Method and system for dynamic desensitization of intermediate layer data stream

Abstract

The present invention relates to the field of computer vision technology, specifically to a method and system for dynamic desensitization of middle-layer data streams, comprising the following steps: accessing a virtual database with built-in ODBC and JDBC drivers, capturing input medical data streams, analyzing data content and labels, calculating the sensitivity index of each data field, determining the data sensitivity level based on the calculation results combined with the historical frequency of medical data access and user roles, and outputting a medical data sensitivity level association table by calculating the correlation between keyword density and access frequency in the field. In the present invention, by dynamically capturing medical data streams, desensitization rules are adjusted in real time according to sensitivity indicators and user behaviors, effectively improving the adaptability and efficiency of processing data, updating desensitization strategies in real time, increasing sensitivity responses to changing data streams, and ensuring the integrity and security of data during transmission and processing by continuously monitoring data stream structures and access patterns.

Description

Technical Field

[0001] The present invention relates to the field of computer vision technology, and in particular to a method and system for dynamic desensitization of intermediate layer data streams. Background Art

[0002] The field of data desensitization technology involves a series of methods and systems that aim to modify sensitive data that is processed, transmitted or stored, thereby preventing the data from leaking its original content in an unauthorized environment. This technology is often used in scenarios where personal privacy or corporate secrets need to be protected, such as financial services, health care, and public services. Data desensitization methods include but are not limited to data masking, falsification, randomization, and segmentation, which can ensure that even if the data is illegally accessed, its true information cannot be identified or used.

[0003] Among them, the dynamic desensitization method of the middle-layer data flow is crucial to protecting the privacy of patients, especially in the medical industry, while ensuring that medical service providers can access and analyze necessary data in compliance with regulations, and can perform desensitization in real time during the transmission of data from one system to another. Dynamic desensitization ensures the security and privacy of data during transmission, and supports medical institutions to perform data analysis and processing without exposing patients' sensitive information. Its main purpose is to improve the efficiency and flexibility of medical data processing while ensuring data security, and support medical institutions to improve service quality and operational efficiency.

[0004] Existing technologies have shown limitations in processing dynamically changing data streams, mainly because static desensitization rules cannot respond to changes in data access patterns in a timely manner, which may lead to insufficient data protection or reduced processing efficiency. In a demanding medical environment, this deficiency will hinder timely access to critical information and reduce the efficiency of medical decision-making. In actual operations, static methods often lack the ability to adapt to the dynamics of data stream access, making it difficult to support efficient data analysis and real-time monitoring, and may not be able to provide the necessary data support in emergency medical situations. In addition, fixed desensitization strategies are not adapted to the rapid changes in regulations and policies, limiting the flexibility of medical institutions in adjusting data processing and privacy protection, resulting in the failure to fully realize the potential of data utilization. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a method and system for dynamic desensitization of intermediate layer data streams.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: a method for dynamic desensitization of intermediate layer data stream, comprising the following steps:

[0007] S1: Access the virtual database with built-in ODBC and JDBC drivers, capture the input medical data flow, analyze the data content and labels, calculate the sensitivity index of each data field, and determine the data sensitivity level based on the calculation results combined with the historical frequency of medical data access and user roles. By calculating the correlation between the keyword density in the field and the access frequency, the medical data sensitivity level association table is output;

[0008] S2: Based on the medical data sensitivity level association table, analyze the sensitive content of the data field, map the sensitivity level with the desensitization rule, dynamically adjust the desensitization rule according to the mapping result combined with the characteristics of the medical data and the user access frequency, regularly update the desensitization strategy according to the data type and user access mode, and output the dynamic desensitization strategy configuration;

[0009] S3: Apply the dynamic desensitization strategy configuration to perform streaming data processing, perform buffer management according to the time window and flow of medical data inflow, adjust the residence time and processing order of data in the buffer, perform instant desensitization processing according to the new desensitization rules, and generate a streaming desensitized data snapshot;

[0010] S4: Based on the streaming desensitized data snapshot, analyze the structure of the data stream and the changes in access patterns, monitor the continuity of the desensitization processing and data integrity in real time, identify abnormal patterns and data leakage risks in the dynamic desensitization process, and output the dynamic desensitization monitoring records of the middle-layer medical data stream.

