A medical data collection method based on blockchain

Abstract

The present application relates to the field of medical data management, and provides a medical data collection method based on blockchain, which first registers a personal account through the personal basic information uploaded by the user, then verifies and verifies when the user logs in to determine whether a secondary SMS verification is required, then collects the user's medical data through a sensor group, and uses a public key to encrypt the collected medical data in pieces, and then uses blockchain technology to distribute and store it in various storage nodes, then extracts the encrypted pieces and decrypts them to obtain the decrypted medical data, and then analyzes them to generate an analysis report and display it. In this way, the safe management and effective application of user medical data can be achieved.

Description

Technical Field

[0001] The present application relates to the field of medical data management, and more specifically, to a medical data collection method based on blockchain. Background Art

[0002] With the rapid development of medical information technology, the security and privacy protection of medical data have become key issues that need to be addressed. Traditional centralized storage methods are susceptible to single point failures, data tampering, and privacy leakage, and are difficult to meet the high requirements of modern medical services for data security. At the same time, with the advancement of Internet of Things technology, more and more smart devices are used to collect personal health data, and the large amount of data generated by these devices requires an efficient and secure management mechanism.

[0003] Patent CN112509655B proposes a blockchain-based intelligent medical data collection system. Specifically, after users upload their personal information and register through mobile phone verification, they obtain a pair of public and private keys to ensure communication security. The system verifies user information when logging in. Subsequently, the smart device collects medical data in real time, and these data are stored in a decentralized manner after being fragmented and encrypted. Users can use their own private keys to decrypt these data fragments and verify the authenticity of the data through public keys. Finally, the system will intelligently analyze these data and generate a detailed health report for users to review.

[0004] When verifying the logged-in personal account, the above patent uses the difference between the average value of historical password input time and the actual verification time to perform password verification and determine whether to trigger mobile phone SMS verification. However, this method assumes that the time difference of all password inputs is only determined by the user's behavioral habits, and ignores the changes in input time that may be caused by different password combinations (such as complex characters, uppercase and lowercase conversions). For example, complex password combinations usually take longer to enter, while simple passwords are faster. In addition, the analysis is performed only by calculating the average value of all historical input times, ignoring the uniqueness of each historical input time and its importance in different contexts. This approach may cause recent and more representative behavior patterns to be diluted by earlier historical data, thereby reducing the accuracy and adaptability of the model.

[0005] Therefore, an optimized blockchain-based medical data collection solution is desired. Summary of the invention

[0006] In view of the shortcomings of the prior art, this application provides a medical data collection method based on blockchain.

[0007] According to one aspect of the present application, a blockchain-based medical data collection method is provided, comprising:

[0008] Register a personal account based on the personal basic information uploaded by the user, wherein the personal account includes personal basic information, account password, public key and private key;

[0009] After the user inputs the personal account for login, the personal account for login is verified, including: embedding and quickly scanning the semantic query encoding of the personal account for login to obtain password verification query response encoding features; based on the password verification query response encoding features, a determination result is obtained, and the determination result is used to indicate whether SMS verification is required through a mobile phone number;

[0010] Collecting multiple medical data of the user through a sensor group, the medical data including blood pressure data, heart rate data, pulse data, body temperature data and movement information of the user during sleep;

[0011] After the public key is used to perform sharding encryption on the user's medical data, the obtained sharded medical data of the user is encrypted and distributedly stored in the storage nodes through blockchain technology;

[0012] Extracting the encrypted slices of each user's medical data in the storage node, and decrypting the encrypted slices of each user's medical data to obtain decrypted medical data;

[0013] The decrypted medical data is analyzed to obtain an analysis report, and the analysis report is displayed.

[0014] Furthermore, the logged-in personal account is embedded and quickly scanned for semantic query encoding to obtain password verification query response encoding features, including:

[0015] Acquire multiple historical login account passwords continuously input by the user, and record the input time of each of the historical login account passwords to obtain multiple password input times;

[0016] Adding each password input time of the plurality of password input times to the end of the corresponding historical login account password to obtain a set of login account password data containing password input time information;

[0017] respectively performing embedded coding and semantic association analysis on the set of login account password data containing password input time information to obtain a set of login account password data association feature vectors;

[0018] Obtain the login account password to be verified and the time when the password to be verified was entered by the user during the login process;

[0019] After adding the password input time to be verified to the end of the login account password to be verified, embedding coding is performed on the obtained password data to obtain an embedded coding vector of the password data to be verified;

[0020] The fast semantic response query encoding is performed on the set of the embedded coding vector of the password data to be verified and the associated feature vector of the login account password data to obtain a password verification query response coding vector as the password verification query response coding feature.

[0021] Furthermore, embedding coding and semantic association analysis are performed on the set of login account password data containing password input time information to obtain a set of login account password data association feature vectors, including:

[0022] Using the password embedding matrix, each of the login account password data containing the password input time information in the set of login account password data containing the password input time information is embedded and encoded to obtain a set of login account password data embedded encoding vectors;

[0023] The set of the login account password data embedded encoding vectors is passed through a password semantic association analyzer based on a stacked LSTM to obtain a set of login account password data associated feature vectors.

[0024] Further, after adding the password input time to be verified to the end of the login account password to be verified, the obtained password data to be verified is embedded and encoded to obtain an embedded encoding vector of the password data to be verified, including:

[0025] Add the password input time to be verified to the end of the login account password to be verified to obtain the password data to be verified;

[0026] The password embedding matrix is ​​used to embed the password data to be verified to obtain an embedding coding vector of the password data to be verified.

[0027] Further, performing fast semantic response query encoding on the set of the embedded encoding vector of the password data to be verified and the associated feature vector of the login account password data to obtain the password verification query response encoding vector includes:

[0028] Calculate the mutual information between the embedded coding vector of the password data to be verified and each login account password data associated feature vector in the set of login account password data associated feature vectors to obtain a set of semantic fast semantic matching factors of the password data to be verified-login account password;

[0029] Based on the set of to-be-verified-login account password semantics fast semantic matching factors, determining a subset of fast-matched login account password data associated feature vectors;

[0030] Semantic query encoding is performed on a subset of the embedded encoding vector of the password data to be verified and the associated feature vector of the login account password data to obtain the password verification query response encoding vector.

[0031] Further, based on the set of the to-be-verified-login account password semantics fast semantic matching factors, a subset of fast-matched login account password data associated feature vectors is determined, including:

[0032] Identify the first to-be-verified-login account password semantic maximum value and the second to-be-verified-login account password semantic maximum value from the set of to-be-verified-login account password semantic fast semantic matching factors;

[0033] Based on the positions of the first to-be-verified-login account password semantic maximum value and the second to-be-verified-login account password semantic maximum value in the set of to-be-verified-login account password semantic fast semantic matching factors, a subset of the login account password data associated feature vectors for fast matching is determined from the set of login account password data associated feature vectors.

