Intelligent underwear data processing method and system based on artificial intelligence

Abstract

The invention provides an intelligent underwear data processing method and system based on artificial intelligence. The method comprises the steps that intelligent underwear collects physiological and environmental data of a user through a built-in sensor; the edge computing chip preprocesses the collected data to obtain first data, performs sensitivity grading marking to obtain second data, performs processing according to sensitivity grades to obtain third data, and selects fourth data needing to be uploaded from the third data; the server receives and decrypts the fourth data, integrates the multi-source data according to a timestamp to construct a health data set, inputs a preset model for analysis, and sends a result to the intelligent underwear; the edge computing chip provides health suggestions through a feedback device according to the analysis result, and collects a user feedback optimization model; and the server simulates the health state of the user by using a digital twinning technology, predicts the health problem and formulates a management scheme. According to the invention, the accuracy of intelligent underwear data analysis is improved, and the security and privacy protection of the user data in the transmission and processing process are enhanced.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an artificial intelligence-based intelligent underwear data processing method and system. Background Art

[0002] With the development of science and technology, smart underwear has gradually emerged. It has built-in sensors that can collect physiological data such as heart rate and body temperature. However, there are many problems with the current data processing of smart underwear. For example, privacy protection is insufficient and a large amount of sensitive physiological data is at risk of leakage. At the same time, data processing fails to fully tap the value of data, making it difficult to provide users with accurate and personalized health analysis and feedback. Therefore, a more accurate, secure and intelligent data processing method is urgently needed. Summary of the invention

[0003] Based on the above problems, the present invention proposes an artificial intelligence-based smart underwear data processing method and system. Through the scheme of the present invention, multiple advanced technologies such as edge computing, deep learning, blockchain and digital twins are combined, which not only improves the accuracy of smart underwear data analysis, but also enhances the security and privacy protection of user data during transmission and processing.

[0004] In view of this, one aspect of the present invention proposes an intelligent underwear data processing method based on artificial intelligence, comprising: The user's physiological data and environmental perception data are collected using smart sensors installed in the smart underwear, wherein the collection frequency is set to different preset values ​​according to the sensor type, and the collected data is time-stamped; After preprocessing the collected physiological data and environmental perception data in the edge computing chip of the smart underwear, first data is obtained; The edge computing chip performs sensitivity classification and labeling on the first data to obtain second data; The edge computing chip processes the second data according to different sensitivity levels to obtain third data; The edge computing chip determines fourth data to be uploaded to the server from the third data; The server decrypts the fourth data after receiving it, and synchronously integrates the multi-source data according to the timestamp to construct a user health data set; The server inputs the user health data set into a preset health analysis model to obtain the user's health analysis result, and sends the health analysis result to the smart underwear; In the entire data processing process, data access logs are recorded with the help of blockchain technology, and each data access generates an unalterable record; The edge computing chip provides personalized health advice and reminders to the user through the feedback device of the smart underwear based on the health analysis results; at the same time, collects the user's interactive feedback on the feedback information, corrects the parameters of the artificial intelligence model in real time, and optimizes the subsequent health analysis results; The server uses digital twin technology to drive a virtual digital model with user physiological data, simulates the user's physical responses in different health states and sports scenarios, predicts potential health problems, and develops a health management plan.

[0005] Optionally, the step of the edge computing chip performing sensitivity classification and labeling on the first data to obtain second data includes: extracting temporal features, spatial features, and semantic features of the first data; Using a preset feature vector extraction algorithm, different types of sensor data are converted into standardized feature vectors; Construct a multimodal feature fusion matrix to fuse data features of different dimensions; Analyze the feature fusion matrix based on the preset sensitivity assessment model; Calculate the privacy exposure risk value of the data, which takes into account the identifiability, relevance and harmfulness of the data; According to the corresponding relationship between the risk value and the preset sensitivity level, the sensitivity level of the data is determined and marked, thereby obtaining the second data.

[0006] Optionally, the step of the edge computing chip processing the second data according to different sensitivity levels to obtain third data includes: extracting first level data from the second data; The server sends a first processing program to the edge computing chip; After the edge computing chip receives the first processing program and processes the first level data using the first processing program, it sends first feedback data to the server; The server receives the first feedback data and verifies the first feedback data; If the verification is passed, a first processing program failure instruction is sent; If the verification fails, adjusting the first processing procedure based on the first feedback data to obtain a second processing procedure; Sending an invalidation instruction of the second processing program and the first processing program to the edge computing chip; The edge computing chip receives the second processing program and processes the first level data using the second processing program; Repeat the above steps until the feedback data sent by the edge computing chip passes the verification.

[0007] Optionally, the step of the edge computing chip determining fourth data to be uploaded to the server from the third data includes: Determining the service processing type and service processing content of the server; Detecting the data duplication degree of the third data, performing data downsampling, retaining key time node data, and obtaining intermediate data with timestamps; Performing data value evaluation on the intermediate data, including: calculating the information entropy value of the intermediate data; evaluating the contribution of the intermediate data to the business processing type and business processing content of the server; generating a data value score for the intermediate data; and establishing a data value priority queue for the intermediate data according to the data value score; Determine the upload priority threshold based on the current network bandwidth status, server load, and data transmission cost; Segmenting the intermediate data in combination with the data value priority queue, extracting data feature vectors, performing data compression, and generating multiple compressed data packets; The compressed data packets are prioritized based on the preset rule screening data and the upload priority threshold to determine an upload data set, thereby obtaining fourth data.

[0008] Optionally, the server decrypts the fourth data after receiving it, and synchronously integrates the multi-source data according to the timestamp to construct the user health data set, including: Receiving and verifying the fourth data, including: verifying the integrity of data transmission: checking the signature of the data packet, checking the integrity of the data packet sequence, and verifying the validity of the timestamp; identifying the source and type of data; parsing the format of the data packet; and generating a receipt confirmation message; Decrypting the fourth data includes: obtaining a decryption key from a security key management system; performing decryption operations on different encryption types: decrypting homomorphically encrypted data, processing differential privacy data, and restoring desensitized data; verifying the correctness of the decryption result; and recording a decryption operation log; The steps of aligning the decrypted data timestamp include: extracting the data timestamp information; unifying the time standard: converting timestamps in different time zones, calibrating the device clock deviation, and processing the sampling time error; establishing the time index structure; and identifying the data timing relationship; The steps of multi-source data synchronization after timestamp alignment include: identifying the association relationship between data; performing time window division: setting the synchronization time window size, handling data sampling frequency differences, and aligning data sampling points; filling missing data: applying interpolation algorithms, marking filling data, and evaluating filling reliability; handling data conflicts; The steps of quality assessment of synchronized data include: detecting outliers: statistical feature analysis, time series pattern recognition, marking abnormal data points; assessing data consistency; calculating data reliability scores; generating quality assessment reports; The steps of data integration after synchronization include: building a unified data model: defining data field mapping, unifying measurement units, and standardizing data formats; merging multi-source data; establishing data association indexes; and optimizing storage structures; Metadata management steps include: recording data processing; updating data catalogs; generating data summaries; The steps of constructing a user health data set include: organizing a hierarchical data structure; establishing a fast retrieval mechanism; implementing an incremental update strategy; and optimizing access performance.