[0011] As a further solution of the present invention, the steps for calculating the sensitivity index of each data field are:

[0012] S111: Connect to the virtual database through ODBC and JDBC, synchronously input the medical data stream, read and classify the medical information and metadata tags, and generate a medical data field set;

[0013] S112: Perform content analysis on the medical data field set, using the formula:

[0014]

[0015] Calculating Information Entropy , evaluate the complexity and diversity of data fields, and generate data field information entropy index data, where, Representation field The probability of the target information appearing in is the diversity factor of information categories, increasing the weight adjustment of information categories, Represents a collection of medical data fields;

[0016] S113: Based on the information entropy index data of the data field and the density of sensitive words in the data field, a formula is used:

[0017]

[0018] Calculate the sensitivity index for each field , get the data field sensitivity index data, where, and is the adjustment factor, is the number of sensitive words, is the total number of words, represents the natural exponential function.

[0019] As a further solution of the present invention, the steps for obtaining the medical data sensitivity level association table are:

[0020] S121: Using the data field sensitivity index data, combined with historical access frequency and user role information, using a formula,

[0021]

[0022] Calculate a preliminary sensitivity level for each data field , and obtain the preliminary assessment results of the sensitivity level of the data field, where: and is a coefficient adjusted based on role and frequency, is the frequency of visits;

[0023] S122: Based on the preliminary evaluation result of the sensitivity level of the data field, the formula is used:

[0024]

[0025] Calculate field keyword density ,in, is the total number of words, is the number of keywords, is a scaling factor used to adjust the impact of keyword density;

[0026] S123: Based on the preliminary evaluation results of the data field sensitivity level and the field keyword density, the formula is used:

[0027]

[0028] Calculate the final medical data sensitivity level , output the medical data sensitivity level association table, where, and It is an influence coefficient used to balance the sensitivity level and keyword density.

[0029] As a further solution of the present invention, the steps of mapping the sensitivity level and the desensitization rule are:

[0030] S211: Based on the medical data sensitivity level association table, access and query, extract the sensitivity level of each data field, and obtain a list of each data field and its sensitivity level;

[0031] S212: Analyze each data field and its sensitivity level list, identify the sensitive content type in the data field, including personal information and medical records, and use the formula:

[0032]

[0033] Calculate the coverage ratio of sensitive content , generate sensitive content analysis results, where is the number of sensitive content items, is the total number of field items;

[0034] S213: According to the sensitive content analysis result, the sensitivity level is mapped to the corresponding desensitization rule, using the formula:

[0035]

[0036] Calculate weighted index of mapped values , generate a sensitivity level and desensitization rule mapping table, where, For the The weight of desensitization rules corresponding to sensitive content. Represents the total number of classes of sensitive content.

[0037] As a further solution of the present invention, the steps of obtaining the dynamic desensitization strategy configuration are:

[0038] S221: Based on the sensitivity level and desensitization rule mapping table, using the formula:

[0039]

[0040] Calculate the quantitative evaluation value of the preliminary desensitization rule configuration , generate the initial configuration of desensitization rules, where Adjust the sensitivity of dynamic adjustment, is the frequency of visits;

[0041] S222: Based on the preliminary configuration of the dynamic desensitization rules, the desensitization rules are periodically re-evaluated and updated according to the real-time data of user access patterns and data update frequency, using the formula,

[0042]

[0043] Calculate the updated desensitization policy configuration value , generate desensitization rule update configuration, where, is the adjustment factor, is the change in access frequency;

[0044] S223: Apply the desensitization rule to update the configuration and evaluate the effect using the formula:

[0045]

[0046] Calculate the quantitative evaluation value of desensitization effect , and dynamically adjust to output dynamic desensitization strategy configuration, where: It is the effect evaluation coefficient, which quantifies the effectiveness of strategy execution.