[0034] Furthermore, semantic query encoding is performed on a subset of the embedded coding vector of the password data to be verified and the feature vector associated with the login account password data to obtain the password verification query response coding vector, including: linearly transforming the embedded coding vector of the password data to be verified to obtain a query vector of the password data to be verified as a query vector and a value vector of the password data to be verified as a value vector, and taking a subset of the feature vector associated with the login account password data as a subset of the key vector, and performing cross-domain query encoding based on a converter structure on the query vector, the value vector and a subset of the key vector to obtain the password verification query response coding vector.

[0035] Furthermore, based on the password verification query response coding feature, a discrimination result is obtained, including: passing the password verification query response coding vector through a classifier-based SMS verification discriminator to obtain the discrimination result.

[0036] Furthermore, after the public key is used to perform shard encryption processing on the various medical data of the user, the obtained encrypted shards of the various medical data of the user are distributedly stored in the storage nodes through the blockchain technology, including:

[0037] Slicing each item of medical data of the user to obtain multiple medical data slices of the user;

[0038] Using the public key and asymmetric encryption technology, encrypt each of the multiple medical data slices to obtain encrypted slices of each medical data of the user;

[0039] The blockchain technology is used to encrypt and fragment the user's medical data and store them in storage nodes at different locations.

[0040] Further, extracting the encrypted slices of each user's medical data in the storage node, and decrypting the encrypted slices of each user's medical data to obtain decrypted medical data, including:

[0041] Extracting encrypted slices of each item of medical data of the user from each storage node;

[0042] Decrypting the encrypted slices of each medical data of the user using the private key to obtain a first medical data set;

[0043] Using the public key to verify the encrypted slices of each medical data of the user to obtain a second medical data set;

[0044] The contents of the first medical data and the second medical data corresponding to each group in the first medical data set and the second medical data set are compared and analyzed to perform medical data verification to obtain the decrypted medical data, wherein if the contents of the first medical data and the second medical data are consistent, it is determined that the medical data is complete and authentic; if the contents of the first medical data and the second medical data are inconsistent, it is determined that the medical data is distorted.

[0045] This application has significant technical effects due to the adoption of the above technical solutions:

[0046] The blockchain-based medical data collection method provided by the present application first registers a personal account through the personal basic information uploaded by the user, then verifies and verifies when the user logs in to determine whether a secondary SMS verification is required, then collects the user's medical data through a sensor group, and uses the public key to encrypt the collected medical data in slices, and then uses blockchain technology to distribute and store it in various storage nodes, then extracts the encrypted slices and decrypts them to obtain the decrypted medical data, and then analyzes them to generate an analysis report and display it. In this way, the safe management and effective application of user medical data can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0048] Figure 1This is a flowchart of a blockchain-based medical data collection method according to an embodiment of the present application.

[0049] Figure 2 This is a flowchart of step S120 in the blockchain-based medical data collection method according to an embodiment of the present application.

[0050] Figure 3 This is a flowchart of step S121 in the blockchain-based medical data collection method according to an embodiment of the present application.

[0051] Figure 4 This is a flowchart of step S1216 in the blockchain-based medical data collection method according to an embodiment of the present application.

[0052] Figure 5 This is a flowchart of step S140 in the blockchain-based medical data collection method according to an embodiment of the present application.

[0053] Figure 6 This is a flowchart of step S150 in the blockchain-based medical data collection method according to an embodiment of the present application. DETAILED DESCRIPTION

[0054] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0055] Based on the problems in the above background technology, this application provides a medical data collection method based on blockchain. Figure 1 Flowchart of a blockchain-based medical data collection method according to an embodiment of the present application. Figure 1As shown, according to the embodiment of the present application, the medical data collection method based on blockchain includes: S110, registering a personal account according to the personal basic information uploaded by the user, wherein the personal account includes personal basic information, account password, public key and private key; S120, after the user enters the personal account for login, verifying the logged-in personal account; S130, collecting multiple medical data of the user through a sensor group, the medical data including blood pressure data, heart rate data, pulse data, human body temperature data and human motion information during sleep; S140, after using the public key to perform sharding encryption processing on the various medical data of the user, the obtained encrypted shards of the various medical data of the user are distributedly stored in the storage node through blockchain technology; S150, extracting the encrypted shards of the various medical data of the user in the storage node, and decrypting the encrypted shards of the various medical data of the user to obtain the decrypted medical data; S160, analyzing the decrypted medical data to obtain an analysis report, and displaying the analysis report.

[0056] In step S110, a personal account is registered according to the basic personal information uploaded by the user, wherein the personal account includes basic personal information, account password, public key and private key. It should be understood that by collecting the basic personal information uploaded by the user, such as name, age, gender, contact information, etc., the system builds a preliminary understanding of the user, laying the foundation for subsequent personalized services and precision medical data management. These basic information helps to accurately distinguish different users in the huge medical data system, ensuring that the data of each user is independent and traceable. The setting of account passwords is the first line of defense for users to access personal medical data, ensuring that only authorized users can log in to the system to view and manage their own data. At the same time, the generation of public and private keys is based on cryptographic principles, which provides a guarantee for the communication security of users during data transmission and storage. The public key can be publicly used to encrypt data, and only users with the corresponding private key can decrypt it, ensuring the confidentiality, integrity and non-repudiation of the data during network transmission and storage. For example, in the process of transmitting medical data from smart devices to distributed storage nodes, the data will be encrypted by the public key, and only the user's own private key can decrypt and view it, preventing the data from being stolen or tampered with midway. In general, registering a personal account based on personal information uploaded by users is the basis and key link of the blockchain-based medical data collection method, which is conducive to realizing personalized medical customized services.

[0057] In step S120, after the user inputs the personal account for logging in, the personal account for logging in is verified.

[0058] Correspondingly, after the user inputs the personal account for login, the personal account for login is verified. The technical concept of the present application is to obtain multiple historical login account passwords continuously input by the user, and record the input time of each historical login account password, and at the same time obtain the login account password to be verified and the input time of the password to be verified input by the user during the login process, and use AI-based data processing and encoding methods to add each input time to the tail of the corresponding historical login account password and embed the coding and semantic association of each added data, and similarly add the input time of the password to be verified to the tail of the login account password to be verified and embed the coding, so as to intelligently judge whether SMS verification is required through the mobile phone number according to the query response between the embedded features of the password data to be verified and the associated features of each login account password data. In this way, not only the time when the user enters the password is considered, but also the influence of different password combinations on the input time is paid special attention to, so as to better capture the additional input delay caused by the complex password combination, so as to more accurately reflect the actual behavior habits of the user. In addition, the weight of each historical input can be dynamically adjusted, which helps to better adapt to the changes in user behavior and avoid the dilution of the latest behavior pattern by early data.

[0059] Specifically, Figure 2 FIG. 1 is a flowchart of step S120 in the blockchain-based medical data collection method according to an embodiment of the present application. Figure 2 As shown, the step S120 includes: S121, embedding encoding and quick scanning semantic query encoding of the logged-in personal account to obtain password verification query response encoding characteristics; S122, based on the password verification query response encoding characteristics, obtaining a judgment result, and the judgment result is used to indicate whether SMS verification is required via a mobile phone number.