[0009] Optionally, the server inputs the user health data set into a preset health analysis model to obtain the user's health analysis result, and sends the health analysis result to the smart underwear, including: Preprocessing the user health data set includes: performing data standardization processing: normalizing numerical features, encoding category features, and processing time series features; constructing feature vectors: extracting statistical features, generating time series features, and calculating derived features; processing missing values ​​and outliers; and generating an analysis data set; Model selection and configuration, including: loading the following preset health analysis models according to the analysis task: disease risk prediction model, fatigue assessment model, health status classification model; configuring the model parameters of the loaded model: setting the inference threshold, adjusting the model weight, optimizing the prediction parameters; initializing the operating environment of the loaded model; verifying the model status of the loaded model; Performing health risk assessment according to the disease risk prediction model includes: executing disease risk prediction: analyzing abnormal physiological indicators, evaluating health trend changes, and calculating risk probability distribution; generating risk rating: determining risk level, identifying key risk factors, and calculating risk confidence; generating early warning information; and recording the assessment process; Fatigue analysis is performed according to the fatigue assessment model, including: analyzing physiological data patterns: assessing heart rate variability, analyzing activity intensity, and monitoring recovery status; calculating fatigue index: integrating multi-dimensional indicators, applying fatigue assessment algorithms, and generating fatigue levels; assessing recovery suggestions; and recording analysis results; Comprehensively process the above risk assessment results and fatigue analysis results according to the health status classification model, including: integrating multi-model prediction results: weighted fusion prediction values, eliminating prediction conflicts, and optimizing result reliability; generating health reports: summarizing key findings, providing detailed explanations, and formulating recommended solutions; marking important reminders; and storing comprehensive analysis results; Generate personalized suggestions based on the comprehensive analysis results, including: customize suggestions based on user portraits: consider user health history, analyze living habits, and evaluate exercise ability; formulate improvement plans: generate action suggestions, set health goals, and plan intervention measures; generate reminder strategies; and optimize suggestion content; Generate a health analysis result based on the comprehensive analysis result and the personalized suggestion; The steps for transmitting health analysis results include: packaging analysis results: compressing data packets, adding security identifiers, generating verification codes; establishing a secure channel; performing batch transmission; confirming that the transmission is complete; Feedback tracking steps include: monitoring the implementation of recommendations; collecting user feedback; updating personalized parameters; and optimizing health analysis models.

[0010] Optionally, the step of the edge computing chip processing the second data according to different sensitivity levels to obtain third data further includes: extracting second level data from the second data; dividing the second level data into first data to be processed, second data to be processed and third data to be processed; Initial TEE environment, verify the integrity of the environment, and generate a zero-knowledge proof key pair within the TEE; Extracting features from the first data to be processed and constructing a circuit required for proof; Generating a zero-knowledge proof includes: converting the first to-be-processed data into a polynomial expression; generating proof parameters using an elliptic curve cryptography algorithm; constructing a commitment scheme to ensure data correctness; generating a non-interactive zero-knowledge proof; and verifying the validity of the generated zero-knowledge proof; Generate homomorphic encryption keys based on lattice cryptography; Splitting the second to-be-processed data into data blocks suitable for homomorphic operations; Perform homomorphic encryption on each data block to generate first encrypted data that supports ciphertext domain operations, and record encryption parameters for subsequent decryption; Determine noise parameters based on Laplace mechanism; According to the noise parameter, adding differential privacy noise to the third data to be processed to obtain second encrypted data; Verify the data availability after adding differential privacy noise, and dynamically adjust the noise parameters to balance privacy protection and data availability; Integrate the zero-knowledge proof, the first encrypted data, and the second encrypted data generated above to obtain third data; Classify and store the third data, establish a data index structure, record the metadata of the data processing process, and generate a data processing report; Verify the integrity of third-party data, check privacy protection effects, evaluate data availability, generate processing records and upload them to the blockchain.

[0011] Optionally, the step of the edge computing chip processing the second data according to different sensitivity levels to obtain third data further includes: extracting third level data from the second data; Determine the privacy budget parameters based on the characteristics of the third-level data and the privacy protection requirements; According to the data type of the third level data, select the corresponding noise distribution; Carry out preliminary cleaning of the third-level data to remove erroneous values, outliers, and irrelevant redundant information; Generate the corresponding random noise using a random number generator based on the selected noise distribution and privacy budget parameters; The generated random noise is added to the preprocessed third level data to obtain third data.

[0012] Optionally, the step of the edge computing chip processing the second data according to different sensitivity levels to obtain third data further includes: extracting fourth level data from the second data; Processing the fourth-level data using a preset fourth-level data processing model to obtain third data; The steps of constructing the fourth level data processing model include: Determine a plurality of participants holding fourth-level training data of the fourth level, and pre-process the fourth-level training data in the local edge computing devices or data centers of the participants; wherein the data of each participant is always retained locally, and the original data is not transmitted externally; Each of the above-mentioned participants independently selects the corresponding machine learning model architecture based on its own data characteristics and business needs, and randomly initializes the model parameters to prepare for local training; Using the local fourth-level training data, each participant starts to train the initialized model, using an optimization algorithm to continuously update the model parameters in multiple rounds of iterations, so that the model gradually fits the characteristics and rules of the local data; After a preset round of local training, each of the above-mentioned participants extracts the key parameters of the trained model, and uploads these abstracted key parameters without original data information to the server through a secure communication channel; After receiving the key parameters uploaded by each participant, the server assigns corresponding weights according to the preset aggregation algorithm, comprehensively considers factors such as the data volume and data quality of each participant, aggregates and integrates all key parameters, and updates the global model; Transmitting the updated new parameters of the global model back to each of the participants through a secure channel; Each of the participants uses the received new parameters to replace the original corresponding parts in the local model, completes the update of the local model, and obtains the fourth-level data processing model of each party; Based on the updated fourth-level data processing model, each of the participants continues to use new data for training, upload parameters, receive new global models, and continuously optimizes model performance, so that the fourth-level data processing model's processing capabilities for fourth-level data gradually improve.

[0013] Another aspect of the present invention provides an artificial intelligence-based smart underwear data processing system for executing an artificial intelligence-based smart underwear data processing method, comprising: a server, and smart underwear provided with a smart sensor and an edge computing chip; wherein, The smart sensor is configured to: collect physiological data and environmental perception data of the user, wherein the collection frequency is set to different preset values ​​according to the sensor type, and the collected data is time-stamped; The edge computing chip is configured as: After preprocessing the collected physiological data and environmental perception data, first data is obtained; Performing sensitivity classification and labeling on the first data to obtain second data; Processing the second data according to different sensitivity levels to obtain third data; Determining fourth data to be uploaded to the server from the third data; The server is configured to: Decrypting the fourth data after receiving it, synchronously integrating the multi-source data according to the timestamp, and constructing a user health data set; Inputting the user health data set into a preset health analysis model to obtain the user's health analysis result, and sending the health analysis result to the smart underwear; In the entire data processing process, data access logs are recorded with the help of blockchain technology, and each data access generates an unalterable record; The edge computing chip is further configured to: provide personalized health advice and reminders to the user through the feedback device of the smart underwear based on the health analysis results; at the same time, collect interactive feedback from the user on the feedback information, correct the parameters of the artificial intelligence model in real time, and optimize subsequent health analysis results; The server is also configured to: utilize digital twin technology to drive a virtual digital model with user physiological data, simulate the user's physical responses in different health states and sports scenarios, predict potential health problems, and develop a health management plan.

[0014] By adopting the technical solution of the present invention, the data processing method of smart underwear based on artificial intelligence includes: using the smart sensor arranged in the smart underwear to collect the physiological data and environmental perception data of the user; pre-processing the collected physiological data and environmental perception data in the edge computing chip of the smart underwear to obtain the first data; the edge computing chip performs sensitivity classification and marking on the first data to obtain the second data; the edge computing chip processes the second data according to different sensitivity levels to obtain the third data; the edge computing chip determines the fourth data uploaded to the server from the third data; the server decrypts the fourth data after receiving it, and synchronously integrates the multi-source data according to the timestamp to construct the user health data set; the server The user health data set is input into a preset health analysis model to obtain the user's health analysis results, and the health analysis results are sent to the smart underwear; in the entire data processing process, the data access log is recorded with the help of blockchain technology, and each data access generates an unalterable record; the edge computing chip provides personalized health advice and reminders to the user through the feedback device of the smart underwear based on the health analysis results; at the same time, the user's interactive feedback on the feedback information is collected, the parameters of the artificial intelligence model are corrected in real time, and the subsequent health analysis results are optimized; the server uses digital twin technology to drive the virtual digital model with the user's physiological data, simulate the user's physical response in different health states and sports scenes, predict potential health problems, and formulate a health management plan. Through the scheme of the present invention, a variety of advanced technologies such as edge computing, deep learning, blockchain and digital twins are combined, which not only improves the accuracy of smart underwear data analysis, but also enhances the security and privacy protection of user data during transmission and processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is a flow chart of an intelligent underwear data processing method based on artificial intelligence provided by an embodiment of the present invention; Figure 2 It is a schematic block diagram of an intelligent underwear data processing system based on artificial intelligence provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0016] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0017] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0018] The terms "first", "second", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices.

