A medical information sharing method and system based on medical big data

Abstract

The present invention relates to the field of medical information processing technology, and in particular to a medical information sharing method and system based on medical big data. The content includes: real-time collection of medical data to obtain original medical data; intelligent preprocessing of original medical data to obtain intelligent preprocessed medical data; encryption processing of intelligent preprocessed medical data to obtain encrypted medical data; synchronization processing of encrypted medical data to obtain encrypted and real-time synchronized medical data; upon receiving a medical data access request, verifying the identity of the visitor, and after the verification is passed and it is determined that the visitor has access rights, decrypting the encrypted and real-time synchronized medical data to obtain decrypted medical data. The technical problems of inaccurate medical data detection and processing and low security factor of medical information sharing are solved.

Description

Technical Field

[0001] The present invention relates to the field of medical information processing technology, and in particular to a medical information sharing method and system based on medical big data. Background Art

[0002] In the development of modern medical informatization, with the widespread application of various electronic medical record systems (EMR) and medical data management systems, the need for sharing and interoperability of medical information based on medical big data has become increasingly urgent. However, different medical institutions usually use different electronic medical record systems, resulting in significant differences in medical data standards and formats, forming an information island phenomenon. Secondly, the medical information sharing process based on medical big data needs to attach great importance to the privacy protection of medical information. Medical information is sensitive data, and a sharing model that is not properly protected may lead to the leakage of patient privacy and even cause legal risks.

[0003] However, in the existing implementation process of medical information sharing based on medical big data, there are problems such as inaccurate medical data detection and processing and low security factor of medical information sharing. Summary of the invention

[0004] The present invention provides a medical information sharing method and system based on medical big data to solve the technical problems of inaccurate medical data detection and processing and low safety factor of medical information sharing.

[0005] The present invention provides a medical information sharing method and system based on medical big data, which specifically includes the following technical solutions:

[0006] A medical information sharing method based on medical big data comprises the following steps:

[0007] S1. Collect medical data in real time to obtain original medical data; perform intelligent preprocessing of data analysis, data cleaning, data format conversion and mapping, outlier detection and processing, data standardization and normalization on the original medical data to obtain medical data after intelligent preprocessing; and introduce a multi-level dynamic recursive anomaly detection algorithm to perform outlier detection and processing on the medical data that has completed format conversion and mapping; the multi-level dynamic recursive anomaly detection algorithm performs anomaly detection on the basis of multi-level data decomposition through dynamic threshold generation, recursive clustering detection, anomaly score calculation and reverse verification steps;

[0008] S2. Encrypt the medical data after intelligent preprocessing to obtain encrypted medical data; synchronize the encrypted medical data to obtain encrypted and real-time synchronized medical data;

[0009] S3. When a medical data access request is received, the identity of the visitor is verified. After the verification is passed and it is determined that the visitor has access rights, the encrypted and real-time synchronized medical data is decrypted to obtain the decrypted medical data.

[0010] Preferably, the S1 specifically includes:

[0011] For the data decomposition in the multi-level dynamic recursive anomaly detection algorithm, the medical data that has completed format conversion and mapping is hierarchically decomposed to generate two or more levels of medical data subsets; based on the medical data subsets, the adaptive threshold required for the medical data subsets in each level is calculated through the dynamic threshold adaptive generation mechanism; the formula is as follows:

[0012] ,

[0013] in, It is The first Medical data subset Adaptive threshold of Indicates The first level A subset of medical data; It is The first Medical data subset The mean of It is The first level Medical data subset The standard deviation of Indicates medical data that has completed format conversion and mapping The Medical data; Represents a subset of medical data The total amount of data; It is Weight factors for the layers.

[0014] Preferably, the S1 specifically includes:

[0015] In the recursive clustering detection process, the cluster deviation score is used to determine the abnormal possibility of medical data in the medical data subset; the deviation of each layer of medical data subset is calculated based on the current adaptive threshold to detect and strip abnormal medical data layer by layer, and the stripping results are recursively brought into the next layer of clustering detection to form a layer-by-layer recursive detection mode; the specific formula is as follows:

[0016] ,

[0017] in, For the The first Deviation scores for clustering of medical data subsets; It is Hierarchical recursive clustering deviation factor; Indicates Recursive weighting factor for the levels.

[0018] Preferably, the S1 specifically includes:

[0019] After each layer of cluster detection, the abnormality score is calculated for the medical data that has completed format conversion and mapping, and recursive calculation is performed based on the layered deviation score; the abnormality score calculation formula is as follows:

[0020] ,

[0021] in, Indicates medical data that has completed format conversion and mapping The Abnormality score of medical data; is the number of levels; For the The weight coefficient of the level is used to regulate the impact of different levels on the abnormality score; is the weight coefficient, ensuring the recursive increase and decrease of the abnormality score; the variable is the adaptive factor; Indicates medical data that has completed format conversion and mapping The Abnormality score of medical data;

[0022] Based on the abnormality score, normal medical data and abnormal medical data are distinguished, and the probability density function of abnormal medical data and normal medical data is calculated.