[0060] More specifically, Figure 3 FIG. 1 is a flowchart of step S121 in the blockchain-based medical data collection method according to an embodiment of the present application. Figure 3As shown, the step S121 includes: S1211, obtaining multiple historical login account passwords continuously input by the user, and recording the input time of each of the historical login account passwords to obtain multiple password input times; S1212, adding each password input time of the multiple password input times to the tail of the historical login account password corresponding thereto to obtain a set of login account password data containing password input time information; S1213, respectively performing embedded coding and semantic association analysis on the set of login account password data containing password input time information to obtain a set of login account password data associated feature vectors; S1214, obtaining the login account password to be verified and the password input time to be verified input by the user during the login process; S1215, after adding the password input time to be verified to the tail of the login account password to be verified, embedding coding is performed on the obtained password data to be verified to obtain an embedded coding vector of the password data to be verified; S1216, performing fast semantic response query coding on the set of the embedded coding vector of the password data to be verified and the associated feature vector of the login account password data to obtain a password verification query response coding vector as the password verification query response coding feature.

[0061] In step S1211, multiple historical login account passwords continuously input by the user are obtained, and the input time of each of the historical login account passwords is recorded to obtain multiple password input times. It should be understood that by obtaining multiple continuous historical login account passwords and corresponding input times, the behavioral data patterns of the user's login operations under normal circumstances can be accumulated. These historical data can reflect the user's login habits at different times and in different situations, including input speed, possible password memory methods (such as whether certain parts are input regularly), etc., thereby providing a benchmark reference for subsequent verification analysis, so as to accurately determine whether the current login behavior is consistent with the user's normal pattern. And by recording each historical password input time, it is helpful to capture the dynamic characteristics of the user when entering the password, such as whether there is a time fluctuation caused by thinking, recalling or input errors. This is not just a simple time record, but it can dig out the details of the user's behavior pattern from these dynamic changes, which in turn helps to accurately identify the user's identity and judge the login security. That is, the behavior pattern baseline formed based on multiple historical data points can more comprehensively and accurately evaluate the rationality of login behavior when verifying the current login, reduce misjudgments caused by accidental factors (such as slow input due to temporary distraction), improve the recognition accuracy of legitimate users, and reduce false alarm rates (legitimate users are misjudged as abnormal) and false negative rates (illegal users are misjudged as legitimate). And the continuously recorded historical data can be continuously updated and adapted to new behavior patterns. For example, if the user's input speed slows down for a period of time due to some reason (such as finger injury), it will not be immediately judged as abnormal, but through long-term historical data trend analysis to determine whether this change is within a reasonable range, thereby maintaining the dynamic adaptability of the verification process, ensuring that interference with the user's normal use is minimized, while maintaining sensitivity to abnormal login behavior.

[0062] In step S1212, each of the multiple password input times is added to the tail of the historical login account password corresponding thereto to obtain a set of login account password data containing password input time information. Accordingly, considering that the password input time expresses the specific time point at which the user enters the password each time, if multiple failed login attempts are detected in a short period of time, especially the time distribution of these attempts does not conform to the user's normal behavior pattern, the system can identify that this may be a brute force or automated attack. The historical login account password is used to verify the user's identity and ensure that only authorized users can access system resources. Therefore, in order to more accurately capture the user's login mode, to more comprehensively understand and analyze the user's login situation, and to improve the security of personal medical data, in the technical solution of the present application, each of the multiple password input times is added to the tail of the historical login account password corresponding thereto to obtain a set of login account password data containing password input time information.

[0063] In step S1213, embedding coding and semantic association analysis are performed on the set of login account password data containing password input time information to obtain a set of login account password data associated feature vectors. Specifically, in the embodiment of the present application, step S1213 includes: using a password embedding matrix to embedding coding each login account password data containing password input time information in the set of login account password data containing password input time information to obtain a set of login account password data embedded coding vectors; passing the set of login account password data embedded coding vectors through a password semantic association analyzer based on a stacked LSTM to obtain a set of login account password data associated feature vectors.

[0064] It should be understood that each login account password data containing password input time information contains key and important data information and content, and there are complex association patterns and relationships between these data. Based on this, in the technical solution of the present application, a password embedding matrix is ​​used to embed and encode each login account password data containing password input time information in the set of login account password data containing password input time information to obtain a set of login account password data embedded encoding vectors. That is to say, the password embedding matrix can effectively capture the complex relationship between the password and the timestamp by converting the original login account password data and its input time information into a numerical vector in a high-dimensional space, rather than just a simple character matching or time comparison. In particular, the password embedding matrix is ​​obtained by using a deep learning framework to train a neural network model so that the model can learn useful feature representations from a large amount of unlabeled data.

[0065] Next, since the user login behavior is essentially a process that changes over time, and there is a mutual correlation between the user login behaviors at different time points, therefore, in order to effectively capture and extract the user login information at different time points, in the technical solution of the present application, the set of the login account password data embedded in the encoding vector is passed through a password semantic association analyzer based on stacked LSTM to obtain a set of login account password data associated feature vectors. It can be understood that stacked LSTM refers to a deep neural network structure formed by stacking multiple LSTM layers in sequence. Each layer of LSTM receives the output of the previous layer as input and gradually extracts higher-level feature representations. This structure can capture complex patterns in the data more deeply. Therefore, by stacking multiple layers of LSTM, the system can abstract higher-level semantic information in the login account password data layer by layer, that is, each layer of LSTM can capture the pattern of login accounts on different time scales, and finally form a deep understanding of user login behavior.

[0066] In step S1214, the login account password to be verified and the time of password input are obtained. It should be understood that by obtaining the login account password to be verified and the time of password input are obtained, the user's current operation behavior data can be captured in real time, so as to compare and analyze with the historical behavior pattern. This is the most direct data basis for subsequent judgment of whether it conforms to the normal behavior pattern. Specifically, once the login information to be verified is obtained, the system can start the verification process to quickly and accurately determine whether there is a potential risk in this login. This instant startup verification can effectively prevent unauthorized access attempts, protect the user account security at the first time, prevent malicious attackers from using stolen passwords or account information obtained by other illegal means to log in to the system, and avoid the risk of user medical data leakage.

[0067] In step S1215, after adding the password input time to be verified to the tail of the login account password to be verified, the obtained password data to be verified is embedded and encoded to obtain the password data embedded encoding vector to be verified. Specifically, in the embodiment of the present application, step S1215 includes: adding the password input time to be verified to the tail of the login account password to be verified to obtain the password data to be verified; and using the password embedding matrix to embed the password data to obtain the password data embedded encoding vector to be verified.

