The present invention relates to the field of smart elderly care technology, and in particular to a method for constructing smart elderly care services based on Fabric blockchain and big data AI. Background Art

[0002] Elderly care resources are unevenly distributed across regions, urban and rural areas, and different groups, leading to significant disparities in the quality of elderly care services. Traditional elderly care service models suffer from data silos, high risks of privacy leaks, and insufficient personalized services.

[0003] While existing IoT devices can collect health data, they face challenges such as inconsistent data formats, high risks of privacy leaks, and difficulty in cross-platform collaboration. Furthermore, the application of big data analysis and AI technologies is often limited to single scenarios and lacks systematic integration. While blockchain technology has found some applications in data security, it has yet to fully integrate distributed storage and smart contract capabilities into smart elderly care systems. Therefore, there is an urgent need for a smart elderly care system that integrates Fabric, big data, AI, and the IoT to address core issues such as data security, cross-domain collaboration, and personalized services. Summary of the Invention

[0004] The purpose of this invention is to provide a method for constructing smart elderly care services based on Fabric blockchain and big data AI to solve the above problems.

[0005] To achieve the above object, the present invention adopts the following technical solutions: A method for constructing smart elderly care services based on Fabric blockchain and big data AI, including the following steps: S1. Collect multi-source heterogeneous data of the elderly and standardize the data; S2. Classify and store the standardized data processed in step S1 based on the Hyperledger Fabric blockchain and the IPFS distributed network; S3. Build a multimodal AI health warning and decision-making model architecture, analyze the standardized data stored in step S2, and generate health risk warnings and personalized care plans; S4. Authorize cross-institutional data sharing through blockchain smart contracts, pre-set automated process triggering rules for smart contracts, and conduct real-time cross-institutional data sharing and business collaboration; S5. Configure an aging-friendly interactive terminal to receive real-time health risk warnings, personalized care plans, and business collaboration event processes, capture user interaction operations, and return the operation data to step S3 to optimize the AI model of the multimodal AI health warning and decision-making model architecture, forming a closed-loop iterative mechanism.

[0006] Preferably, the standardization process of the multi-source heterogeneous data is specifically as follows: S11. Deploy data collection ports to obtain multi-source heterogeneous data of the elderly; S12, the acquisition port parses the data, converts the timestamp and unit, and encapsulates it into a standard format; S13. Call the Pandas analysis and processing library to fill missing values; S14. Detect outliers using the Z-score method and replace or directly eliminate outlier records using upper and lower limits; S15. Standardize or normalize the data, remove duplicate data, and generate standardized data.

[0007] Preferably, the standardized data includes structured data and unstructured data. The structured data is directly stored in the Hyperledger Fabric on-chain ledger and permission control is performed through smart contracts. The unstructured data is stored in the IPFS distributed network and the hash value is recorded in the blockchain.

[0008] Preferably, the standardized data is transmitted using an end-to-end encrypted transmission link, and sensitive fields are desensitized.

[0009] Preferably, the multimodal AI health warning and decision-making model architecture includes a time series prediction model, a behavior recognition model, a multimodal fusion model and a personalized recommendation model; the time series prediction model analyzes physiological time series information based on the LSTM or Transformer algorithm to predict future health risks; the behavior recognition model processes millimeter wave radar signals based on the CNN deep learning algorithm, and combines the XGBoost classifier to perform fall action recognition and classification; the multimodal fusion model evaluates the comprehensive health status based on standardized data and triggers a graded health risk warning; the personalized recommendation model analyzes the comprehensive health status information through a collaborative filtering algorithm to generate a personalized care plan.

[0010] Preferably, the standardized data is pre-set with role-based access rights, and the access rights are adjusted and updated in real time through smart contracts.

[0011] Preferably, the smart contract integrates a multi-protocol communication interface for real-time data sharing and business collaboration across institutions, and the new data generated by the business collaboration flows back to step S3 to perform AI model optimization on the multimodal AI health warning and decision-making model architecture.