[0023] Preferably, the S1 specifically includes:

[0024] In the reverse verification process, the recursive clustering detection results at each level are verified based on the probability density functions of abnormal medical data and normal medical data.

[0025] Preferably, the S2 specifically includes:

[0026] The medical data after intelligent preprocessing is encrypted using a dynamic layered composite encryption algorithm; in the process of implementing the dynamic layered composite encryption algorithm, the medical data after intelligent preprocessing is first processed to obtain the characteristics of the medical data after intelligent preprocessing; then a random basis matrix is ​​generated based on the characteristics of the medical data after intelligent preprocessing; a dynamic key vector is generated based on the random basis matrix; a nonlinear mapping matrix is ​​constructed based on the dynamic key vector; based on the nonlinear mapping matrix, the medical data after intelligent preprocessing is nonlinearly mapped through matrix multiplication to obtain the medical data after nonlinear transformation; the dynamic key matrix is ​​introduced to perform layered encryption on the medical data after nonlinear transformation to obtain a ciphertext vector; the ciphertext vector is subjected to compound random mapping through a random perturbation function to obtain the encrypted medical data;

[0027] The random perturbation function is defined as:

[0028] ,

[0029] in, is the random perturbation function of the composite random mapping; is input data or intermediate results;

[0030] The specific formula of composite random mapping is:

[0031] ,

[0032] in, Represents encrypted medical data; Random perturbation function representing a composite random mapping; Indicates The ciphertext generated after layer-by-layer encryption; It is The ciphertext generated after layer-by-layer encryption; Indicates the total number of layered encryption layers.

[0033] Preferably, the S3 specifically includes:

[0034] The encrypted and real-time synchronized medical data is decrypted, and the decryption process includes de-perturbation, layered inverse decryption, nonlinear inverse mapping, and dynamic key verification.

[0035] A medical information sharing system based on medical big data, applied to the above-mentioned medical information sharing method based on medical big data, includes the following parts:

[0036] Data acquisition module, data intelligent processing module, data encryption and security module, data synchronization module, authorized access and decryption module;

[0037] The data acquisition module is responsible for collecting medical data, obtaining raw medical data, and sending the raw medical data to the data intelligent processing module;

[0038] The data intelligent processing module performs intelligent preprocessing on the original medical data and sends the intelligently preprocessed medical data to the data encryption and security module;

[0039] The data encryption and security module encrypts the medical data after intelligent preprocessing and sends the encrypted medical data to the data synchronization module;

[0040] The data synchronization module synchronizes the encrypted medical data to obtain encrypted and real-time synchronized medical data, and transmits the encrypted and real-time synchronized medical data to the authorized access and decryption module;

[0041] The authorized access and decryption module verifies the identity of the visitor upon receiving a request for access to medical data. After the verification is passed and it is determined that the visitor has access rights, the encrypted and real-time synchronized medical data is decrypted to obtain the decrypted medical data.

[0042] The beneficial effects of the technical solution of the present invention are:

[0043] 1. By introducing a multi-level dynamic recursive anomaly detection algorithm, the problem of the difficulty in accurately detecting and identifying abnormal data in traditional medical data is effectively solved. Through the steps of dynamic threshold generation, recursive cluster detection, anomaly score calculation and reverse verification, more sophisticated anomaly detection is achieved, so that anomalies at different data levels can dynamically adjust adaptive thresholds and respond sensitively. In the steps of data decomposition, dynamic threshold generation and recursive cluster detection, the generation of adaptive thresholds for each subset and multi-level cluster detection make anomaly detection highly adaptable and noise-resistant, ensuring the accuracy of anomaly detection and providing strong support for the intelligent preprocessing of medical data.

[0044] 2. In the data encryption link, a dynamic layered composite encryption algorithm is adopted. The feature-driven random matrix generation technology is used to generate a random base matrix, and a dynamic key vector is generated based on the random base matrix. The encrypted medical data is further generated through nonlinear transformation, layered encryption and composite random mapping. The dynamic nature of dynamic key generation ensures the unpredictability and personalization of the key. Nonlinear transformation not only improves the encryption strength of medical data after intelligent preprocessing, but also avoids the singleness and fragility of traditional encryption algorithms. Layered encryption enables the medical data after intelligent preprocessing to be encrypted with a unique dynamic key matrix at each layer, which increases the difficulty of cracking exponentially and greatly improves the security of medical data in the process of medical information sharing. At the same time, the composite random mapping introduces irreversible random perturbations in the final ciphertext, further enhancing the privacy protection of the data. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a structural diagram of a medical information sharing system based on medical big data described in the present invention;

[0046] Figure 2 This is a flow chart of a medical information sharing method based on medical big data described in the present invention. DETAILED DESCRIPTION

[0047] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0049] The following is a detailed description of a specific solution of a medical information sharing method and system based on medical big data provided by the present invention in conjunction with the accompanying drawings.