[0019] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0020] Refer to the following Figure 1 to Figure 2 To describe an artificial intelligence-based smart underwear data processing method and system provided according to some embodiments of the present invention.

[0021] like Figure 1 As shown, an embodiment of the present invention provides a smart underwear data processing method based on artificial intelligence, comprising: The user's physiological data and environmental perception data are collected using smart sensors installed in the smart underwear, wherein the collection frequency is set to different preset values ​​according to the sensor type, and the collected data is time-stamped; In this step, the smart sensors include but are not limited to heart rate sensors, body temperature sensors, pressure sensors, image acquisition sensors, etc., through which the user's physiological data and environmental perception data can be collected. For example, the image acquisition sensor can be used to collect skin image data; and the edge computing chip in the smart underwear can use image recognition technology to process the skin image data, monitor the skin condition, identify abnormal skin conditions (such as rash, redness and swelling), and fuse image features with other physiological data for analysis.

[0022] After preprocessing the collected physiological data and environmental perception data in the edge computing chip of the smart underwear, first data is obtained; In this step, the preprocessing includes filtering and denoising to remove motion artifacts caused by body movement, dynamically adjusting filter parameters using an adaptive filtering algorithm, and data normalization to map data of different scales to the [0, 1] interval. The edge computing chip is a customized low-power, high-computing chip with a built-in trusted execution environment (TEE), which can perform encryption operations related to zero-knowledge proofs and sensitive data processing operations and other data processing operations in the TEE.

[0023] The edge computing chip performs sensitivity classification and labeling on the first data to obtain second data; In this step, the privacy / sensitivity level of the smart underwear data can be graded using a multimodal data fusion algorithm, wherein the privacy / sensitivity level includes at least five levels.

[0024] The edge computing chip processes the second data according to different sensitivity levels to obtain third data; In this step, you can choose whether to use zero-knowledge proof technology to process private data based on the data type and real-time requirements. When processing, generate a proof key and a verification key according to the predetermined zero-knowledge proof algorithm, use the proof key to process private data and generate a zero-knowledge proof, and temporarily store it together with the processing results. The data that has been initially processed by edge computing is encrypted using the Advanced Encryption Standard (AES) algorithm and transmitted to the server through the IoT communication module in accordance with the secure transmission protocol (SSL / TLS); using differential privacy technology, add controllable random noise to the data during big data analysis to protect the statistical characteristics of group data while concealing individual privacy data; if there is zero-knowledge proof-related data, the verification key, zero-knowledge proof and processing results are transmitted in sequence.

[0025] The edge computing chip determines fourth data to be uploaded to the server from the third data; The server decrypts the fourth data after receiving it, and synchronously integrates the multi-source data according to the timestamp to construct a user health data set; The server inputs the user health data set into a preset health analysis model to obtain the user's health analysis results (including but not limited to disease risk warning and physical fatigue level assessment), and sends the health analysis results to the smart underwear; It is understandable that the health analysis model is pre-trained based on a deep learning algorithm, and can identify physiological data patterns corresponding to different health states and output the user's health analysis results. The health analysis model training method includes: collecting a large number of user physiological data samples labeled with health states, using transfer learning, first pre-training the model infrastructure on a public health big data set, and then fine-tuning the model parameters for the specific data collected by the smart underwear, and optimizing the loss function to convergence.

[0026] In the entire data processing process, data access logs are recorded with the help of blockchain technology, and each data access generates an unalterable record; It is understandable that users can view access details and set third-party data access permissions through the smart terminal APP; the blockchain technology adopts the form of a consortium chain, with smart underwear manufacturers, cloud service providers, and authorized medical institutions acting as consortium nodes to jointly maintain data access ledgers.

[0027] The edge computing chip provides personalized health advice and reminders to the user based on the health analysis results through the feedback device of the smart underwear (such as a micro vibration motor, a voice device, a flexible display screen, etc.); at the same time, collects the user's interactive feedback on the feedback information, corrects the parameters of the artificial intelligence model in real time, and optimizes the subsequent health analysis results; The server uses digital twin technology to drive a virtual digital model with user physiological data, simulates the user's physical responses in different health states and sports scenarios, predicts potential health problems, and develops a health management plan.

[0028] It can be understood that with the help of three-dimensional modeling technology, a three-dimensional pressure distribution model is constructed based on the pressure data collected by the pressure sensor equipped with the smart underwear to assist in judging the comfort of the underwear and body changes; there is a mapping relationship between the digital twin model and the body's physiological system, and the accuracy of the simulation is ensured by continuously collecting and comparing (processed) physiological data with model simulation data, and dynamically calibrating and updating.

[0029] The solution of this embodiment combines multiple advanced technologies such as edge computing, deep learning, blockchain and digital twins, which not only improves the accuracy of smart underwear data analysis, but also enhances the security and privacy protection of user data during transmission and processing.

[0030] In some possible implementations of the present invention, the step of the edge computing chip performing sensitivity classification and labeling on the first data to obtain the second data includes: extracting temporal features, spatial features, and semantic features of the first data; Using a preset feature vector extraction algorithm, different types of sensor data are converted into standardized feature vectors; Construct a multimodal feature fusion matrix to fuse data features of different dimensions; Analyze the feature fusion matrix based on the preset sensitivity assessment model; Calculate the privacy exposure risk value of the data, which takes into account the identifiability, relevance and harmfulness of the data; According to the corresponding relationship between the risk value and the preset sensitivity level, the sensitivity level of the data is determined and marked, thereby obtaining the second data.

[0031] In this embodiment, the sensitivity level can be divided into the following five levels: Level 1: Direct identification information (e.g. biometric data such as fingerprints, facial recognition features, iris information, genetic information, etc.); Level 2: Indirect identification information (such as precise physiological index data, undisclosed major disease information, etc.); Level 3: sensitive behavioral data (such as activity patterns, dressing habits, etc.); Level 4: General behavioral data (such as basic physiological status; age distribution of smart underwear users, regional distribution data, average sleep duration of users in a certain area, and other aggregated data); Level 5: Environmental perception data (such as temperature and humidity), basic parameters disclosed by the equipment, and other data not included in the above four levels.

[0032] In this embodiment, the privacy / sensitivity level is divided into five levels so that corresponding processing methods can be used for data belonging to different levels in the future. It should be understood that the division of privacy / sensitivity levels in this embodiment is only for illustrative purposes; the division of privacy / sensitivity levels can be defined as needed, or can be intelligently divided into levels through big data analysis and artificial intelligence, or other level division methods can be used, and the embodiments of the present invention do not limit this.

[0033] In some possible implementations of the present invention, a dynamic adjustment step and a log recording step are also included; the dynamic adjustment step specifically includes: continuously monitoring data usage and privacy risk changes; dynamically adjusting classification standards based on user feedback and privacy requirements; updating data processing strategies in real time; optimizing sensitivity assessment model parameters; the log recording step specifically includes: recording data classification results on the blockchain; recording each data access and processing operation; saving classification adjustment history; and establishing a data processing audit trail.

[0034] The solution of this embodiment can achieve hierarchical and refined protection of data privacy, reduce the risk of sensitive data leakage, and ensure user privacy security; optimize computing resource utilization through differentiated processing strategies, reduce encryption and decryption overhead, and improve data processing efficiency; achieve refined control of data access, provide a flexible permission management mechanism, and support multi-scenario data application; ensure that data processing is traceable throughout the process, support privacy compliance audits, and facilitate problem location and processing; support dynamic adjustment of privacy protection strategies to adapt to the privacy needs of different users and meet the requirements of different application scenarios.