[0012] Preferably, the triggering rules of the smart contract include abnormal blood pressure thresholds, fall behavior recognition results, and automatic execution of emergency response, medical record retrieval and drug tracing processes.

[0013] Preferably, the aging-friendly interactive terminal includes a voice interaction module, a one-button alarm function module, an aging-friendly interface interaction module and a behavior data collection module; the voice interaction module integrates an NLP natural language processing algorithm, and has trained a dialect voice analysis model for regional differences, supporting natural language commands to query health reports or call services; the one-button alarm module embeds a physical button on the user device side, supports a single press to trigger an emergency alarm, and sends positioning information and health status to the emergency center and family terminal; the aging-friendly interface interaction module uses a UI smart screen with large fonts and high-contrast colors to provide a visual display of health data, medication reminder pop-ups and activity suggestions; the behavior data collection module is used to capture user interaction operations, and return the operation data to step S3 to perform AI model optimization on the multimodal AI health warning and decision-making model architecture, and dynamically optimize the structure and layout of the aging-friendly interactive terminal according to user operation habits.

[0014] Preferably, the data sources of the multi-source heterogeneous data include smart wearable devices, environmental sensors, behavior monitoring equipment, hospital medical records and community health archive data, and the data collection methods include real-time collection and timed batch collection. The real-time collection is transmitted in real time through the MQTT / HTTP protocol, and the timed batch collection is uploaded in batches according to preset intervals.

[0015] After adopting the above technical solution, the present invention has the following advantages compared with the background technology: The present invention provides a method for constructing smart elderly care services based on Fabric blockchain and big data AI. Blockchain technology is used to ensure that health data cannot be tampered with and is securely shared across institutions. Multimodal AI models are combined to achieve accurate health early warnings, and smart contracts are used to automatically trigger emergency responses and cross-departmental collaboration, significantly shortening rescue time. The system integrates elderly-friendly voice interaction, one-click alarms, and a visual interface, significantly reducing the operational threshold for the elderly. At the same time, AI models and resource allocation are continuously optimized through closed-loop data feedback, thereby improving the efficiency of elderly care services and reducing overall costs. It also comprehensively solves the problems of data silos, delayed responses, and high usage barriers in traditional elderly care models. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 Flow chart of the method of the present invention. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0018] Please refer to Figure 1As shown, the present invention discloses a method for constructing smart elderly care services based on Fabric blockchain and big data AI, comprising the following steps: S1. Collect multi-source heterogeneous data of the elderly and standardize the data; The sources of multi-source heterogeneous data include smart wearable devices (such as smart bracelets and smart watches, which collect physiological data such as heart rate, blood pressure, blood oxygen, number of steps, sleep quality, etc.), environmental sensors (such as temperature and humidity sensors, air quality monitors, etc., which collect parameters such as temperature, humidity, PM2.5 in the elderly care environment), behavior monitoring equipment (such as millimeter-wave radars and cameras (in privacy protection mode), which monitor the activity trajectories and falls of the elderly), and hospital medical records and community health archive data (such as blood sugar records and medication history). Data collection methods include real-time collection and scheduled batch collection. Real-time collection targets key health indicators (such as abnormal heart rate and fall events) and transmits them in real time through the MQTT / HTTP protocol. Scheduled batch collection targets stable data (such as daily blood pressure and blood sugar values) and uploads them in batches according to preset intervals.

[0019] The specific process of standardizing multi-source heterogeneous data is as follows: S11. Deploy data collection ports to obtain multi-source heterogeneous data of the elderly; S12, the acquisition port parses the data, converts the timestamp and unit, and encapsulates it into a standard format; S13. Call the Pandas analysis and processing library to fill missing values; S14. Detect outliers using the Z-score method and replace or directly eliminate outlier records using upper and lower limits; S15. Standardize or normalize the data, remove duplicate data, and generate standardized data. Data normalization includes Min-Max normalization and Z-score normalization.