[0068] It should be understood that in order to further capture and fully understand the behavior of the current user login account, in the technical solution of the present application, the input time of the password to be verified is added to the tail of the login account password to be verified. That is, by combining time information with the password, the user's behavior pattern can be captured more comprehensively. For example, the user's input speed, input time period, etc. can be added to the model as additional features to better understand the user's current behavior. Further, the password data to be verified is embedded and encoded using a password embedding matrix to better understand the semantic association between the login password and the input time, and an embedded encoding vector of the password data to be verified is obtained. In particular, the embedded encoded vector can help the system detect abnormal behavior more sensitively. For example, if a user's login time suddenly deviates from its historical pattern, the model can determine whether there is a risk by analyzing the distance of the embedded vector to decide whether an additional identity authentication step is required.

[0069] In step S1216, the set of the embedded coding vector of the password data to be verified and the associated feature vector of the login account password data is subjected to fast semantic response query coding to obtain the password verification query response coding vector as the password verification query response coding feature. In particular, the embedded coding vector of the password data to be verified reflects the semantic information of the account password entered by the user in this login attempt. The set of associated feature vectors of the login account password data comprehensively reflects the stable behavior pattern formed by the user in multiple logins in the past. The password data to be verified has different degrees of importance and association in each historical login account password data. Therefore, in order to complete the query comparison between the password data to be verified and the historical login data in a short time, and to more accurately capture the key differences between the password data to be verified and the historical pattern, in the technical solution of the present application, the set of the embedded coding vector of the password data to be verified and the associated feature vector of the login account password data is processed using fast semantic response query coding. In this way, the system can monitor and quickly evaluate the user's behavior in real time, identify potential abnormal activities, and dynamically adjust the security policy according to the risk assessment results. For example, when an abnormal login pattern is detected, the system can trigger additional security checks, such as SMS verification codes, to ensure account security.

[0070] Specifically, Figure 4 FIG. 1 is a flowchart of step S1216 in the blockchain-based medical data collection method according to an embodiment of the present application. Figure 4 As shown, the step S1216 includes: S12161, calculating the mutual information between the password data embedded coding vector to be verified and each login account password data associated feature vector in the set of feature vectors associated with the login account password data to obtain a set of fast semantic matching factors of the semantics of the to-be-verified-login account password; S12162, determining a subset of fast matching login account password data associated feature vectors based on the set of fast semantic matching factors of the semantics of the to-be-verified-login account password; S12163, performing semantic query encoding on the password data embedded coding vector to be verified and the subset of the login account password data associated feature vector to obtain the password verification query response coding vector.

[0071] Specifically, first, the mutual information between the embedded coding vector of the password data to be verified and each login account password data associated feature vector in the set of login account password data associated feature vectors is calculated to obtain a set of semantic fast semantic matching factors of the password data to be verified and the login account password. The above process can be expressed as:

[0072] ;

[0073] ;

[0074] ;

[0075] ;

[0076] in, is a set of feature vectors associated with the login account password data, , , and are the first, second, and third in the set of the associated feature vectors of the login account and password data. and Login account password data associated feature vector, is the embedded coding vector of the cryptographic data to be verified, yes function, is the feature vector of the embedded coding transformation of the password data to be verified, yes The corresponding login account password data association transformation feature vector, express The information entropy of express The information entropy of express and The joint entropy between express and The semantic quick semantic matching factor between the to-be-verified-login account password.

[0077] It should be understood that by calculating the mutual information between the embedded coding vector of the password data to be verified and the associated feature vectors of each login account password data, the semantic similarity between the embedded coding vector of the password data to be verified and the associated feature vectors of different login account password data can be evaluated. The larger the calculated mutual information value is, the more information overlaps between the two feature vectors, that is, the user's current login attempt behavior and a certain historical login behavior are highly correlated, that is, it can help identify the login account password data associated feature vector that is most relevant to the embedded coding vector of the password data to be verified. In addition, calculating the mutual information also provides a quantitative method to provide an objective standard for subsequent fast matching.

[0078] Specifically, in the embodiment of the present application, the step S12162 includes: identifying the first to-be-verified-login account password semantic maximum value and the second to-be-verified-login account password semantic maximum value from the set of the to-be-verified-login account password semantic fast semantic matching factors. The process can be expressed as:

[0079] ;

[0080] ;

[0081] in, It is a collection of quick semantic matching factors of the login account password to be verified. is the number of the to-be-verified-login account password semantics quick semantic matching factors in the set of to-be-verified-login account password semantics quick semantic matching factors, To get the maximum value in a set, It is the first to be verified-the maximum semantic value of the login account password. It is the second value to be verified - the maximum value of the semantics of the login account password;

[0082] Based on the positions of the first to-be-verified-login account password semantic maximum value and the second to-be-verified-login account password semantic maximum value in the set of to-be-verified-login account password semantic fast semantic matching factors, a subset of the login account password data associated feature vectors for fast matching is determined from the set of the login account password data associated feature vectors. This process can be expressed as:

[0083] ;

[0084] ;

[0085] ;

[0086] in, It is a collection of quick semantic matching factors of the login account password to be verified. To find the position of the maximum value in the set, It is the first to be verified-the maximum semantic value of the login account password. To first determine the quick match position, Determine the quick match position for the second, and Represents different login account password data associated feature vectors for determining quick matching locations, It is a subset of the feature vectors associated with the login account and password data.

[0087] That is, by identifying the first to-be-verified-login account password semantic maximum value and the second to-be-verified-login account password semantic maximum value from the set of the to-be-verified-login account password semantic fast semantic matching factors, two login account password data associated feature vectors with the most semantic relevance to the to-be-verified password data embedded coding vector can be screened out. Then, based on the positions of the first to-be-verified-login account password semantic maximum value and the second to-be-verified-login account password semantic maximum value in the set of the to-be-verified-login account password semantic fast semantic matching factors, a subset of the login account password data associated feature vectors that are quickly matched is determined from the set of the login account password data associated feature vectors. In this way, resources can be concentrated to process those login account password data associated feature vectors that are most likely to have interactive significance with the to-be-verified password data embedded coding vector, which is conducive to improving data processing efficiency and the quality of the results obtained. And the subset of login account password data associated feature vectors obtained by fast matching screening can reflect the model's judgment on the importance of historical login account password data, which helps to reduce noise interference and improve the generalization ability of the model.

[0088] Specifically, in the embodiment of the present application, the step S12163 includes: performing a linear transformation on the embedded coding vector of the password data to be verified to obtain a query vector of the password data to be verified as a query vector and a value vector of the password data to be verified as a value vector, and taking a subset of the feature vector associated with the login account password data as a subset of the key vector, performing cross-domain query coding based on the converter structure on the query vector, the value vector and the subset of the key vector to obtain the password verification query response coding vector. The above process can be expressed as:

[0089] ;

[0090] ;

[0091] ;

[0092] in, The first character in the set of associated feature vectors of the login account password data Login account password data associated feature vector, and are the query embedding matrix and the query bias vector respectively, is the query vector of the password data to be verified, and are the value embedding matrix and value bias vector respectively, is the password data value vector to be verified, yes The transposed vector of yes Length, yes function, To first determine the quick match position, Determine the quick match position for the second, is matrix multiplication, is the cryptographic verification query response encoding vector.