[0035] In some possible implementations of the present invention, the step of the edge computing chip processing the second data according to different sensitivity levels to obtain third data includes: Extracting first-level data (i.e., data with a sensitivity level marked as the first level) from the second data; The server sends a first processing program to the edge computing chip; After the edge computing chip receives the first processing program and processes the first level data using the first processing program, it sends first feedback data to the server; The server receives the first feedback data and verifies the first feedback data; If the verification is passed, a first processing program failure instruction is sent; If the verification fails, adjusting the first processing procedure based on the first feedback data to obtain a second processing procedure; Sending an invalidation instruction of the second processing program and the first processing program to the edge computing chip; The edge computing chip receives the second processing program and processes the first level data using the second processing program; Repeat the above steps until the feedback data sent by the edge computing chip passes the verification.

[0036] In this embodiment, the first level of data is processed using a one-time processing program sent by the server to the edge computing chip, specifically: the server sends a one-time first processing program to the edge computing chip; the edge computing chip receives the first processing program and uses the first processing program to process the data, and then sends the first feedback data to the server; the server receives the first feedback data and verifies the first feedback data; if the verification is passed, an invalidation instruction of the first processing program is sent; if the verification is not passed, the first processing program is adjusted based on the first feedback data to obtain a second processing program; the second processing program and the first processing program are invalidated to the edge computing chip; the edge computing chip receives the second processing program and uses the second processing program to process the data; the above steps are repeated until the feedback data sent by the edge computing chip passes the verification.

[0037] In some possible implementations of the present invention, the step of the edge computing chip determining, from the third data, fourth data to be uploaded to the server includes: Determining the service processing type and service processing content of the server; Detecting the degree of data duplication of the third data (including: calculating a data similarity matrix, identifying redundant data segments, and marking mergeable data), and performing data downsampling, retaining key time node data, and obtaining intermediate data with timestamps; Performing data value evaluation on the intermediate data, including: calculating the information entropy value of the intermediate data; evaluating the contribution of the intermediate data to the business processing type and business processing content of the server (including: calculating feature importance score, analyzing data novelty, evaluating data representativeness, etc.); generating a data value score for the intermediate data; establishing a data value priority queue for the intermediate data according to the data value score; Determine the upload priority threshold based on the current network bandwidth status, server load, and data transmission cost; Segment the intermediate data in combination with the data value priority queue, extract data feature vectors, perform data compression (including: applying a lossless compression algorithm, evaluating compression efficiency, and ensuring data reproducibility), and generate multiple compressed data packets; The data is screened based on preset rules (including: checking data integrity, verifying data timeliness, and evaluating privacy risks) and the upload priority threshold value, and priority sorting is performed on the compressed data packets to determine an upload data set to obtain fourth data.

[0038] The solution of this embodiment can reduce the amount of data transmission, optimize network bandwidth utilization, reduce transmission delay, and improve data throughput; ensure the value of uploaded data, reduce data redundancy, maintain data integrity, and improve data availability; optimize storage space usage, reduce processing overhead, improve computing efficiency, and balance system load; support dynamic network environments, adapt to different data characteristics, meet diverse needs, and facilitate policy adjustments.

[0039] In some possible implementations of the present invention, the server decrypts the fourth data after receiving it, and synchronously integrates the multi-source data according to the timestamp to construct the user health data set, including: Receiving and verifying the fourth data, including: verifying the integrity of data transmission: checking the signature of the data packet, checking the integrity of the data packet sequence, and verifying the validity of the timestamp; identifying the source and type of data; parsing the format of the data packet; and generating a receipt confirmation message; Decrypting the fourth data includes: obtaining a decryption key from a security key management system; performing decryption operations on different encryption types: decrypting homomorphically encrypted data, processing differential privacy data, and restoring desensitized data; verifying the correctness of the decryption result; and recording a decryption operation log; The steps of aligning the decrypted data timestamp include: extracting the data timestamp information; unifying the time standard: converting timestamps in different time zones, calibrating the device clock deviation, and processing the sampling time error; establishing the time index structure; and identifying the data timing relationship; The steps of multi-source data synchronization after timestamp alignment include: identifying the association relationship between data; performing time window division: setting the synchronization time window size, handling data sampling frequency differences, and aligning data sampling points; filling missing data: applying interpolation algorithms, marking filling data, and evaluating filling reliability; handling data conflicts; The steps of quality assessment of synchronized data include: detecting outliers: statistical feature analysis, time series pattern recognition, marking abnormal data points; assessing data consistency; calculating data reliability scores; generating quality assessment reports; The steps of data integration after synchronization include: building a unified data model: defining data field mapping, unifying measurement units, and standardizing data formats; merging multi-source data; establishing data association indexes; and optimizing storage structures; Metadata management steps include: recording data processing; updating data catalogs; generating data summaries; The steps of constructing a user health data set include: organizing a hierarchical data structure; establishing a fast retrieval mechanism; implementing an incremental update strategy; and optimizing access performance.

[0040] The solutions of the embodiments of the present invention can ensure data integrity, improve data accuracy, enhance data consistency, and optimize data availability; can also improve data processing speed, optimize resource utilization, reduce processing delays, and support large-scale data processing; facilitate data retrieval, support complex queries, provide real-time access, and optimize data analysis efficiency; ensure data security, provide fault-tolerant mechanisms, support data recovery, and ensure service stability; support data source expansion, adapt to data scale growth, facilitate functional upgrades, and be compatible with new data types; ensure data security processing, control access rights, and protect user privacy.

[0041] In some possible implementations of the present invention, the server inputs the user health data set into a preset health analysis model to obtain the user's health analysis result, and sends the health analysis result to the smart underwear, including: Preprocessing the user health data set includes: performing data standardization processing: normalizing numerical features, encoding category features, and processing time series features; constructing feature vectors: extracting statistical features, generating time series features, and calculating derived features; processing missing values ​​and outliers; and generating an analysis data set; Model selection and configuration, including: loading the following preset health analysis models according to the analysis task: disease risk prediction model, fatigue assessment model, health status classification model; configuring the model parameters of the loaded model: setting the inference threshold, adjusting the model weight, optimizing the prediction parameters; initializing the operating environment of the loaded model; verifying the model status of the loaded model; It is understandable that the preset health analysis model includes but is not limited to one or more of a disease risk prediction model, a fatigue level assessment model, and a health status classification model.

[0042] Performing health risk assessment according to the disease risk prediction model includes: executing disease risk prediction: analyzing abnormal physiological indicators, evaluating health trend changes, and calculating risk probability distribution; generating risk rating: determining risk level, identifying key risk factors, and calculating risk confidence; generating early warning information; and recording the assessment process; Fatigue analysis is performed according to the fatigue assessment model, including: analyzing physiological data patterns: assessing heart rate variability, analyzing activity intensity, and monitoring recovery status; calculating fatigue index: integrating multi-dimensional indicators, applying fatigue assessment algorithms, and generating fatigue levels; assessing recovery suggestions; and recording analysis results; Comprehensively process the above risk assessment results and fatigue analysis results according to the health status classification model, including: integrating multi-model prediction results: weighted fusion prediction values, eliminating prediction conflicts, and optimizing result reliability; generating health reports: summarizing key findings, providing detailed explanations, and formulating recommended solutions; marking important reminders; and storing comprehensive analysis results; Generate personalized suggestions based on the comprehensive analysis results, including: customize suggestions based on user portraits: consider user health history, analyze living habits, and evaluate exercise ability; formulate improvement plans: generate action suggestions, set health goals, and plan intervention measures; generate reminder strategies; and optimize suggestion content; Generate a health analysis result based on the comprehensive analysis result and the personalized suggestion; The steps for transmitting health analysis results include: packaging analysis results: compressing data packets, adding security identifiers, generating verification codes; establishing a secure channel; performing batch transmission; confirming that the transmission is complete; Feedback tracking steps include: monitoring the implementation of recommendations; collecting user feedback; updating personalized parameters; and optimizing health analysis models.