[0050] Refer to the attached Figure 1 , which shows a structure diagram of a medical information sharing system based on medical big data provided by an embodiment of the present invention, the system includes the following parts:

[0051] Data acquisition module, data intelligent processing module, data encryption and security module, data synchronization module, authorized access and decryption module;

[0052] The data acquisition module is responsible for collecting medical data from different electronic medical record systems of various medical institutions in real time, obtaining the original medical data, and sending the original medical data to the data intelligent processing module;

[0053] The data intelligent processing module performs intelligent preprocessing on the original medical data to obtain the intelligent preprocessed medical data, and sends the intelligent preprocessed medical data to the data encryption and security module;

[0054] The data encryption and security module encrypts the medical data after intelligent preprocessing to obtain the encrypted medical data, and sends the encrypted medical data to the data synchronization module;

[0055] The data synchronization module synchronizes the encrypted medical data to ensure that the encrypted medical data can be synchronized in different medical institutions, obtains encrypted and real-time synchronized medical data, and transmits the encrypted and real-time synchronized medical data to the authorized access and decryption module;

[0056] The authorized access and decryption module verifies the identity of the visitor upon receiving a request to access medical data. After the verification is passed and it is determined that the visitor has access rights, the encrypted and real-time synchronized medical data is decrypted to obtain the decrypted medical data and realize medical information sharing.

[0057] Refer to the attached Figure 2 , which shows a flow chart of a medical information sharing method based on medical big data provided by an embodiment of the present invention, the method comprising the following steps:

[0058] S1. Collect medical data in real time to obtain original medical data; perform intelligent preprocessing on the original medical data to obtain intelligent preprocessed medical data;

[0059] Medical data is collected in real time from different electronic medical record systems of various medical institutions to obtain original medical data. The original medical data is subjected to intelligent preprocessing such as data analysis, data cleaning, data format conversion and mapping, abnormal value detection and processing, and data standardization and normalization to obtain medical data after intelligent preprocessing; in the process of intelligent preprocessing, data analysis, data cleaning, data format conversion and mapping, and data standardization and normalization all use existing technical means, which will not be elaborated here.

[0060] For outlier detection and processing, a multi-level dynamic recursive anomaly detection algorithm is introduced to detect and process outliers in medical data that has completed format conversion and mapping. The core idea of ​​the multi-level dynamic recursive anomaly detection algorithm is to achieve refined anomaly detection through dynamic threshold generation, recursive clustering detection, anomaly score calculation and reverse verification on the basis of multi-level data decomposition. The specific implementation process is as follows:

[0061] First, data decomposition and initialization are performed; the medical data that has completed format conversion and mapping Use expert experience to perform hierarchical decomposition according to characteristics and distribution characteristics to form multiple levels of medical data subsets The main purpose of data decomposition is to classify data with different characteristics into their respective levels by dividing the medical data after format conversion and mapping, so as to achieve more refined detection in the future. The specific implementation method is as follows:

[0062] ,

[0063] in, Indicates The first level A subset of medical data is the medical data that has completed format conversion and mapping. The medical data subset obtained by decomposition; Indicates medical data that has completed format conversion and mapping The Medical data; yes The main characteristic function of The main features, such as the patient’s age, medical history and other key information, are determined based on domain knowledge; For the The weight function of the medical data subset is used to measure Secondary characteristics of It is The level weight factor of the level is used to control the Level pair Characteristics of medical data The impact of (for example, the weight of the first level can be set to 1.5, and the weight of the second level can be set to 2.0); for The weight factors of the medical data subsets are used to control the influence of different subsets and ensure the flexibility of subset division (for example, the weight factor of the first medical data subset is 0.5, and the weight factor of the second medical data subset is 0.3); is the benchmark value of hierarchical decomposition. After data decomposition, The output is used for threshold adaptive setting in the next step.

[0064] Data decomposition considers the weighted processing of different features of medical data that have completed format conversion and mapping, and realizes the division of medical data subsets through threshold setting. The specific process is as follows:

[0065] 1. Set the medical data to be converted and mapped as ;

[0066] 2. Main characteristic functions and the weight function for the medical data subset Determination of:

[0067] Main characteristic functions : Used to describe medical data that has completed format conversion and mapping The Medical data For example, for patients, It can be a key indicator reflecting the patient's health status, such as age, medical history severity score, etc. It will be determined based on the actual application scenario.