[0047] As a further solution of the present invention, the steps of obtaining the streaming desensitized data snapshot are:

[0048] S311: Apply the dynamic desensitization strategy configuration, and use the formula according to the time window and flow of medical data inflow,

[0049]

[0050] Calculate buffer record processing speed , basic configuration of output buffer, where, represents the time window of data inflow, is the flow rate, which means the number of data records per unit time;

[0051] S312: Based on the basic configuration of the buffer and in combination with the dynamic characteristics of the real-time data stream, a formula is used:

[0052]

[0053] Calculate and output the adjusted data residence time ,in, Used to adjust the residence time changes caused by traffic changes to avoid processing delays caused by sudden traffic increases;

[0054] S313: Desensitizing the medical data using the adjusted data residence time, using the formula:

[0055]

[0056] Calculating the sensitivity of immediate desensitization treatment , generate a streaming desensitized data snapshot, where, Used to adjust the sensitivity of desensitization intensity relative to residence time.

[0057] As a further solution of the present invention, the steps for obtaining the dynamic desensitization monitoring record of the intermediate layer medical data flow are:

[0058] S411: Collect and analyze the streaming desensitized data snapshots, identify changes in the structure and access patterns of the data stream, and use the formula:

[0059]

[0060] Calculate the average comprehensive index of each data point in the data stream , generate data flow structure and access pattern analysis records, where, Representative The access frequency of data points, Representative data points The weight of the position in the data stream, is the total number of data points;

[0061] S412: Based on the data stream structure and access pattern analysis records, the continuity and integrity of the desensitization processing in the data stream are monitored in real time, using the formula,

[0062]

[0063] Calculate the average integrity score of the data block , output continuity and integrity monitoring records, where is the coefficient for adjusting the sensitivity of continuity monitoring, is the threshold factor for integrity detection, used to adjust the impact of integrity score, For the The integrity score of the data block, is the total number of data blocks;

[0064] S413: Analyze the records and the continuity and integrity monitoring records according to the data flow structure and access pattern, identify abnormal patterns and potential data leakage risks in the dynamic desensitization process, and use the formula,

[0065]

[0066] Calculate the risk assessment value of the anonymized medical data stream , Generate dynamic desensitized monitoring records of medical data flows in the middle layer, where is the adjustment factor for risk assessment, is the abnormal pattern sensitivity factor, which determines the impact of abnormal data on the overall risk assessment. Indicates The abnormal indicators of the monitoring points, is the total number of monitoring points.

[0067] The middle-layer data stream dynamic desensitization system includes:

[0068] The data capture and analysis module connects to the virtual database through ODBC and JDBC, synchronously inputs the medical data stream and calculates the information entropy, combines the density of sensitive words in the data field, calculates the sensitivity index of each field, and obtains the data field sensitivity index data;

[0069] The sensitivity level assessment module uses the data field sensitivity index data, combined with historical access frequency and user role information, to conduct a preliminary assessment of the data field sensitivity level and calculate the field keyword density, and output a medical data sensitivity level association table;

[0070] The desensitization rule mapping module accesses and queries the medical data sensitivity level association table, identifies the sensitive content type in the data field, calculates the coverage ratio of the sensitive content, maps the sensitivity level with the corresponding desensitization rule, and generates a sensitivity level and desensitization rule mapping table;

[0071] The desensitization strategy update module periodically re-evaluates and updates the desensitization rules based on the sensitivity level and desensitization rule mapping table, according to the real-time data of user access mode and data update frequency, and performs effect evaluation to output dynamic desensitization strategy configuration;

[0072] The data stream desensitization module applies the dynamic desensitization strategy configuration, calculates the buffer record processing speed according to the time window and flow of the medical data inflow, calculates the adjusted data residence time and desensitizes the medical data in combination with the dynamic characteristics of the real-time data stream, and generates a streaming desensitized data snapshot;

[0073] The monitoring and evaluation module collects and analyzes the streaming desensitized data snapshots, identifies changes in the structure and access patterns of the data stream, performs continuity and integrity monitoring, calculates the risk assessment value of the desensitized medical data stream, and generates dynamic desensitization monitoring records for the intermediate layer medical data stream.