[0093] It should be understood that by utilizing the self-attention mechanism of the converter structure, when processing the collection data of the login account password data associated feature vector, it is possible to dynamically focus on the parts at different positions in the collection, which is conducive to improving the parallel processing capability of the data. The password data query vector to be verified obtained by calculation is mainly responsible for finding the login account password data associated feature vector that is most relevant to itself, while the password data value vector to be verified carries the specific information of these passwords to be verified. By processing the password data query vector to be verified, the password data value vector to be verified and the login account password data associated feature vector through the converter structure, it is possible to effectively capture the long-distance dependency relationship between the subset of the password data embedded encoding vector to be verified and the login account password data associated feature vector, and then generate an accurate password verification query response encoding vector.

[0094] In step S122, based on the coding features of the password verification query response, a discrimination result is obtained, and the discrimination result is used to indicate whether SMS verification is required through a mobile phone number. Specifically, in an embodiment of the present application, the step S122 includes: passing the password verification query response coding vector through a classifier-based SMS verification discriminator to obtain the discrimination result. It should be understood that after a series of complex coding processes, the password verification query response coding vector is finally obtained. At this time, the system still needs a clear decision-making mechanism to determine whether the current login situation is safe and reliable, and whether further SMS verification is required to enhance security. The classifier-based SMS verification discriminator provides such a data-driven decision-making function, which can comprehensively consider the feature information contained in the password verification query response coding vector, and through the judgment ability of the classifier, make a final qualitative evaluation of the login behavior to determine whether to trigger the SMS verification as an additional security measure. Here, the classifier is a machine learning model, and its core function is to classify the input data into predefined categories according to its characteristics. Specifically, it extracts key features and patterns from a large number of sample data of known categories (such as normal login behavior categories and abnormal login behavior categories) to build a mathematical model that can predict the category of new input data. During the training process, it adjusts its own parameters to minimize the classification error rate, thereby accurately classifying unknown data. The discrimination result directly determines whether the user needs to be verified by SMS. If the discrimination result shows that the current login behavior has certain risks (for example, it is judged as an abnormal login), the SMS verification is triggered to require the user to enter the mobile phone verification code, which is equivalent to adding an extra security line for the user account. The SMS verification code is usually sent to the user's pre-registered mobile phone number. Only the legitimate user of the mobile phone can obtain and enter the correct verification code, thereby effectively preventing illegal users from logging into the system by stealing passwords, protecting the user's medical data from being leaked, and ensuring that the medical data collection process is carried out in a safe and reliable environment.

[0095] In particular, considering that the embedded coding vector of the password data to be verified represents the embedded coding features of the login account password to be verified and the input time of the password to be verified, and the set of the login account password data association feature vectors represents the embedded coding short-range and long-range bidirectional temporal association features of multiple historical login account passwords and their input times, when performing semantic query encoding based on the fast scanning mechanism, the insufficient correspondence of the temporal scanning mechanism of the embedded coding features in different time domain spaces will also lead to the lack of instance determination of the query coding feature of the password verification query response coding vector, thereby affecting the accuracy of the discrimination result obtained by the classifier-based SMS verification discriminator.

[0096] Preferably, passing the password verification query response encoding vector through a classifier-based SMS verification discriminator to obtain a discrimination result comprises:

[0097] Calculate the distance between each pair of eigenvalues ​​of the password verification query response encoding vector, such as the L2 distance, and take the square root of the distance to obtain the password verification query response encoding distance representation matrix. This process can be expressed by the formula:

[0098] ;

[0099] in, represents the password verification query response encoding vector, and They represent the first and The eigenvalues ​​at the positions, express and The distance between express and The square root of the distance between

[0100] The process of obtaining the password verification query response encoding auto-association matrix of the password verification query response encoding vector can be expressed by the formula:

[0101] ;

[0102] in, represents the password verification query response encoding vector, and is a row vector, represents matrix multiplication, express The transposed vector of represents the password verification query response encoding autocorrelation matrix;

[0103] The password verification query response encoding vector is matrix-multiplied with the password verification query response encoding distance representation matrix to obtain the password verification query response encoding primary mapping vector. The process can be expressed by the formula:

[0104] ;

[0105] in, represents the password verification query response encoding vector, represents the password verification query response encoding distance representation matrix, represents matrix multiplication, Represents a first-level mapping vector for encoding a password verification query response;

[0106] The password verification query response code primary mapping vector is matrix-multiplied with the matrix product of the password verification query response code distance representation matrix and the password verification query response code self-association matrix to obtain the password verification query response code multi-level mapping vector. The process can be expressed by the formula:

[0107] ;

[0108] in, represents matrix multiplication, represents the first-level mapping vector of the password verification query response encoding, represents the password verification query response encoding distance representation matrix, represents the password verification query response encoding autocorrelation matrix, Represents a multi-level mapping vector encoding a password verification query response;

[0109] The password verification query response code multi-level mapping vector is interpolated with the password verification query response code associated eigenvector (if the eigenvalue is insufficient, interpolation or zero padding) composed of the eigenvalue of the password verification query response code auto-association matrix to obtain an optimized password verification query response code vector;

[0110] The optimized password verification query response encoding vector is passed through the classifier-based SMS verification discriminator to obtain the discrimination result.

[0111] That is, through the linear target mapping representation based on the similarity distance representation matrix of the password verification query response coding vector, the self-correlation complete similarity instantiation of the password verification query response coding vector is represented by a secondary target mapping representation based on a multi-level distribution hierarchy, and the association mismatch negative influence factor is compensated by the association fusion kernel bias to improve the eigenvalue instance determination degree of the password verification query response coding vector under similarity constraints, that is, the significance degree of the eigenvalue as an instance for classification regression determination, and improve the accuracy of the discrimination result of the password verification query response coding vector obtained by the classifier-based SMS verification discriminator.

[0112] In summary, step S120 is clearly explained, which uses AI-based data processing and encoding to add the input time of each user's historical login account password to the end of the corresponding historical login account password and embed the added data and semantically associate them. Similarly, the input time of the password to be verified entered during the user login process is added to the end of the login account password to be verified and embedded and encoded, so as to intelligently determine whether SMS verification is required through the mobile phone number based on the query response between the embedded features of the password data to be verified and the associated features of each login account password data. In this way, not only the time when the user enters the password is considered, but also the impact of different password combinations on the input time is paid special attention to, so as to better capture the additional input delay caused by complex password combinations, thereby more accurately reflecting the user's actual behavior habits. In addition, the weight of each historical input can be dynamically adjusted, which helps to better adapt to changes in user behavior and avoid the dilution of the latest behavior patterns by early data.