[0043] The solution of this embodiment can improve the accuracy of risk prediction, enhance the precision of fatigue assessment, optimize the reliability of health assessment, and improve the timeliness of early warning; it can realize personalized customization of suggestions, adapt to individual differences, provide accurate health guidance, and optimize user experience; support real-time health monitoring, quickly respond to abnormal situations, issue early warning information in time, and dynamically adjust suggestions; provide explanations of analysis results, clarify sources of risks, support tracking of decision-making basis, and enhance user understanding; optimize computing resource utilization, improve processing efficiency, ensure service stability, and support large-scale analysis; support dynamic updating of models, adapt to changes in health status, optimize analysis strategies, and improve system robustness.

[0044] In some possible implementations of the present invention, the method for constructing the health analysis model includes: Training data collection and preprocessing steps include: collecting multi-source training data (including: physiological data collected by smart underwear, user health record data, annotated data provided by medical institutions, and public health data sets); data cleaning and standardization (including: removing outliers and noise, processing missing values, and unifying data formats and units); building a standard data set; generating a data quality report; Feature engineering steps include: time series feature extraction (including: calculating statistical features, extracting trend features, constructing period features, and generating frequency domain features); feature selection and dimensionality reduction (including: calculating feature importance, performing correlation analysis, and applying dimensionality reduction algorithms); constructing feature combinations; and verifying feature validity. Model architecture design steps, including: designing the deep learning network structure (including: defining the input layer structure, designing the feature extraction layer, building the attention mechanism, designing the output layer structure); configuring model parameters (including: setting the learning rate strategy, selecting the optimizer type, defining the loss function); implementing model components; verifying the model structure; Model training steps, including: performing transfer learning (including: loading pre-trained models, freezing base layer parameters, and adapting to target tasks); implementing phased training (performing pre-training phases, performing fine-tuning training, and optimizing model parameters); monitoring the training process; and recording training logs; Model evaluation and optimization steps include: performing multi-dimensional evaluation (including: calculating accuracy indicators, evaluating generalization ability, analyzing prediction bias); performing model optimization (including: adjusting network structure, optimizing hyperparameters, improving training strategies); performing comparative experiments; and generating evaluation reports. The steps of model fusion include: building an integrated learning framework (including: selecting basic models, designing fusion strategies, and optimizing weight distribution); implementing model integration (including: training sub-models, optimizing fusion parameters, and verifying integration effects); evaluating fusion performance; and selecting the optimal solution. Model deployment steps, including: model conversion and optimization (including: compressing model size, quantifying model parameters, optimizing computing efficiency); deployment environment configuration (including: setting up the operating environment, configuring computing resources, optimizing inference performance); implementing performance testing; monitoring operating status; Continuous optimization steps include: collecting feedback from actual applications; updating training data; optimizing model structure; and improving deployment strategies.

[0045] The solution of the embodiments of the present invention can improve prediction accuracy, enhance model stability, improve generalization ability, and reduce prediction delay; it can support personalized prediction, adapt to changes in data distribution, handle diversified scenarios, and enhance migration capabilities; optimize reasoning speed, reduce resource consumption, improve concurrent processing capabilities, and reduce storage overhead; it is easy to update the model, support online learning, simplify the deployment process, and facilitate fault diagnosis.

[0046] In some possible implementations of the present invention, the step of the edge computing chip processing the second data according to different sensitivity levels to obtain third data further includes: Extracting second-level data (i.e., data with a sensitivity level marked as the second level) from the second data; dividing the second level data into first data to be processed, second data to be processed and third data to be processed; It is understandable that in order to keep data confidential more finely and accurately, the second-level data that needs to be transmitted to the server can be further refined. According to the different protection strengths / characteristics of data using zero-knowledge proof technology, homomorphic encryption technology, and differential privacy noise encryption technology, the second-level data can be divided into first data to be processed, second data to be processed, and third data to be processed. The first data to be processed, the second data to be processed, and the third data to be processed are processed using zero-knowledge proof technology, homomorphic encryption technology, and differential privacy noise encryption technology, respectively.

[0047] Initial TEE environment, verify the integrity of the environment, and generate a zero-knowledge proof key pair within the TEE; It is understandable that TEE is the abbreviation of Trusted Execution Environment, which is a hardware-based secure computing environment designed to protect the security of code and data running in it, preventing unauthorized access and tampering. TEE provides a secure execution environment by isolating part of the processor from the main operating system, ensuring the security of sensitive operations and data.

[0048] Extracting features from the first data to be processed and constructing a circuit required for proof; It is understandable that, in essence, the "circuit" here is not a circuit in the real physical sense, but an abstract model of calculation, similar to the combination of logic gates (AND gates, OR gates, NOT gates, etc.) in digital circuits, which is used to define the process and rules of a series of calculation operations. It breaks down complex data processing tasks into logical links composed of basic operation units, converts and operates the input data according to the predetermined logic, and finally outputs the results. Polynomial functions are often used to represent these "circuits". Polynomials have many excellent properties in cryptography, such as high computational efficiency and mathematical problem characteristics based on security. For a simple addition circuit, the corresponding polynomial may be f(x,y)=x+y, and the multiplication circuit is g(x,y)=x*y. Complex operations are the result of multiple polynomials nested and combined. In the context of zero-knowledge proof, to prove a statement about private data to the verifier, you must first build such a circuit. For example, to prove that the user's age is greater than 18 years old (assuming that age is private data), build a comparison circuit with the user's age as input and a Boolean value as output. If it is greater than 18, the output is "true", otherwise it is "false". This circuit defines the target relationship of the proof, and all subsequent proof processes revolve around it. Different zero-knowledge proof protocols (such as zk-SNARKs and zk- STARKs) have different requirements and optimization methods for circuit construction. Some protocols prefer simple circuits to reduce computational complexity and proof size; others focus on scalability and can cope with large-scale and complex data processing scenarios. When building a circuit, it is necessary to adjust the circuit structure and select appropriate polynomial representations according to the selected protocol to ensure that zero-knowledge proofs can be generated and verified smoothly in the future. When the circuit is running, only the abstract representation of the data is processed, and the real private data will not be exposed. Taking cryptocurrency transactions as an example, to prove that the transaction amount is within a certain legal range, the constructed circuit processes the encrypted form of the amount or the result after homomorphic encryption. The outside world cannot see the real amount, but the legality of the transaction can be verified through the circuit output.

[0049] Generating a zero-knowledge proof includes: converting the first to-be-processed data into a polynomial expression; generating proof parameters using an elliptic curve cryptography algorithm; constructing a commitment scheme to ensure data correctness; generating a non-interactive zero-knowledge proof; and verifying the validity of the generated zero-knowledge proof; Generate homomorphic encryption keys based on lattice cryptography; Splitting the second to-be-processed data into data blocks suitable for homomorphic operations; Perform homomorphic encryption on each data block (including: mapping the data to a polynomial ring; adding random noise; performing lattice cryptographic operations), generate first encrypted data that supports ciphertext domain operations, and record encryption parameters for subsequent decryption; Determine the noise parameters based on the Laplace mechanism (including: evaluating the privacy budget ε value; calculating the noise distribution parameters; generating a random noise sequence); According to the noise parameter, adding differential privacy noise to the third data to be processed to obtain second encrypted data; Verify the data availability after adding differential privacy noise, and dynamically adjust the noise parameters to balance privacy protection and data availability; Integrate the zero-knowledge proof, the first encrypted data, and the second encrypted data generated above to obtain third data; Classify and store the third data, establish a data index structure, record the metadata of the data processing process, and generate a data processing report; Verify the integrity of third-party data, check privacy protection effects, evaluate data availability, generate processing records and upload them to the blockchain.