[0020] S2. The standardized data processed in step S1 is categorized and stored based on the Hyperledger Fabric blockchain and the IPFS distributed network. On-chain storage in Hyperledger Fabric ensures the immutability of structured data (such as user IDs and health indicators). All operation records (such as data access and modification) are traceable through a hash chain to prevent data forgery. Unstructured data (such as medical images) is stored on the IPFS network and linked to the blockchain via content identifiers (CIDs). This avoids the single point of failure risk of traditional centralized storage and addresses the performance bottleneck of large file storage on the blockchain.

[0021] The Hyperledger Fabric blockchain is composed of institutions such as medical institutions, community centers, and family representatives, and each institution participates in consensus as a node.

[0022] Standardized data includes structured data (user ID, health indicators, etc.) and unstructured data (medical images, videos, etc.). Structured data is directly stored in the Hyperledger Fabric on-chain ledger and permission control is performed through smart contracts. Unstructured data is stored in the IPFS distributed network, and the hash value is recorded in the blockchain.

[0023] Standardized data is transmitted using an end-to-end encrypted transmission link. For example, HTTPS is enabled in device-to-server and inter-node communications to prevent man-in-the-middle attacks. Sensitive data (such as user ID) is encrypted using AES-256 before transmission, and the key is managed by the authorized party. Sensitive fields are also desensitized, such as replacing names and addresses with numbers and obfuscating addresses. SHA-256 hashing is performed on associated identifiers (such as device IDs) to avoid direct exposure of device information.

[0024] S3. Build a multimodal AI health warning and decision-making model architecture, analyze the standardized data stored in step S2, and generate health risk warnings and personalized care plans; The multimodal AI health warning and decision-making model architecture includes a time series prediction model, a behavior recognition model, a multimodal fusion model and a personalized recommendation model; the time series prediction model analyzes physiological time series information based on the LSTM or Transformer algorithm to predict future health risks; the behavior recognition model processes millimeter wave radar signals based on the CNN deep learning algorithm and combines the XGBoost classifier to identify and classify fall actions; the multimodal fusion model evaluates the overall health status based on standardized data and triggers graded health risk warnings; the personalized recommendation model analyzes the overall health status information through a collaborative filtering algorithm to generate personalized care plans.

[0025] The specific implementation methods of model training and deployment include: Data preparation: Training set construction: 80% of the historical data is divided as the training set and 20% as the validation set to ensure balanced data distribution.

[0026] Feature engineering: extract statistical features (mean, variance) of time series data, frequency domain features (Fourier transform) of behavioral data, and word vectors (Word2Vec) of text data.

[0027] Model training: LSTM time series prediction: Input the blood pressure sequence of the past 72 hours and output the predicted value for the next 24 hours. The loss function uses MAE (mean absolute error).

[0028] XGBoost fall detection: Input the feature vector of the radar signal (such as acceleration, angle change), output the binary classification result (fall / normal), and optimize the cross entropy loss.

[0029] Transformer multimodal fusion: Physiological data, behavioral features, and text vectors are concatenated and input into the Transformer encoder to output a comprehensive health score.

[0030] Model deployment: Containerized service: Use Docker to encapsulate the trained model as a RESTful API and deploy it to a Kubernetes cluster to support high-concurrency requests.

[0031] Edge computing: Deploy lightweight models (such as TensorFlow Lite) on local servers in retirement communities to reduce cloud transmission latency.

[0032] S4. Authorize cross-institutional data sharing through blockchain smart contracts, pre-set automated process triggering rules for smart contracts, and conduct real-time cross-institutional data sharing and business collaboration; S5. Configure an aging-friendly interactive terminal to receive real-time health risk warnings, personalized care plans, and business collaboration event processes, capture user interaction operations, and return the operation data to step S3 to optimize the AI model of the multimodal AI health warning and decision-making model architecture, forming a closed-loop iterative mechanism.

[0033] Standardized data is pre-set with role-based access permissions, such as setting data access permissions for family members, medical staff, and community managers. Family members can only view basic health data, medical staff can access complete medical records, and administrators have data maintenance permissions to achieve automated processes in scenarios such as emergency response and drug traceability; and access permissions are adjusted and updated in real time through smart contracts, such as temporarily enhancing family members' permissions in an emergency.