[0113] In step S130, a plurality of medical data of the user is collected through a sensor group, and the medical data includes blood pressure data, heart rate data, pulse data, human body temperature data and human motion information during sleep. It should be understood that the use of a variety of sensors to collect a plurality of medical data including blood pressure, heart rate, pulse, body temperature and sleep motion information is intended to monitor the health status of the user in real time and continuously from multiple dimensions. Different physiological indicators can reflect the functional status of different systems of the user's body. For example, blood pressure data can reflect the working condition of the cardiovascular system, heart rate and pulse data can provide information about heart function and blood circulation, body temperature data can help understand the body's metabolism and immune status, and motion information during sleep can assist in evaluating the user's sleep quality and potential sleep-related health problems. By combining these data, a relatively complete user health portrait can be constructed, which helps to promptly discover potential health risks or signs of disease. These collected user data are conducive to the subsequent development of personalized and precise medical services.

[0114] In the process of medical data collection, the rational use of the sensor group and effective data acquisition are key links. The following is the specific implementation process of collecting multiple medical data of the user through the sensor group:

[0115] The first is the careful selection and scientific deployment of sensors. According to the characteristics of various medical data, suitable sensors are selected and reasonably placed on the user's body or surrounding environment. The high-precision electronic blood pressure monitor sensor is based on the principle of oscillometric method and is accurately worn on the user's arm at the same height as the heart to accurately measure blood pressure data. For heart rate and pulse information, the photoelectric capacitance pulse wave sensor emits light and detects changes in reflected light, or the electrocardiogram sensor captures the electrical activity of the heart, and is placed in appropriate parts such as the wrist, finger or chest according to measurement needs to achieve accurate data collection. The user's body temperature is measured with the help of thermistors or infrared temperature sensors, which are placed in the armpits, mouth, ear canal or forehead according to different types to ensure accurate acquisition of body temperature data. In order to record movement information during sleep, accelerometers or micro-motion sensors are cleverly placed under the mattress or near the body to sensitively perceive subtle changes such as turning over and limb movements during sleep.

[0116] When sensors collect data, they strictly follow the set time intervals or continue in real time. The blood pressure sensor automatically measures at regular intervals, the heart rate and pulse sensors monitor and record changes in real time, the body temperature sensor also obtains data at the corresponding frequency, and the sleep movement sensor captures the dynamics of sleep throughout the process. The collected data is quickly transmitted through wired (such as USB, Bluetooth, etc.) or wireless (such as Wi-Fi, Bluetooth low energy, ZigBee, etc.) communication technology. The data is encrypted during the transmission process to ensure its security and privacy, and effectively prevent the risk of theft and tampering.

[0117] After the data is transmitted to the system, preprocessing and calibration work is immediately carried out. The data cleaning process effectively removes abnormal values ​​or erroneous data caused by sensor failure, interference, etc., to ensure reliable data quality. For example, abnormal blood pressure data points are corrected or eliminated after identification. At the same time, sensors are calibrated regularly to ensure measurement accuracy. Different sensors use corresponding calibration methods, such as comparing blood pressure sensors with standard sphygmomanometers, and calibrating temperature sensors under standard temperature environments. The calibration frequency depends on the accuracy requirements and usage, such as once a week or once a month, to ensure that the sensor always maintains accurate measurement performance.

[0118] Finally, various types of medical data from different sensors are integrated and stored according to time sequence and data type to build a complete user medical data set for subsequent analysis and processing. At the same time, a strict data quality monitoring mechanism is established to monitor data quality in real time through threshold setting, trend analysis and other means to promptly detect sensor failures or data transmission anomalies. Once discovered, an alarm is immediately issued to notify maintenance personnel to handle the problem, ensuring that the collected data is always stable and reliable, thereby providing a solid data foundation for medical diagnosis, health management, etc., thereby strongly supporting the precise development of medical services.

[0119] In step S140, the user's medical data is encrypted in pieces using the public key, and then the encrypted pieces of the user's medical data are distributed and stored in storage nodes using blockchain technology. Specifically, Figure 5 FIG. 1 is a flowchart of step S140 in the blockchain-based medical data collection method according to an embodiment of the present application. Figure 5 As shown, the step S140 includes: S141, sharding the various medical data of the user to obtain multiple medical data shards of the user; S142, using the public key and using asymmetric encryption technology to encrypt each of the multiple medical data shards to obtain encrypted shards of the various medical data of the user; S143, using the blockchain technology to disperse and store the encrypted shards of the various medical data of the user on storage nodes at different locations.

[0120] In step S141, the various medical data of the user are sharded to obtain multiple medical data shards of the user. It should be understood that sharding the various medical data of the user means decomposing the complete medical data (such as a continuous blood pressure data sequence, a heart rate data sequence, etc.) into smaller, relatively independent data fragments according to specific rules, that is, multiple medical data shards. These shards can be evenly divided based on the amount of data, or divided according to the logical structure of the data (such as according to time intervals, data types, etc.). For example, for blood pressure data collected continuously over a period of time, it can be divided into different shards according to the amount of data per hour or day; for medical data records containing multiple types of information (such as physiological indicators, detection time, etc.), different types of information can be divided into different shards. The distributed storage characteristics of blockchain technology require that data be stored and managed in an appropriate form. By sharding medical data, it can better match the distributed ledger structure of blockchain, and facilitate storage and management on multiple storage nodes. Each shard can be propagated, verified and stored in the blockchain network as an independent unit, which improves the flexibility and efficiency of data storage and processing. Moreover, during the storage and transmission of the sharded data, even if a shard is attacked or leaked, it is difficult for the attacker to obtain the complete medical data because it only contains partial data information, thus reducing the risk of data leakage and effectively protecting the privacy of users. At the same time, through a reasonable sharding strategy, sensitive information can be controlled and protected in a more fine-grained manner, such as storing the part containing personal identity identification separately from the medical test data in different shards, further enhancing data security.

[0121] In step S142, the public key is used to encrypt each of the multiple medical data slices by asymmetric encryption technology to obtain the encrypted slices of the user's medical data. It should be understood that asymmetric encryption technology is a cryptographic technology that uses two different keys, a public key and a private key, for encryption and decryption operations. The public key can be made public to anyone for encryption of data; while the private key is kept confidential by a specific user, and only the user with the private key can decrypt the data encrypted with the corresponding public key. This key separation feature makes it impossible to easily decrypt the data during data transmission and storage, even if the public key is obtained by others, thereby ensuring the confidentiality of the data. Medical data contains extremely sensitive personal health information of the user, and its privacy and security are of vital importance. Using the public key for asymmetric encryption, even if the data is illegally obtained during the distributed storage process or the storage node is attacked, the attacker cannot decrypt the encrypted slices because he does not have the corresponding private key, and thus cannot obtain the original medical data content, effectively preventing the leakage of the user's medical data and protecting the user's privacy rights.