[0050] Through the solution of this embodiment, zero-knowledge proof of data can be achieved to ensure that the original data is not exposed; support encrypted data calculation, provide quantifiable differential privacy protection, and prevent data re-identification attacks; retain the core statistical characteristics of the data, support data analysis in an encrypted state, ensure the accuracy of data analysis results, and achieve a balance between privacy protection and data utility; optimize the efficiency of cryptographic operations, reduce encryption overhead, support parallel processing, and reduce storage space occupancy; ensure the security of the data processing environment, support integrity verification of the data processing process, provide reliable audit tracking, and prevent unauthorized data access; support different types of sensitive data processing, adapt to different application scenarios, facilitate function expansion and upgrades, and be compatible with mainstream privacy computing frameworks.

[0051] In some possible implementations of the present invention, the step of the edge computing chip processing the second data according to different sensitivity levels to obtain third data further includes: Extracting third-level data (i.e., data with a sensitivity level marked as the third level) from the second data; According to the characteristics of the third-level data and the privacy protection requirements, determine the privacy budget parameters (the smaller the privacy budget, the greater the added noise intensity, the higher the degree of data privacy protection, but it will also have a greater impact on data availability; conversely, the larger the privacy budget, the smaller the noise, the higher the data availability, and the weaker the privacy protection); According to the data type of the third-level data, select the corresponding noise distribution (if it is continuous data, Gaussian distribution is usually used; if it is discrete data, Laplace distribution is often used); Carry out preliminary cleaning of the third-level data to remove erroneous values, outliers, and irrelevant redundant information (to ensure data quality and ensure that subsequent noise addition and processing are based on reliable data); Based on the selected noise distribution and privacy budget parameters, a random number generator is used to generate the corresponding random noise (for example, if a Laplace distribution is used, a noise value sequence that conforms to the distribution is generated given a scale parameter for mixing into the data); The generated random noise is added to the preprocessed third level data to obtain third data.

[0052] It is understandable that in key links such as data aggregation and statistical calculation, the noise is added to the corresponding elements of the original data one by one (for numerical data), so that each data point after processing is integrated with the noise component, thereby confusing the real data and preventing attackers from inferring individual sensitive information through tiny data differences. The third data with added noise is used for subsequent data analysis, mining, or machine learning model training. Even if a third party obtains this data, it is difficult to accurately restore the original data due to the interference of noise. At the same time, after setting the privacy budget reasonably, the retained data statistical characteristics can still support general business needs, such as group trend analysis and correlation research. At a specific stage of the data processing process, the evaluation indicators of differential privacy are used, such as calculating the probability of change in the output results of adjacent data sets before and after processing, to check whether the pre-set privacy budget is met. If it is found that privacy protection is insufficient or data availability is excessively damaged, the privacy budget parameters or noise distribution are readjusted to optimize the processing effect.

[0053] The solution of this embodiment effectively conceals the precise information of individual data by adding carefully designed noise, so that even if a third party has a large amount of relevant data, it is extremely difficult to reversely deduce the original data of a specific individual, thereby protecting the privacy content in the third-level data; although noise is introduced, a reasonable setting of the privacy budget can ensure that the approximate statistical characteristics of the data are retained, and it can still be used for group-level data analysis and macro trend derivation, so that the business will not completely lose the value of data utilization while protecting privacy, and take into account both privacy and functional requirements; the application of differential encryption technology helps to meet the requirements of various data privacy regulations, and in data disclosure and sharing scenarios, it proves to regulators and users that sufficient privacy protection measures have been taken, reduces compliance risks, and enhances user trust; because noise destroys the accuracy of the data, it is difficult for attackers to obtain sensitive information using common means such as differential attacks, thereby enhancing the security and robustness of data in the face of potential privacy attacks.

[0054] In some possible implementations of the present invention, the step of the edge computing chip processing the second data according to different sensitivity levels to obtain third data further includes: Extracting fourth-level data (i.e., data with a sensitivity level marked as the fourth level) from the second data; Processing the fourth-level data using a preset fourth-level data processing model to obtain third data; The steps of constructing the fourth level data processing model include: Identify multiple participants who hold Level 4 training data of Level 4 (which may be data centers of smart wearable device manufacturers, medical and health institutions, etc.; or various smart wearable devices (such as smart underwear, smart bracelets, etc.), various testing equipment, etc.), and pre-process the Level 4 training data in the local edge computing devices or data centers of the participants (pre-processing includes basic operations such as noise removal and normalization to make the data suitable for model training); the data of each participant is always kept locally, and the original data is not transmitted externally; Each of the above-mentioned participants independently selects the corresponding machine learning model architecture based on its own data characteristics and business needs (e.g., if focusing on health trend prediction, the long short-term memory network (LSTM) can be selected; if doing simple classification tasks, the decision tree model is also one of the options; etc.), and randomly initializes the model parameters to prepare for local training; Using the local fourth-level training data, each participant starts to train the initialized model, using optimization algorithms (such as stochastic gradient descent (SGD) and its variants Adagrad, Adam, etc.), and continuously updating the model parameters in multiple rounds of iterations to gradually fit the characteristics and rules of the local data; After a preset round of local training, each of the above-mentioned participants extracts the key parameters of the trained model (such as the weight matrix and bias vector in the neural network), and uploads these abstracted key parameters that do not contain original data information to the server through a secure communication channel (in this process, to ensure the security of transmission, encrypted transmission protocols such as SSL / TLS can be used); After receiving the key parameters uploaded by each participant, the server assigns corresponding weights according to the preset aggregation algorithm (such as weighted average method), comprehensively considers the data volume and data quality of each participant, aggregates and integrates all key parameters, and updates the global model (this process is only a mathematical operation of abstract parameters, and it is impossible to reversely deduce the original data of each participant); Transmitting the updated new parameters of the global model back to each of the participants through a secure channel; Each of the participants uses the received new parameters to replace the original corresponding parts in the local model, completes the update of the local model, and obtains the fourth-level data processing model of each party; Based on the updated fourth-level data processing model, each of the participants continues to use new data for training, upload parameters, receive new global models, and continuously optimizes model performance, so that the fourth-level data processing model's processing capabilities for fourth-level data gradually improve.

[0055] In the scheme of this embodiment, the original fourth-level data does not leave the local environment throughout the process, avoiding the risk of privacy leakage caused by data sharing. All participants can jointly improve the model effect and protect the privacy rights and interests of the data owner. By aggregating the data features of multiple parties, compared with a single participant relying only on its own limited data training model, the global model integrates richer information and can capture more comprehensive laws and trends, and the accuracy and robustness of data prediction, classification and other processing are significantly enhanced; each participant does not need to spend a lot of resources alone to collect massive data to train high-order models. Through the federated learning method of sharing model parameters, the cost of data collection, storage and processing is reduced, and the overall business efficiency is improved; as long as the new participant holds the same type of fourth-level data, it can easily join the federated learning system, quickly integrate and contribute its own data features, promote the continuous evolution of the model, and adapt to the dynamic development of the business.

[0056] In some possible implementations of the present invention, the step of the edge computing chip processing the second data according to different sensitivity levels to obtain third data includes: Extract the fifth-level data (i.e., the data with the sensitivity level marked as the fifth level) from the second data: Performing basic encryption on the fifth level data; Data classification and aggregation are performed on the encrypted data, and general access control is established on the classified and aggregated data to obtain third data.

[0057] In an embodiment of the present invention, the second level of data is processed in combination with zero-knowledge proof technology and other technologies; the third level of data is processed using differential encryption technology; the fourth level of data is processed using federated learning technology; and the fifth level of data is processed using ordinary encryption technology.