[0068] Weight function for medical data subset : Used to measure medical data after format conversion and mapping The Medical data Secondary characteristics of the patient, such as the patient's gender, family medical history, environmental factors, etc. The specific details will be determined according to the actual application scenario.

[0069] 3. Hierarchical weight factor and the weight factor for the medical data subset Settings:

[0070] Hierarchical weight factor : Used to control the main characteristic function The weight in the hierarchical division is the different attention paid to the main features at each level. As the level increases, a certain feature (such as age) may be given a greater weight, so that subsequent anomaly detection focuses on this feature.

[0071] Weighting factors for medical data subsets : Weight function for controlling subsets of medical data The influence of the feature in the partitioning of medical data subsets. Its size affects the proportion of secondary features in the hierarchical partitioning, ensuring the adaptability of different medical data subsets to features when decomposing.

[0072] 4. Hierarchical decomposition and condition setting:

[0073] In the hierarchical decomposition, As a metric, the weight and the weighted sum of the primary and secondary features are used to determine the data grouping.

[0074] Set the base value for hierarchical decomposition , whose value is the mean or median of the medical data after format conversion and mapping, and is used to identify the qualified Divide into corresponding medical data subsets.

[0075] 5. Data division:

[0076] satisfy Complete format conversion and mapping of medical data The Medical data Divided into medical data subsets In each Represents a hierarchically divided subset of medical data.

[0077] Next, the adaptive thresholds required for the medical data subsets in each layer are calculated through the dynamic threshold adaptive generation mechanism. , ensuring that medical data at different levels have adaptable threshold ranges. As the anomaly detection standard for the medical data subset in each level; by calculating the mean, standard deviation and deviation degree of each medical data subset, an adaptive threshold is generated for each level of medical data subset The formula is as follows:

[0078] ,

[0079] in, It is The first level Medical data subset Adaptive threshold of It is The first level Medical data subset The mean of It is The first level Medical data subset The standard deviation of Represents a subset of medical data The total amount of data; It is The weight factor of each level is used to adjust the sensitivity of each level. Different values ​​are used to assign dynamic deviation sensitivity to medical data at different levels. It is determined according to specific scenarios. For example, , then the deviation part is simplified to the absolute average of the deviation values, indicating that it is less sensitive to abnormal medical data; , then the larger deviation value has a greater impact on the overall deviation, which is suitable for anomaly detection with medium sensitivity; , which will significantly amplify abnormal medical data and is particularly suitable for anomaly detection that is sensitive to extreme anomalies; the generated adaptive threshold provides the basis for subsequent anomaly detection.

[0080] In the process of dynamic threshold generation, the mean and standard deviation of the medical data subset are first determined, and then the deviation part is calculated by the weight factor. Specifically, the medical data subset Each medical data that has completed format conversion and mapping The Medical data With the medical data subset The mean The deviation value Promoted to The power increases the weight of the medical data with large deviations, which helps to amplify the impact of abnormal medical data on the overall deviation. The deviation of the entire medical data subset is generated by accumulating and averaging the deviation values, and finally the deviation is analyzed. The power operation is used to further adjust its sensitivity while keeping the final deviation consistent with the data distribution characteristics. The adaptive threshold of the final medical data subset is the combined result of the mean, standard deviation and deviation.

[0081] In recursive cluster detection, the cluster deviation score is used to determine the possibility of abnormal medical data in the medical data subset. The deviation of each layer of medical data subset will be calculated based on the current adaptive threshold to detect and remove abnormal medical data layer by layer, and the removal results will be recursively brought into the next layer of cluster detection to form a layer-by-layer recursive detection mode. The specific formula is as follows:

[0082] ,

[0083] in, For the The first level The deviation score of each medical data subset cluster is used to evaluate the degree to which the medical data at this level deviates from the normal distribution; It is The hierarchical recursive clustering deviation factor is used to adjust the deviation sensitivity of different levels. , indicating that medical data with larger deviations have a more significant impact on the deviation score; , which means reducing the impact of large deviations and making the calculated deviation score more conservative. The value of can be determined according to the specific scenario. , the square of the absolute difference of the deviation is processed to make the impact of outlier medical data more obvious; Indicates The recursive weighting factor of the hierarchy ensures that when the medical data in the medical data subset is far away from the mean, the deviation score increases exponentially, as shown in , the deviation score is more sensitive to larger deviations. The calculated deviation score It is used to further screen abnormal medical data at this level, and at the same time, the recursively peeled medical data subset is taken as input to the next level of detection.