[0122] In step S143, the blockchain technology is used to encrypt and fragment the user's medical data and store them in storage nodes at different locations. It should be understood that by utilizing the distributed storage characteristics of the blockchain, the storage resources of each node can be fully utilized, avoiding the resource bottleneck problem that may be caused by centralized storage. Different storage nodes can participate in data storage according to their own storage capacity and performance, realizing flexible allocation and efficient utilization of resources. And the encrypted fragments of medical data are stored in storage nodes at different locations to avoid the risk of single point failure. Even if a storage node is attacked or fails, the data on other nodes is still safe and available, ensuring the integrity and confidentiality of the medical data.

[0123] In step S150, the encrypted slices of each user's medical data in the storage node are extracted, and the encrypted slices of each user's medical data are decrypted to obtain decrypted medical data. Specifically, Figure 6 FIG. 1 is a flowchart of step S150 in the blockchain-based medical data collection method according to an embodiment of the present application. Figure 6As shown, the step S150 includes: S151, extracting the encrypted slices of the user's medical data from each of the storage nodes; S152, using the private key to decrypt the encrypted slices of the user's medical data to obtain a first medical data set; S153, using the public key to verify the encrypted slices of the user's medical data to obtain a second medical data set; S154, comparing and analyzing the contents of each group of corresponding first medical data and second medical data in the first medical data set and the second medical data set to verify the medical data to obtain the decrypted medical data, wherein, if the contents of the first medical data and the second medical data are consistent, it is determined that the medical data is complete and authentic, and if the contents of the first medical data and the second medical data are inconsistent, it is determined that the medical data is distorted.

[0124] In step S151, the encrypted slices of the user's medical data are extracted from each of the storage nodes. It should be understood that the encrypted slices of the user's medical data are scattered and stored on each storage node. The purpose of extracting the encrypted slices from these nodes is to reintegrate the fragmented stored data, provide a complete data basis for subsequent decryption operations, and finally restore the user's complete medical data. This step is also an important starting step for subsequent verification of data integrity and security. Only by accurately and completely obtaining the encrypted slices on each node can we ensure that the data has not been tampered with during the storage process through decryption and further data verification processes (such as decrypting with a private key and comparing with the public key for verification, etc.), and that the data source is authentic and reliable, providing a solid guarantee for the safe use of medical data, which is conducive to maintaining the authority and credibility of medical data.

[0125] In step S152, the encrypted slices of the medical data of the user are decrypted using the private key to obtain the first medical data set. It should be understood that in the previous data storage process, the medical data slices are encrypted using the public key to ensure the security of the data in the distributed storage environment and prevent others from obtaining the data content without authorization. The private key is the only decryption key corresponding to the public key, and only the user with the private key (i.e., the data owner) can decrypt the encrypted slices. This mechanism strictly limits the access rights to the data and ensures the confidentiality of the medical data. Only authorized users can obtain and use their sensitive health information, maintaining the user's ownership and control over their medical data. Specifically, the medical data exists in the form of encrypted slices on the distributed storage nodes. These slices alone cannot provide meaningful information, and in order to prevent data leakage and tampering, the corresponding private key must be used for decryption operations to restore the encrypted slices to the original, understandable and usable medical data format, ensuring the security and integrity of the data throughout the entire collection, storage and use process, so that the medical data can be used for legitimate purposes such as diagnosis, treatment, and health management under the premise of safety. The first medical data set decrypted by the private key contains various original medical data of the user collected by the sensor, which is unmodified and reflects the detailed information of the user's real health status.

[0126] In step S153, the public key is used to verify the encrypted slices of the user's medical data to obtain a second medical data set. It should be understood that although the private key decryption can restore the encrypted slices to the original data, there may be a risk of data tampering during data storage and transmission. The public key and the private key exist in pairs and have a specific mathematical relationship. Using the public key to verify the encrypted slices can be used to check whether the data has been illegally modified after encryption based on this mathematical relationship. If the data is tampered with, the public key verification will not be able to obtain the correct result, thereby timely discovering the integrity problem of the data. That is, private key decryption mainly focuses on restoring the data content, while public key verification checks the security of the data from another perspective. The combination of the two forms a double protection mechanism. Even if the private key is leaked under certain circumstances, if the attacker tampers with the data, because the public key verification cannot pass, it is still impossible to successfully forge or tamper with the data without being discovered, which further enhances the security of the entire medical data processing process, prevents unauthorized modification and falsification of data from misleading medical decisions, and ensures that medical services are based on accurate and reliable data. The second medical data set contains medical data obtained after public key processing. If the public key verification is successful, the second medical data set will contain the same complete original medical data as the first medical data set. If the verification fails, the second medical data set contains some medical data with erroneous data.

[0127] In step S154, the contents of the first medical data and the second medical data corresponding to each group in the first medical data set and the second medical data set are compared and analyzed to verify the medical data to obtain the decrypted medical data, wherein if the contents of the first medical data and the second medical data are consistent, it is determined that the medical data is complete and true, and if the contents of the first medical data and the second medical data are inconsistent, it is determined that the medical data is distorted. It should be understood that in the medical field, the accuracy of the data is directly related to the correctness of the medical decision, which in turn affects the health and treatment effect of the patient. By comparing the contents of each group of corresponding medical data in the first medical data set (private key decryption result) and the second medical data set (public key verification result), the data is cross-checked from two independent decryption and verification paths, which can ensure the accuracy of the acquired medical data to the greatest extent. If the two are consistent, it means that the data has not been damaged or tampered with in a series of processes such as encryption, storage, transmission and decryption, so that the accuracy and reliability of the medical data can be ensured. If the first medical data and the second medical data are inconsistent, it is determined that the medical data has been distorted, which enables rapid adoption of appropriate measures, such as starting a data recovery mechanism, investigating the cause of data distortion (which may be a storage node failure, network transmission problem, or malicious attack, etc.), and preventing distorted data from spreading in the system and being misused for medical decision-making. This proactive data risk prevention mechanism helps maintain system stability and data integrity, avoids a series of medical problems caused by the use of erroneous data, reduces medical risks, and ensures the security and effectiveness of medical data throughout its life cycle.

[0128] In step S160, the decrypted medical data is analyzed to obtain an analysis report, and the analysis report is displayed. It should be understood that the decrypted medical data contains rich health information of the user, but these raw data themselves are often complicated and isolated, and it is difficult to directly provide clear guidance for medical decision-making or health management. Through analysis, these raw data can be converted into meaningful information, and then the health trends, disease risk factors and potential health problems hidden behind the data can be excavated, so as to maximize the value of medical data. For example, potential cardiovascular disease risk trends can be found from long-term blood pressure and heart rate data, thereby providing a basis for early intervention. And medical professionals need to integrate various information when diagnosing diseases, formulating treatment plans or evaluating treatment effects. The analysis report of the decrypted medical data can provide a comprehensive, accurate and quantitative basis to help doctors more accurately judge the condition, choose the most appropriate treatment method, and improve the accuracy of diagnosis and the effectiveness of treatment. At the same time, for the user himself, the analysis report helps to formulate a personalized health management plan, such as diet adjustment, exercise advice, disease prevention measures, etc., which is conducive to promoting users to actively participate in health management and thus improve the overall health level.