[0074] Compared with the prior art, the advantages and positive effects of the present invention are:

[0075] In the present invention, by dynamically capturing medical data streams, desensitization rules are adjusted in real time according to sensitivity indicators and user behaviors, effectively improving the adaptability and efficiency of data processing, allowing precise control of data access requirements in different scenarios, adapting to the specific needs of different users, and updating desensitization strategies in real time. The sensitivity response to changing data streams is increased to ensure immediate data protection and compliance. Through continuous monitoring of data stream structures and access patterns, data security is further enhanced, potential security threats are discovered and responded to in a timely manner, and the integrity and security of data during transmission and processing are ensured. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 It is a schematic diagram of the steps of the present invention;

[0077] Figure 2 A calculation flow chart of the sensitivity index of each data field of the present invention;

[0078] Figure 3 This is a flowchart for obtaining the medical data sensitivity level association table of the present invention;

[0079] Figure 4 It is a mapping flow chart of the sensitivity level and desensitization rules of the present invention;

[0080] Figure 5 A flowchart for obtaining the configuration of the dynamic desensitization strategy of the present invention;

[0081] Figure 6 The following is a flow chart of obtaining a snapshot of streaming desensitized data according to the present invention;

[0082] Figure 7 The present invention is a flowchart for obtaining the dynamic desensitization monitoring record of the intermediate layer medical data flow. DETAILED DESCRIPTION

[0083] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0084] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0085] See also Figure 1 ,The middle layer data stream dynamic desensitization method includes the following steps:

[0086] S1: Access the virtual database with built-in ODBC and JDBC drivers, capture the input medical data flow, analyze the data content and labels, calculate the sensitivity index of each data field, and determine the data sensitivity level based on the calculation results combined with the historical frequency of medical data access and user roles. By calculating the correlation between the keyword density in the field and the access frequency, the medical data sensitivity level association table is output;

[0087] S2: Based on the medical data sensitivity level association table, analyze the sensitive content of the data field, map the sensitivity level with the desensitization rules, dynamically adjust the desensitization rules according to the mapping results combined with the characteristics of the medical data and the user access frequency, regularly update the desensitization strategy according to the data type and user access mode, and output the dynamic desensitization strategy configuration;

[0088] S3: Apply dynamic desensitization policy configuration to process streaming data, manage the buffer according to the time window and flow of medical data inflow, adjust the data residence time in the buffer and the processing order, perform instant desensitization processing according to the new desensitization rules, and generate a streaming desensitized data snapshot;

[0089] S4: Based on the streaming desensitized data snapshot, analyze the structure and access pattern changes of the data stream, monitor the continuity and data integrity of the desensitization process in real time, identify abnormal patterns and data leakage risks in the dynamic desensitization process, and output the dynamic desensitization monitoring records of the middle-layer medical data stream.

[0090] The medical data sensitivity level association table includes the sensitivity classification record set, access rights and keyword density. The dynamic desensitization policy configuration includes the desensitization level, update frequency and policy effectiveness evaluation results. The streaming desensitized data snapshot includes the data snapshot time point, data format and desensitization quality analysis results. The middle-layer medical data stream dynamic desensitization monitoring record includes the monitoring timestamp and data integrity verification results.

[0091] See also Figure 2 , the calculation steps of the sensitivity index of each data field are:

[0092] S111: Connect to the virtual database through ODBC and JDBC, synchronously input the medical data stream, read and classify the medical information and metadata tags, and generate a medical data field set;

[0093] S112: Perform content analysis on the medical data field set using the formula,

[0094]

[0095] Calculating Information Entropy , evaluate the complexity and diversity of data fields, and generate data field information entropy index data, where, Representation field The probability of the target information appearing in is the diversity factor of information categories, increasing the weight adjustment of information categories, Represents a collection of medical data fields;

[0096] S113: Based on the information entropy index data of the data field and the density of sensitive words in the data field, the formula is used:

[0097]

[0098] Calculate the sensitivity index for each field , get the data field sensitivity index data, where, and is the adjustment factor, is the number of sensitive words, is the total number of words, represents the natural exponential function.

[0099] The database was successfully connected through ODBC and JDBC drivers, and 100 medical data records were captured.

[0100] The data fields of interest include the following field information:

[0101]

[0102] Fields Appears 50 times, field Appears 30 times, field Occurs 20 times.

[0103] Total calculation times . Therefore, the probability of each class is:

[0104]

[0105] Set the diversity factor for information categories .