[0058] See also Figure 2 Another embodiment of the present invention provides an artificial intelligence-based smart underwear data processing system for executing an artificial intelligence-based smart underwear data processing method, including: a server, a smart underwear provided with a smart sensor and an edge computing chip; wherein, The smart sensor is configured to: collect physiological data and environmental perception data of the user, wherein the collection frequency is set to different preset values ​​according to the sensor type, and the collected data is time-stamped; The edge computing chip is configured as: After preprocessing the collected physiological data and environmental perception data, first data is obtained; Performing sensitivity classification and labeling on the first data to obtain second data; Processing the second data according to different sensitivity levels to obtain third data; Determining fourth data to be uploaded to the server from the third data; The server is configured to: Decrypting the fourth data after receiving it, synchronously integrating the multi-source data according to the timestamp, and constructing a user health data set; Inputting the user health data set into a preset health analysis model to obtain the user's health analysis result, and sending the health analysis result to the smart underwear; In the entire data processing process, data access logs are recorded with the help of blockchain technology, and each data access generates an unalterable record; The edge computing chip is further configured to: provide personalized health advice and reminders to the user through the feedback device of the smart underwear according to the health analysis results; at the same time, collect interactive feedback from the user on the feedback information, correct the parameters of the artificial intelligence model in real time, and optimize the subsequent health analysis results; The server is also configured to: utilize digital twin technology to drive a virtual digital model with user physiological data, simulate the user's physical responses in different health states and sports scenarios, predict potential health problems, and develop a health management plan.

[0059] It should be known that Figure 2 The block diagram of the intelligent underwear data processing system based on artificial intelligence is only for illustration, and the number of modules shown does not limit the protection scope of the present invention. The intelligent underwear data processing system based on artificial intelligence provided in this embodiment can be used to execute the corresponding embodiments of the intelligent underwear data processing method based on artificial intelligence. For the specific implementation process, please refer to the description of the embodiments of each method, which will not be repeated here.

[0060] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0061] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0062] In the several embodiments provided in the present application, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are only schematic, such as the division of the above-mentioned units, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0063] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0064] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0065] If the above-mentioned integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a memory, including a number of instructions to enable a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the above-mentioned methods of each embodiment of the present application. The aforementioned memory includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or CD-ROM and other media that can store program codes.

[0066] A person skilled in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable memory, and the memory can include: a flash drive, a read-only memory (English: Read-Only Memory, abbreviated as: ROM), a random access memory (English: Random Access Memory, abbreviated as: RAM), a magnetic disk or an optical disk, etc.

[0067] The embodiments of the present application are introduced in detail above. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method and core idea of ​​the present application. At the same time, for general technical personnel in this field, according to the idea of ​​the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

[0068] Although the present invention is disclosed as above, the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions without departing from the spirit and scope of the present invention, and can make various changes and modifications, including the combination of the above-mentioned different functions and implementation steps, including software and hardware implementation methods, all of which are within the scope of protection of the present invention.

Claims

1. A method for processing intelligent underwear data based on artificial intelligence, characterized in that: include: The user's physiological data and environmental perception data are collected using smart sensors installed in the smart underwear, wherein the collection frequency is set to different preset values ​​according to the sensor type, and the collected data is time-stamped; After preprocessing the collected physiological data and environmental perception data in the edge computing chip of the smart underwear, first data is obtained; The edge computing chip performs sensitivity classification and labeling on the first data to obtain second data; The edge computing chip processes the second data according to different sensitivity levels to obtain third data; The edge computing chip determines fourth data to be uploaded to the server from the third data; The server decrypts the fourth data after receiving it, and synchronously integrates the multi-source data according to the timestamp to construct a user health data set; The server inputs the user health data set into a preset health analysis model to obtain the user's health analysis result, and sends the health analysis result to the smart underwear; In the entire data processing process, data access logs are recorded with the help of blockchain technology, and each data access generates an unalterable record; The edge computing chip provides personalized health advice and reminders to the user through the feedback device of the smart underwear based on the health analysis results; at the same time, collects the user's interactive feedback on the feedback information, corrects the parameters of the artificial intelligence model in real time, and optimizes the subsequent health analysis results; The server uses digital twin technology to drive a virtual digital model with user physiological data, simulates the user's physical responses in different health states and sports scenarios, predicts potential health problems, and develops a health management plan.

2. The method for processing intelligent underwear data based on artificial intelligence according to claim 1, characterized in that: The step of the edge computing chip performing sensitivity classification and labeling on the first data to obtain second data includes: extracting temporal features, spatial features, and semantic features of the first data; Using a preset feature vector extraction algorithm, different types of sensor data are converted into standardized feature vectors; Construct a multimodal feature fusion matrix to fuse data features of different dimensions; Analyze the feature fusion matrix based on the preset sensitivity assessment model; Calculate the privacy exposure risk value of the data, which takes into account the identifiability, relevance and harmfulness of the data; According to the corresponding relationship between the risk value and the preset sensitivity level, the sensitivity level of the data is determined and marked, thereby obtaining the second data.

3. The method for processing intelligent underwear data based on artificial intelligence according to claim 2, characterized in that: The step of the edge computing chip processing the second data according to different sensitivity levels to obtain third data includes: extracting first level data from the second data; The server sends a first processing program to the edge computing chip; After the edge computing chip receives the first processing program and processes the first level data using the first processing program, it sends first feedback data to the server; The server receives the first feedback data and verifies the first feedback data; If the verification is passed, a first processing program failure instruction is sent; If the verification fails, adjusting the first processing procedure based on the first feedback data to obtain a second processing procedure; Sending an invalidation instruction of the second processing program and the first processing program to the edge computing chip; The edge computing chip receives the second processing program and processes the first level data using the second processing program; Repeat the above steps until the feedback data sent by the edge computing chip passes the verification.

4. The method for processing intelligent underwear data based on artificial intelligence according to claim 3, characterized in that: The step of determining, by the edge computing chip, fourth data uploaded to the server from the third data includes: Determining the service processing type and service processing content of the server; Detecting the data duplication degree of the third data, performing data downsampling, retaining key time node data, and obtaining intermediate data with timestamps; Performing data value evaluation on the intermediate data, including: calculating the information entropy value of the intermediate data; evaluating the contribution of the intermediate data to the business processing type and business processing content of the server; generating a data value score for the intermediate data; and establishing a data value priority queue for the intermediate data according to the data value score; Determine the upload priority threshold based on the current network bandwidth status, server load, and data transmission cost; Segmenting the intermediate data in combination with the data value priority queue, extracting data feature vectors, performing data compression, and generating multiple compressed data packets; The compressed data packets are prioritized based on the preset rule screening data and the upload priority threshold to determine an upload data set, thereby obtaining fourth data.

5. The method for processing intelligent underwear data based on artificial intelligence according to claim 4, characterized in that: The server decrypts the fourth data after receiving it, and synchronously integrates the multi-source data according to the timestamp to construct the user health data set, including: Receiving and verifying the fourth data, including: verifying the integrity of data transmission: checking the signature of the data packet, checking the integrity of the data packet sequence, and verifying the validity of the timestamp; identifying the source and type of data; parsing the format of the data packet; and generating a receipt confirmation message; Decrypting the fourth data includes: obtaining a decryption key from a security key management system; performing decryption operations on different encryption types: decrypting homomorphically encrypted data, processing differential privacy data, and restoring desensitized data; verifying the correctness of the decryption result; and recording a decryption operation log; The steps of aligning the decrypted data timestamp include: extracting the data timestamp information; unifying the time standard: converting timestamps in different time zones, calibrating the device clock deviation, and processing the sampling time error; establishing the time index structure; and identifying the data timing relationship; The steps of multi-source data synchronization after timestamp alignment include: identifying the association relationship between data; performing time window division: setting the synchronization time window size, handling data sampling frequency differences, and aligning data sampling points; filling missing data: applying interpolation algorithms, marking filling data, and evaluating filling reliability; handling data conflicts; The steps of quality assessment of synchronized data include: detecting outliers: statistical feature analysis, time series pattern recognition, marking abnormal data points; assessing data consistency; calculating data reliability scores; generating quality assessment reports; The steps of data integration after synchronization include: building a unified data model: defining data field mapping, unifying measurement units, and standardizing data formats; merging multi-source data; establishing data association indexes; and optimizing storage structures; Metadata management steps include: recording data processing; updating data catalogs; generating data summaries; The steps of constructing a user health data set include: organizing a hierarchical data structure; establishing a fast retrieval mechanism; implementing an incremental update strategy; and optimizing access performance.