[0034] Smart contracts integrate multi-protocol communication interfaces such as 5G, Wi-Fi, GPRS / eMTC, Bluetooth, Ethernet, Zigbee, and LoRa for real-time cross-institutional data sharing and business collaboration. New data generated by business collaboration is fed back to step S3 to optimize the AI model for the multimodal AI health warning and decision-making model architecture. A universal adapter module is developed to support protocol conversion between devices from different manufacturers (e.g., converting Zigbee data to MQTT messages). Apache Kafka is used as a message queue to ensure reliable transmission of high-throughput data (e.g., thousands of devices concurrently). For example: 5G network: used for real-time video consultations and high-speed transmission of health data (bandwidth > 100 Mbps, latency < 10 ms).

[0035] LoRa protocol: connects low-power environmental sensors (such as temperature and humidity monitoring), with a transmission distance of 2-5 kilometers and a battery life of >2 years.

[0036] Wi-Fi: supports near-field interaction with smart home devices (such as smart medicine boxes), and the Bluetooth Mesh network covers the entire retirement community.

[0037] The triggering rules of the smart contract include abnormal blood pressure thresholds and fall behavior recognition results, and automatically execute emergency response, medical record retrieval, and drug traceability processes. The cross-institutional collaboration scenario is implemented as follows after the triggering: Automate emergency response: (1) After the fall alarm is triggered, the smart contract notifies the community emergency station, family members and the hospital; (2) Emergency vehicles receive real-time monitoring images via 5G and optimize route planning; (3) The hospital prepares rescue equipment in advance and retrieves the patient's allergy history and medication records on the chain.

[0038] Drug traceability and management: (1) Drug circulation data (production, transportation, and distribution) are stored on the entire chain, and family members can scan the code to verify the authenticity (blockchain hash comparison); (2) The smart contract automatically reminds you to replace the drug 30 days before its expiration date and links with the procurement system to generate an order.

[0039] The aging-friendly interactive terminal includes a voice interaction module, a one-button alarm function module, an aging-friendly interface interaction module and a behavior data collection module; the voice interaction module integrates the NLP natural language processing algorithm, and has trained a dialect voice analysis model for regional differences, supporting natural language commands to query health reports or call services; the one-button alarm module embeds a physical button on the user device side, supports a single press to trigger an emergency alarm, and sends positioning information and health status to the emergency center and family terminal; the aging-friendly interface interaction module uses a UI smart screen with large fonts and high-contrast colors to provide health data visualization, medication reminder pop-ups and activity suggestions; the behavior data collection module is used to capture user interaction operations, and reflow the operation data to step S3 to perform AI model optimization on the multimodal AI health warning and decision-making model architecture, and dynamically optimize the structure and layout of the aging-friendly interactive terminal according to user operation habits.

[0040] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A method for constructing smart elderly care services based on Fabric blockchain and big data AI, characterized by: The following steps are involved: S1. Collect multi-source heterogeneous data of the elderly and standardize the data; S2. Classify and store the standardized data processed in step S1 based on the Hyperledger Fabric blockchain and the IPFS distributed network; S3. Build a multimodal AI health warning and decision-making model architecture to analyze the standardized data stored in step S2 and generate health risk warnings and personalized care plans; S4. Authorize and share data across institutions through blockchain smart contracts, pre-set automated process triggering rules for smart contracts, and conduct real-time data sharing and business collaboration across institutions. S5. Configure an aging-friendly interactive terminal to receive real-time health risk warnings, personalized care plans, and business collaboration event processes, capture user interaction operations, and return the operation data to step S3 to optimize the AI model of the multimodal AI health warning and decision-making model architecture, forming a closed-loop iterative mechanism.