[0084] The deviation score combines the mean, standard deviation, recursive weighting factor, etc. of the medical data subset to evaluate whether there are abnormalities in the medical data at this level. Calculate the mean and standard deviation; then for each medical data that has completed format conversion and mapping The Medical data , calculate the absolute difference from the mean , which represents the distance between the medical data after format conversion and mapping and the mean. Power processing is used to adjust the sensitivity of the deviation; the deviation is then compared with the standard deviation. Standardization is performed to obtain dimensionless deviation scores, ensuring that the deviation scores of different levels are calculated on a consistent scale. At the same time, a recursive weighting factor is added to the deviation score. , used to enlarge Distance Mean The exponential function is introduced. , which decreases as the distance from the mean increases. Larger recursive weighting factors This makes the deviation score more sensitive to large deviations, while a smaller recursive weighting factor This will reduce the sensitivity of the deviation score. Finally, for each medical data that has completed format conversion and mapping The deviation scores of the medical data in the sum are formed into the first The first level Deviation scores for clustering of medical data subsets The deviation score reflects the deviation characteristics of the entire medical data subset.

[0085] After each layer of clustering detection, the medical data that has completed format conversion and mapping The Calculate abnormality score for medical data , the calculation of the anomaly score is used to comprehensively judge the degree of data deviation, and recursive calculation is performed based on the layered deviation score. The anomaly score will be recursively fed back to ensure that the anomaly score has high dynamic adaptability. The calculation formula of the anomaly score is as follows:

[0086] ,

[0087] in, Indicates medical data that has completed format conversion and mapping The Abnormality score of medical data; is the number of levels; For the The weight coefficient of each level is used to control the impact of different levels on the abnormality score. It is determined according to the specific scenario. For example, Set to a value that decreases with each level. For example, in the case of three levels, set , , , making higher-level deviations more significant; is the weight coefficient to ensure the recursive increase or decrease of the abnormality score, as follows: The value is 0.8 so that the logarithmic feedback item does not cause a large fluctuation in the score. It is an adaptive factor used to regulate the inhibitory effect when the abnormality score is too high. Specifically, Take 10 and make sure When it is large, the feedback item increase is limited; Indicates the abnormality score of the previous medical data, that is, the medical data after format conversion and mapping The The abnormality scores of medical data are calculated to ensure the recursive correlation of the abnormality scores and make the abnormality scores dynamically responsive.

[0088] By combining the deviation scores of the medical data subset clusters at each level and the abnormality score of the previous medical data, a recursive score is generated, making the abnormality score more flexible and dynamically responsive. First, the adjusted deviation scores of each level are calculated. , the basic calculation of deviation is ,express The distance from the mean of the medical data subset to which it belongs. In order to increase the sensitivity of the hierarchy, the deviation is exponentially operated and given different weight coefficients. , and get the adjusted deviation, This makes different levels of medical data have different contributions in the deviation calculation, thereby improving the adaptive ability of the multi-level dynamic recursive anomaly detection algorithm to different levels. Then divide the adjusted deviation by the standard deviation Standardization can convert the deviations of different scales to the same standard, which is convenient for comparison between different levels. Furthermore, in each level, the calculated standardized deviations are compared with the corresponding deviation scores. Through this product operation, the abnormal deviation contribution at the current level is obtained. Then the abnormal deviation contribution of all levels is averaged and aggregated to obtain the preliminary abnormality score of the medical data. Finally, in order to make the anomaly score have recursive feedback characteristics, a recursive adjustment term is introduced ,This item is based on the anomaly score of the previous medical data, and uses a logarithmic function to limit the recursive increase to prevent the anomaly score from being too high, while ensuring that the anomaly score can dynamically adapt to the changes of adjacent medical data.

[0089] Furthermore, the abnormality score By comparing with the threshold value preset according to the expert experience method, normal medical data and abnormal medical data are simply distinguished, and the normal medical data and abnormal medical data are processed respectively using the existing statistical analysis method to obtain the probability density function of abnormal medical data and normal medical data. and .

[0090] In the reverse verification step, the recursive clustering detection results of each level are verified by the difference between the abnormal data distribution and the normal data distribution to ensure the accuracy of the abnormal judgment. This reverse verification step uses the distribution difference integral calculation to identify the distribution pattern with high similarity by comparing the abnormal data distribution with the normal data distribution, thereby avoiding misjudgment. The reverse verification formula is as follows:

[0091] ,

[0092] in, Indicates The first level The reverse validation bias of a subset of medical data reflects the difference between the distribution of abnormal medical data and the distribution of normal medical data; The probability density function representing abnormal medical data is used to measure the probability density of abnormal medical data distribution and is estimated by existing statistical methods; Indicates medical data that has completed format conversion and mapping Medical data in is the probability density function of normal medical data, which is used to measure the probability density of normal medical data distribution; interval Used to set the verification range to ensure the stability of the calculation results. The specific interval value is determined according to the specific scenario, such as based on the mean and standard deviation, based on the quantile of the anomaly score, a fixed range or a specific threshold. Specifically, based on the quantile of the anomaly score, the values ​​of the anomaly score at the 10% and 90% quantiles are used as the interval. The value of , thereby excluding extremely abnormal outlier medical data and low abnormal medical data within the normal range; parameter To avoid a constant with a zero denominator. Exceeds the set adaptive threshold , it indicates that there is an error in the anomaly detection result of the current level, and the adaptive threshold needs to be readjusted And repeat the recursive cluster detection step.