6. The method for processing intelligent underwear data based on artificial intelligence according to claim 5, characterized in that: The server inputs the user health data set into a preset health analysis model to obtain the user's health analysis result, and sends the health analysis result to the smart underwear, including: Preprocessing the user health data set includes: performing data standardization processing: normalizing numerical features, encoding category features, and processing time series features; constructing feature vectors: extracting statistical features, generating time series features, and calculating derived features; processing missing values ​​and outliers; and generating an analysis data set; Model selection and configuration, including: loading the following preset health analysis models according to the analysis task: disease risk prediction model, fatigue assessment model, health status classification model; configuring the model parameters of the loaded model: setting the inference threshold, adjusting the model weight, optimizing the prediction parameters; initializing the operating environment of the loaded model; verifying the model status of the loaded model; Performing health risk assessment according to the disease risk prediction model includes: executing disease risk prediction: analyzing abnormal physiological indicators, evaluating health trend changes, and calculating risk probability distribution; generating risk rating: determining risk level, identifying key risk factors, and calculating risk confidence; generating early warning information; and recording the assessment process; Fatigue analysis is performed according to the fatigue assessment model, including: analyzing physiological data patterns: assessing heart rate variability, analyzing activity intensity, and monitoring recovery status; calculating fatigue index: integrating multi-dimensional indicators, applying fatigue assessment algorithms, and generating fatigue levels; assessing recovery suggestions; and recording analysis results; Comprehensively process the above risk assessment results and fatigue analysis results according to the health status classification model, including: integrating multi-model prediction results: weighted fusion prediction values, eliminating prediction conflicts, and optimizing result reliability; generating health reports: summarizing key findings, providing detailed explanations, and formulating recommended solutions; marking important reminders; and storing comprehensive analysis results; Generate personalized suggestions based on the comprehensive analysis results, including: customize suggestions based on user portraits: consider user health history, analyze living habits, and evaluate exercise ability; formulate improvement plans: generate action suggestions, set health goals, and plan intervention measures; generate reminder strategies; and optimize suggestion content; Generate a health analysis result based on the comprehensive analysis result and the personalized suggestion; The steps for transmitting health analysis results include: packaging analysis results: compressing data packets, adding security identifiers, generating verification codes; establishing a secure channel; performing batch transmission; confirming that the transmission is complete; Feedback tracking steps include: monitoring the implementation of recommendations; collecting user feedback; updating personalized parameters; and optimizing health analysis models.

7. The method for processing intelligent underwear data based on artificial intelligence according to claim 6, characterized in that: The step of the edge computing chip processing the second data according to different sensitivity levels to obtain third data further includes: extracting second level data from the second data; dividing the second level data into first data to be processed, second data to be processed and third data to be processed; Initial TEE environment, verify the integrity of the environment, and generate a zero-knowledge proof key pair within the TEE; Extracting features from the first data to be processed and constructing a circuit required for proof; Generating a zero-knowledge proof includes: converting the first to-be-processed data into a polynomial expression; generating proof parameters using an elliptic curve cryptography algorithm; constructing a commitment scheme to ensure data correctness; generating a non-interactive zero-knowledge proof; and verifying the validity of the generated zero-knowledge proof; Generate homomorphic encryption keys based on lattice cryptography; Splitting the second to-be-processed data into data blocks suitable for homomorphic operations; Perform homomorphic encryption on each data block to generate first encrypted data that supports ciphertext domain operations, and record encryption parameters for subsequent decryption; Determine noise parameters based on Laplace mechanism; According to the noise parameter, adding differential privacy noise to the third data to be processed to obtain second encrypted data; Verify the data availability after adding differential privacy noise, and dynamically adjust the noise parameters to balance privacy protection and data availability; Integrate the zero-knowledge proof, the first encrypted data, and the second encrypted data generated above to obtain third data; Classify and store the third data, establish a data index structure, record the metadata of the data processing process, and generate a data processing report; Verify the integrity of third-party data, check privacy protection effects, evaluate data availability, generate processing records and upload them to the blockchain.

8. The method for processing intelligent underwear data based on artificial intelligence according to claim 7, characterized in that: The step of the edge computing chip processing the second data according to different sensitivity levels to obtain third data further includes: extracting third level data from the second data; Determine the privacy budget parameters based on the characteristics of the third-level data and the privacy protection requirements; According to the data type of the third level data, select the corresponding noise distribution; Carry out preliminary cleaning of the third-level data to remove erroneous values, outliers, and irrelevant redundant information; Generate the corresponding random noise using a random number generator based on the selected noise distribution and privacy budget parameters; The generated random noise is added to the preprocessed third level data to obtain third data.

9. The method for processing intelligent underwear data based on artificial intelligence according to claim 8, characterized in that: The step of the edge computing chip processing the second data according to different sensitivity levels to obtain third data further includes: extracting fourth level data from the second data; Processing the fourth-level data using a preset fourth-level data processing model to obtain third data; The steps of constructing the fourth level data processing model include: Determine a plurality of participants holding fourth-level training data of the fourth level, and pre-process the fourth-level training data in the local edge computing devices or data centers of the participants; wherein the data of each participant is always retained locally, and the original data is not transmitted externally; Each of the above-mentioned participants independently selects the corresponding machine learning model architecture based on its own data characteristics and business needs, and randomly initializes the model parameters to prepare for local training; Using the local fourth-level training data, each participant starts to train the initialized model, using an optimization algorithm to continuously update the model parameters in multiple rounds of iterations, so that the model gradually fits the characteristics and rules of the local data; After a preset round of local training, each of the above-mentioned participants extracts the key parameters of the trained model, and uploads these abstracted key parameters without original data information to the server through a secure communication channel; After receiving the key parameters uploaded by each participant, the server assigns corresponding weights according to the preset aggregation algorithm, comprehensively considers factors such as the data volume and data quality of each participant, aggregates and integrates all key parameters, and updates the global model; Transmitting the updated new parameters of the global model back to each of the participants through a secure channel; Each of the participants uses the received new parameters to replace the original corresponding parts in the local model, completes the update of the local model, and obtains the fourth-level data processing model of each party; Based on the updated fourth-level data processing model, each of the participants continues to use new data for training, upload parameters, receive new global models, and continuously optimizes model performance, so that the fourth-level data processing model's processing capabilities for fourth-level data gradually improve.

10. An artificial intelligence-based smart underwear data processing system, used to execute the artificial intelligence-based smart underwear data processing method according to any one of claims 1 to 9, characterized in that: include: Server, smart underwear equipped with smart sensors and edge computing chips; The smart sensor is configured to: collect physiological data and environmental perception data of the user, wherein the collection frequency is set to different preset values ​​according to the sensor type, and the collected data is time-stamped; The edge computing chip is configured as: After preprocessing the collected physiological data and environmental perception data, first data is obtained; Performing sensitivity classification and labeling on the first data to obtain second data; Processing the second data according to different sensitivity levels to obtain third data; Determining fourth data to be uploaded to the server from the third data; The server is configured to: Decrypting the fourth data after receiving it, synchronously integrating the multi-source data according to the timestamp, and constructing a user health data set; Inputting the user health data set into a preset health analysis model to obtain the user's health analysis result, and sending the health analysis result to the smart underwear; In the entire data processing process, data access logs are recorded with the help of blockchain technology, and each data access generates an unalterable record; The edge computing chip is further configured to: provide personalized health advice and reminders to the user through the feedback device of the smart underwear based on the health analysis results; at the same time, collect interactive feedback from the user on the feedback information, correct the parameters of the artificial intelligence model in real time, and optimize subsequent health analysis results; The server is also configured to: utilize digital twin technology to drive a virtual digital model with user physiological data, simulate the user's physical responses in different health states and sports scenarios, predict potential health problems, and develop a health management plan.

Images