[0129] In summary, the blockchain-based medical data collection method based on the embodiment of the present application is explained. It first registers a personal account through the personal basic information uploaded by the user, then verifies and verifies when the user logs in to determine whether a secondary SMS verification is required, and then collects the user's medical data through the sensor group, and uses the public key to encrypt the collected medical data in pieces, and then uses blockchain technology to distribute and store it in various storage nodes, and then extracts the encrypted pieces and decrypts them to obtain the decrypted medical data, and then analyzes them to generate an analysis report and display it. In this way, the safe management and effective application of user medical data can be achieved.

Claims

1. A medical data collection method based on blockchain, characterized in that: include: Register a personal account based on the personal basic information uploaded by the user, wherein the personal account includes personal basic information, account password, public key and private key; After the user inputs the personal account for login, the personal account for login is verified, including: embedding and quickly scanning the semantic query encoding of the personal account for login to obtain password verification query response encoding features; based on the password verification query response encoding features, a judgment result is obtained, and the judgment result is used to indicate whether SMS verification is required through a mobile phone number; Collecting a plurality of medical data of the user through a sensor group, the medical data including blood pressure data, heart rate data, pulse data, body temperature data and movement information of the user during sleep; After the public key is used to perform sharding encryption on the user's medical data, the obtained encrypted shards of the user's medical data are distributedly stored in the storage nodes through blockchain technology; Extracting the encrypted slices of each user's medical data in the storage node, and decrypting the encrypted slices of each user's medical data to obtain decrypted medical data; Analyzing the decrypted medical data to obtain an analysis report, and displaying the analysis report; The embedded coding and quick scanning semantic query coding of the logged-in personal account to obtain the password verification query response coding features include: Acquire multiple historical login account passwords continuously input by the user, and record the input time of each of the historical login account passwords to obtain multiple password input times; Adding each password input time of the plurality of password input times to the end of the corresponding historical login account password to obtain a set of login account password data containing password input time information; respectively performing embedded coding and semantic association analysis on the set of login account password data containing password input time information to obtain a set of login account password data association feature vectors; Obtain the login account password to be verified and the time when the password to be verified was entered by the user during the login process; After adding the password input time to be verified to the end of the login account password to be verified, embedding coding is performed on the obtained password data to obtain an embedded coding vector of the password data to be verified; Performing fast semantic response query encoding on the set of the embedded coding vector of the password data to be verified and the associated feature vector of the login account password data to obtain a password verification query response coding vector as the password verification query response coding feature; The method of performing fast semantic response query encoding on the set of the embedded coding vector of the password data to be verified and the associated feature vector of the login account password data to obtain the password verification query response coding vector includes: Calculate the mutual information between the embedded coding vector of the password data to be verified and each login account password data associated feature vector in the set of login account password data associated feature vectors to obtain a set of semantic fast semantic matching factors of the password data to be verified-login account password; Based on the set of to-be-verified-login account password semantics fast semantic matching factors, determining a subset of fast-matched login account password data associated feature vectors; Semantic query encoding is performed on a subset of the embedded encoding vector of the password data to be verified and the associated feature vector of the login account password data to obtain the password verification query response encoding vector.

2. The blockchain-based medical data collection method according to claim 1, characterized in that: The set of login account password data containing password input time information is respectively subjected to embedded coding and semantic association analysis to obtain a set of login account password data association feature vectors, including: Using the password embedding matrix, each of the login account password data containing the password input time information in the set of login account password data containing the password input time information is embedded and encoded to obtain a set of login account password data embedded encoding vectors; The set of the login account password data embedded encoding vectors is passed through a password semantic association analyzer based on a stacked LSTM to obtain a set of login account password data associated feature vectors.

3. The blockchain-based medical data collection method according to claim 2 is characterized in that: After adding the password input time to be verified to the end of the login account password to be verified, embedding coding is performed on the obtained password data to be verified to obtain an embedded coding vector of the password data to be verified, including: Add the password input time to be verified to the end of the login account password to be verified to obtain the password data to be verified; The password embedding matrix is ​​used to embed the password data to be verified to obtain an embedding coding vector of the password data to be verified.

4. The blockchain-based medical data collection method according to claim 3 is characterized in that: Based on the set of quick semantic matching factors of the to-be-verified-login account password semantics, a subset of quick matching login account password data associated feature vectors is determined, including: Identify the first to-be-verified-login account password semantic maximum value and the second to-be-verified-login account password semantic maximum value from the set of to-be-verified-login account password semantic fast semantic matching factors; Based on the positions of the first to-be-verified-login account password semantic maximum value and the second to-be-verified-login account password semantic maximum value in the set of to-be-verified-login account password semantic fast semantic matching factors, a subset of the login account password data associated feature vectors for fast matching is determined from the set of login account password data associated feature vectors.

5. The blockchain-based medical data collection method according to claim 4 is characterized in that: The method comprises: performing semantic query encoding on a subset of the embedded coding vector of the password data to be verified and the feature vector associated with the login account password data to obtain the password verification query response coding vector, including: performing a linear transformation on the embedded coding vector of the password data to be verified to obtain a query vector of the password data to be verified as a query vector and a value vector of the password data to be verified as a value vector, and taking a subset of the feature vector associated with the login account password data as a subset of the key vector, and performing cross-domain query encoding on the query vector, the value vector and a subset of the key vector based on a converter structure to obtain the password verification query response coding vector.

6. The blockchain-based medical data collection method according to claim 5 is characterized in that: Based on the password verification query response coding feature, a discrimination result is obtained, including: passing the password verification query response coding vector through a classifier-based SMS verification discriminator to obtain the discrimination result.

7. The blockchain-based medical data collection method according to claim 6 is characterized in that: After the public key is used to perform sharding encryption on the user's medical data, the obtained user's medical data is encrypted and sharded and distributedly stored in the storage nodes through blockchain technology, including: Slicing each item of medical data of the user to obtain multiple medical data slices of the user; Using the public key and asymmetric encryption technology, encrypt each of the multiple medical data slices to obtain encrypted slices of each medical data of the user; The blockchain technology is used to encrypt and fragment the user's medical data and store them in storage nodes at different locations.

8. The blockchain-based medical data collection method according to claim 7 is characterized in that: Extracting the encrypted slices of each user's medical data in the storage node, and decrypting the encrypted slices of each user's medical data to obtain decrypted medical data, including: Extracting encrypted slices of each item of medical data of the user from each storage node; Decrypting the encrypted slices of each medical data of the user using the private key to obtain a first medical data set; Using the public key to verify the encrypted slices of each medical data of the user to obtain a second medical data set; The contents of the first medical data and the second medical data corresponding to each group in the first medical data set and the second medical data set are compared and analyzed to perform medical data verification to obtain the decrypted medical data, wherein if the contents of the first medical data and the second medical data are consistent, it is determined that the medical data is complete and authentic; if the contents of the first medical data and the second medical data are inconsistent, it is determined that the medical data is distorted.

Images