[0093] S2. Encrypt the medical data after intelligent preprocessing to obtain encrypted medical data; synchronize the encrypted medical data to obtain encrypted and real-time synchronized medical data;

[0094] The medical data after intelligent preprocessing is encrypted using a dynamic layered composite encryption algorithm to obtain encrypted medical data; the dynamic layered composite encryption algorithm encrypts the medical data after intelligent preprocessing through dynamically generated keys, nonlinear transformations, layered encryption, and composite random mapping, and ensures that the output of each layer of operation is directly used as the input of the next step, ensuring the closed-loop and logical rigor of the dynamic layered composite encryption algorithm chain. The specific implementation process is as follows:

[0095] The medical data after intelligent preprocessing is represented as ,in, is the length of the medical data after intelligent preprocessing, element Represents the medical data after intelligent preprocessing. independent data blocks, where .

[0096] Furthermore, in order to ensure the dynamic characteristics of the data encryption process, the medical data after intelligent preprocessing is first processed by the existing feature engineering technology to obtain the characteristics of the medical data after intelligent preprocessing, and then the random basis matrix is ​​determined by using the existing feature-driven random matrix generation technology based on the characteristics of the medical data after intelligent preprocessing. , based on the random basis matrix The structure further generates a dynamic key vector , Indicates the number of keys; the dynamic key vector is used to control subsequent encryption transformations, and the generation formula is as follows:

[0097] ,

[0098] in, It is Dynamic key elements, , used for nonlinear transformation in subsequent encryption; It is the first step of intelligent preprocessing of medical data. data blocks; is a prime number, ensuring that the generated dynamic key element is within a limited range; is a random basis matrix The above formula generates a dynamic key vector with unpredictability through a combination of exponential, cosine and square root functions. , to accommodate data blocks of different dimensions. In the process, the values ​​of independent data blocks in different intelligently preprocessed medical data make the generation of each dynamic key element independent of each other, providing a dynamic key for the block encryption of the intelligently preprocessed medical data.

[0099] Next, the medical data after intelligent preprocessing Perform nonlinear transformation to ensure data complexity. Use dynamic key vector , construct the nonlinear mapping matrix , Indicates a The real matrix space of ; each element Generated by the following formula:

[0100] ,

[0101] in, Represents the nonlinear mapping matrix The Line The elements of the column. Through the combination of sine function, square term and inverse tangent function, the nonlinear mapping matrix Each element in has a highly nonlinear characteristic. The purpose of generating the nonlinear mapping matrix is ​​to use multivariate nonlinear transformation to make the intelligent preprocessed medical data produce complex interactions and disturbances, ensure the dense correlation between the dynamic key and the intelligent preprocessed medical data, and thus increase the encryption difficulty of the intelligent preprocessed medical data. Calculate the nonlinear mapping matrix After that, the medical data after intelligent preprocessing is processed by matrix multiplication. Perform nonlinear mapping to obtain nonlinearly transformed medical data :

[0102] ,

[0103] In the above formula, The operation introduces a nonlinear mapping matrix complexity to obtain medical data after nonlinear transformation , which includes the nonlinear perturbation characteristics of medical data after intelligent preprocessing, ensuring greater resistance to cracking in the subsequent encryption process.

[0104] Medical data after nonlinear transformation After the nonlinear perturbation is obtained, the hierarchical encryption stage is entered. The hierarchical encryption design aims to build a data protection barrier through multiple encryptions, making the cracking difficulty multiplied. Assuming the number of layers is , in In the layer, a dynamic key matrix is ​​introduced , to maintain the uniqueness of each layer of encryption. Define the dynamic key elements in any dynamic key matrix The formula is as follows:

[0105] ,

[0106] in, Indicates The dynamic key elements in the layer dynamic key matrix include multi-level exponentiation operations of the dynamic key elements; Indicates Dynamic Key Elements of Power; Represents the first Data blocks of Power; is a random basis matrix The random angles generated are used to enhance randomness. Layer ciphertext The calculation formula is:

[0107] ,

[0108] in, It is The ciphertext generated by the layer; For the Dynamic key matrix of layers; Represents the medical data after nonlinear transformation Since each layer of encryption uses a unique key matrix, and power encryption further amplifies the complexity of medical data, it ensures that each layer of encrypted medical data cannot be cracked by simple reverse deduction.