2. The method for constructing a smart elderly care service based on Fabric blockchain and big data AI as described in claim 1, characterized in that: The standardization process of the multi-source heterogeneous data is specifically as follows: S11. Deploy data collection ports to obtain multi-source heterogeneous data of the elderly; S12, the acquisition port parses the data, converts the timestamp and unit, and encapsulates it into a standard format; S13. Call the Pandas analysis and processing library to fill missing values; S14. Detect outliers using the Z-score method and replace or directly eliminate outlier records using upper and lower limits; S15. Standardize or normalize the data, remove duplicate data, and generate standardized data.

3. The method for constructing a smart elderly care service based on Fabric blockchain and big data AI as claimed in claim 1, characterized in that: The standardized data includes structured data and unstructured data. The structured data is directly stored in the Hyperledger Fabric on-chain ledger and is controlled by smart contracts. The unstructured data is stored in the IPFS distributed network and the hash value is recorded in the blockchain.

4. The method for constructing a smart elderly care service based on Fabric blockchain and big data AI as described in claim 3 is characterized by: The standardized data is transmitted using an end-to-end encrypted transmission link, and sensitive fields are desensitized.

5. The method for constructing a smart elderly care service based on Fabric blockchain and big data AI as claimed in claim 1, characterized in that: The multimodal AI health warning and decision-making model architecture includes a time series prediction model, a behavior recognition model, a multimodal fusion model, and a personalized recommendation model. The time series prediction model analyzes physiological time series information based on the LSTM or Transformer algorithm to predict future health risks. The behavior recognition model processes millimeter wave radar signals based on the CNN deep learning algorithm and combines it with the XGBoost classifier to identify and classify fall actions. The multimodal fusion model evaluates comprehensive health status based on standardized data and triggers graded health risk warnings. The personalized recommendation model analyzes comprehensive health status information through a collaborative filtering algorithm to generate a personalized care plan.

6. The method for constructing a smart elderly care service based on Fabric blockchain and big data AI as claimed in claim 1, characterized in that: The standardized data is pre-set with role-based access permissions, and the access permissions are adjusted and updated in real time through smart contracts.

7. The method for constructing a smart elderly care service based on Fabric blockchain and big data AI as claimed in claim 1, characterized in that: The smart contract integrates a multi-protocol communication interface for real-time data sharing and business collaboration across institutions, and the new data generated by the business collaboration flows back to step S3 to optimize the AI model of the multimodal AI health warning and decision-making model architecture.

8. The method for constructing a smart elderly care service based on Fabric blockchain and big data AI as claimed in claim 1, characterized in that: The triggering rules of the smart contract include abnormal blood pressure thresholds, fall behavior recognition results, and automatic execution of emergency response, medical record retrieval and drug traceability processes.

9. The method for constructing a smart elderly care service based on Fabric blockchain and big data AI as claimed in claim 1, characterized in that: The aging-friendly interactive terminal includes a voice interaction module, a one-button alarm function module, an aging-friendly interface interaction module and a behavior data collection module; the voice interaction module integrates an NLP natural language processing algorithm, and has trained a dialect voice analysis model for regional differences, supporting natural language commands to query health reports or call services; the one-button alarm module embeds a physical button on the user device side, supports a single press to trigger an emergency alarm, and sends positioning information and health status to the emergency center and family terminal; the aging-friendly interface interaction module uses a UI smart screen with large fonts and high-contrast colors to provide a visual display of health data, medication reminder pop-ups and activity suggestions; the behavior data collection module is used to capture user interaction operations, and reflow the operation data to step S3 to perform AI model optimization on the multimodal AI health warning and decision-making model architecture, and dynamically optimize the structure and layout of the aging-friendly interactive terminal according to user operation habits.

10. The method for constructing a smart elderly care service based on Fabric blockchain and big data AI as claimed in claim 1, characterized in that: The data sources of the multi-source heterogeneous data include smart wearable devices, environmental sensors, behavior monitoring equipment, hospital medical records and community health archive data. The data collection methods include real-time collection and timed batch collection. The real-time collection is transmitted in real time through the MQTT / HTTP protocol, and the timed batch collection is uploaded in batches according to preset intervals.

Images