[0109] When the nonlinearly transformed medical data is After layer-by-layer encryption, the final ciphertext vector is obtained , the ciphertext vector is further encrypted through a composite random mapping to ensure the irreversibility and security of medical data. The random perturbation function of the composite random mapping is defined as:

[0110] ,

[0111] in, It is a random perturbation function of the composite random mapping, used to increase the perturbation of the final ciphertext; It is the input data or intermediate results that make the values ​​generated by the random perturbation function nonlinear and random.

[0112] Finally, the elements of the ciphertext vector Substitute the random perturbation function for composite random mapping:

[0113] ,

[0114] in, Represents the first ciphertext elements, i.e. the encrypted medical data; Indicates The ciphertext generated by the layer. Random perturbation function through composite random mapping The interaction between other ciphertext elements in the ciphertext vector makes each ciphertext element have independent random characteristics, making the encrypted medical data irreversible in data structure. The design of random mapping prevents the leakage of medical data characteristics, and even if it is intercepted, it is difficult to restore the intelligent pre-processed medical data through reverse deduction.

[0115] Furthermore, the encrypted medical data is synchronized using existing private blockchain technology to ensure that the encrypted medical data can be synchronized in different medical institutions to obtain encrypted and real-time synchronized medical data.

[0116] S3. When a medical data access request is received, the identity of the visitor is verified. After the verification is passed and it is determined that the visitor has access rights, the encrypted and real-time synchronized medical data is decrypted to obtain the decrypted medical data.

[0117] When a request for access to medical data is received, the identity of the visitor is verified through existing multi-factor authentication technology (such as two-factor authentication or three-factor authentication) to verify whether the visitor is a normal visitor. When the verification is passed, the visitor is determined to have access to specific medical data by searching the permission database (such as a role-based access control model) based on the visitor's identity. The query of permissions includes the visitor's role (such as doctor, nurse, manager), organization, and the user's data permission level; after confirming that the visitor has access to the requested medical data, the encrypted and real-time synchronized medical data is decrypted. The entire decryption process includes de-perturbation, layered inverse decryption, nonlinear inverse mapping, and dynamic key verification. Specifically, it includes removing the random perturbation added during encryption, and restoring the ciphertext to an unperturbed state by removing the nonlinear perturbation term introduced in the composite random mapping, ensuring that the decrypted data has no additional interference factors and can be accurately restored.

[0118] After de-disturbance is completed, the decryption process enters the layered reverse decryption stage. Starting from the last layer of encryption, the data is decrypted in reverse order layer by layer until it is restored to the unencrypted state. In the decryption of each layer, the encrypted medical data is gradually restored to the unencrypted state by using the inverse operation of the dynamic key matrix corresponding to that layer. Since the dynamic key matrix of each layer is dynamically generated and independent during encryption, the decryption operation of each layer is also unique, and only under the correct dynamic key matrix can the decrypted data be gradually returned to the unencrypted state.

[0119] When all layered decryption is completed, a nonlinear inverse mapping operation is performed on the medical data that has been decrypted to an unencrypted state. The decrypted data is restored to the original intelligently preprocessed medical data using the inverse transformation corresponding to the nonlinear mapping in the encryption process. The execution of the nonlinear inverse mapping ensures that the data structure and content are consistent with the state before encryption. Finally, the decrypted medical data is obtained, and medical information sharing is realized.

[0120] In summary, a medical information sharing method and system based on medical big data has been completed.

[0121] The order of the embodiments of the invention is for description only and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0122] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0123] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. A medical information sharing method based on medical big data, characterized in that: The following steps are involved: S1. Collect medical data in real time to obtain original medical data; perform intelligent preprocessing of the original medical data including data analysis, data cleaning, data format conversion and mapping, outlier detection and processing, data standardization and normalization to obtain medical data after intelligent preprocessing; A multi-level dynamic recursive anomaly detection algorithm is introduced to detect and process outliers for medical data that have completed format conversion and mapping. On the basis of multi-level data decomposition, the multi-level dynamic recursive anomaly detection algorithm performs anomaly detection through dynamic threshold generation, recursive clustering detection, anomaly score calculation and reverse verification steps. For the data decomposition in the multi-level dynamic recursive anomaly detection algorithm, medical data that have completed format conversion and mapping are hierarchically decomposed to generate two or more levels of medical data subsets. Based on the medical data subsets, the adaptive threshold required for the medical data subsets in each level is calculated through the dynamic threshold adaptive generation mechanism. In the recursive clustering detection process, the deviation is calculated based on the adaptive threshold to obtain the deviation score of the cluster. The specific formula is as follows: ， in, For the The first level Deviation scores for clustering of medical data subsets; Represents a subset of medical data The total amount of data; Indicates medical data that has completed format conversion and mapping The Medical data; It is The first level Medical data subset The mean of It is Hierarchical recursive clustering deviation factor; It is The first level Medical data subset The standard deviation of Indicates Recursive weighting factors of the layers; after each layer of cluster detection, calculate the abnormality score of the medical data that has completed format conversion and mapping, and recursively calculate based on the layered deviation score; based on the abnormality score, distinguish between normal medical data and abnormal medical data, and calculate the probability density function of abnormal medical data and normal medical data; S2. Encrypt the medical data after intelligent preprocessing to obtain encrypted medical data; synchronize the encrypted medical data to obtain encrypted and real-time synchronized medical data; S3. When a medical data access request is received, the identity of the visitor is verified. After the verification is passed and it is determined that the visitor has access rights, the encrypted and real-time synchronized medical data is decrypted to obtain the decrypted medical data.

2. The medical information sharing method based on medical big data according to claim 1 is characterized in that: The S1 specifically includes: The formula for calculating the adaptive threshold based on the dynamic threshold adaptive generation mechanism is as follows: ， in, It is The first level Medical data subset Adaptive threshold of Indicates The first level A subset of medical data; It is Weight factors for the layers.

3. The medical information sharing method based on medical big data according to claim 2 is characterized in that: The S1 specifically includes: In the recursive clustering detection process, the cluster deviation score is used to determine the possibility of abnormality of the medical data in the medical data subset; the deviation of each level of the medical data subset is calculated based on the current adaptive threshold to detect and strip off the abnormal medical data layer by layer, and the stripping results are recursively brought into the next level of clustering detection, forming a layer-by-layer recursive detection mode.

4. The medical information sharing method based on medical big data according to claim 3 is characterized in that: The S1 specifically includes: The calculation formula of abnormality score is as follows: ， in, Indicates medical data that has completed format conversion and mapping The Abnormality score of medical data; is the number of levels; For the The weight coefficient of the level is used to regulate the impact of different levels on the abnormality score; is the weight coefficient, ensuring the recursive increase and decrease of the abnormality score; the variable is the adaptive factor; Indicates medical data that has completed format conversion and mapping The Abnormality score of medical data.

5. The medical information sharing method based on medical big data according to claim 4 is characterized in that: The S1 specifically includes: In the reverse verification process, the recursive clustering detection results at each level are verified based on the probability density functions of abnormal medical data and normal medical data.

6. The medical information sharing method based on medical big data according to claim 1 is characterized in that: The S2 specifically includes: The medical data after intelligent preprocessing is encrypted using a dynamic layered composite encryption algorithm; in the process of implementing the dynamic layered composite encryption algorithm, the medical data after intelligent preprocessing is first processed to obtain the characteristics of the medical data after intelligent preprocessing; then a random basis matrix is ​​generated based on the characteristics of the medical data after intelligent preprocessing; a dynamic key vector is generated based on the random basis matrix; a nonlinear mapping matrix is ​​constructed based on the dynamic key vector; based on the nonlinear mapping matrix, the medical data after intelligent preprocessing is nonlinearly mapped through matrix multiplication to obtain the medical data after nonlinear transformation; the dynamic key matrix is ​​introduced to perform layered encryption on the medical data after nonlinear transformation to obtain a ciphertext vector; the ciphertext vector is subjected to compound random mapping through a random perturbation function to obtain the encrypted medical data; The random perturbation function is defined as: ， in, is the random perturbation function of the composite random mapping; is input data or intermediate results; The specific formula of composite random mapping is: ， in, Represents encrypted medical data; Random perturbation function representing a composite random mapping; Indicates The ciphertext generated after layer-by-layer encryption; It is The ciphertext generated after layer-by-layer encryption; Indicates the total number of layered encryption layers.

7. The medical information sharing method based on medical big data according to claim 1 is characterized in that: The S3 specifically includes: The encrypted and real-time synchronized medical data is decrypted, and the decryption process includes de-perturbation, layered inverse decryption, nonlinear inverse mapping, and dynamic key verification.

8. A medical information sharing system based on medical big data, applied to the medical information sharing method based on medical big data as claimed in claim 1, characterized in that: Includes the following parts: Data acquisition module, data intelligent processing module, data encryption and security module, data synchronization module, authorized access and decryption module; The data acquisition module is responsible for collecting medical data, obtaining raw medical data, and sending the raw medical data to the data intelligent processing module; The data intelligent processing module performs intelligent preprocessing on the original medical data and sends the intelligently preprocessed medical data to the data encryption and security module; The data encryption and security module encrypts the medical data after intelligent preprocessing and sends the encrypted medical data to the data synchronization module; The data synchronization module synchronizes the encrypted medical data to obtain encrypted and real-time synchronized medical data, and transmits the encrypted and real-time synchronized medical data to the authorized access and decryption module; The authorized access and decryption module verifies the identity of the visitor upon receiving a request for access to medical data. After the verification is passed and it is determined that the visitor has access rights, the encrypted and real-time synchronized medical data is decrypted to obtain the decrypted medical data.

Images