[0106] Substituting into the formula, we get:

[0107]

[0108]

[0109]

[0110]

[0111] Information Entropy It indicates that the data field has a medium level of information diversity and complexity. The higher the information entropy value, the greater the uncertainty of the information category in the data, which usually indicates that the data field contains rich and diverse information content.

[0112] For example, among 100 words, there are 10 sensitive words. Set the weight and .

[0113] Substituting into the formula, we get:

[0114]

[0115]

[0116]

[0117] Sensitivity index Indicates that the data field has medium sensitivity in terms of security or privacy. A higher sensitivity index means that the data field requires more care and protection when it is processed and shared.

[0118] See also Figure 3 ,The steps for obtaining the medical data sensitivity level association table are:

[0119] S121: Using data field sensitivity indicator data, combined with historical access frequency and user role information, a formula is used to

[0120]

[0121] Calculate a preliminary sensitivity level for each data field , and obtain the preliminary assessment results of the sensitivity level of the data field, where: and is a coefficient adjusted based on role and frequency, is the frequency of visits;

[0122] S122: Based on the preliminary assessment results of the sensitivity level of the data field, the formula is used.

[0123]

[0124] Calculate field keyword density ,in, is the total number of words, is the number of keywords, is a scaling factor used to adjust the impact of keyword density;

[0125] S123: Combined with the preliminary assessment results of the data field sensitivity level and the field keyword density, the formula is used.

[0126]

[0127] Calculate the final medical data sensitivity level , output the medical data sensitivity level association table, where, and It is an influence coefficient used to balance the sensitivity level and keyword density.

[0128] Combined with the above calculation results, the sensitivity index , historical access frequency (i.e. the field is accessed once in every five accesses).

[0129] Setting weights and .

[0130] Substituting into the formula, we get:

[0131]

[0132]

[0133] Initial sensitivity level Indicates that this data field has a high sensitivity in medical databases. This value indicates that additional security measures need to be implemented to protect this data, especially if it is frequently accessed.

[0134] For example, the number of keywords , total number of words , scaling factor .

[0135] Substituting into the formula, we get:

[0136]

[0137] Keyword density This means that there is approximately one keyword in every 20 words. This low density value suggests that although the field contains sensitive words, their frequency of occurrence is not high, which may affect the evaluation of the field's sensitivity level.

[0138] Combined with the above calculation results, the initial sensitivity level , keyword density .

[0139] Setting weights and ,but:

[0140]

[0141]

[0142] Final sensitivity level This clearly indicates that this field needs to be considered highly sensitive in the medical data environment and requires special attention and protection. The value is higher than the initial assessment, which means that the need for data protection has increased after considering the keyword density.

[0143] See also Figure 4 , the mapping steps between sensitivity level and desensitization rules are:

[0144] S211: Based on the medical data sensitivity level association table, access and query, extract the sensitivity level of each data field, and obtain a list of each data field and its sensitivity level;

[0145] S212: Analyze each data field and its sensitivity level list, identify the sensitive content types in the data field, including personal information and medical records, and use the formula,

[0146]

[0147] Calculate the coverage ratio of sensitive content , generate sensitive content analysis results, where is the number of sensitive content items, is the total number of field items;

[0148] S213: Based on the sensitive content analysis results, the sensitivity level is mapped to the corresponding desensitization rules using the formula:

[0149]

[0150] Calculate weighted index of mapped values , generate a sensitivity level and desensitization rule mapping table, where, For the The weight of desensitization rules corresponding to sensitive content. Represents the total number of classes of sensitive content.

[0151] First, the medical data sensitivity level association table is accessed and queried. The query can be completed directly through SQL, and a list of each data field and its sensitivity level is output.

[0152] When checking a specific field, I found:

[0153] Number of sensitive content items , total number of field items .

[0154] Substituting into the formula, we get:

[0155]

[0156] This means that 30% of the content in the data field is marked as sensitive. The 30% coverage ratio indicates that a considerable portion of the data in the field contains sensitive information, which is an important basis for desensitization operations.

[0157] For example, there are two types of sensitive content to consider, with the following weights:

[0158] , corresponding to personal information; , corresponding to the medical history.

[0159] Sensitive content coverage ratio , sensitive content coverage ratio .

[0160] but:

[0161]

[0162]

[0163]

[0164] Calculated mapping value It is a weighted indicator used to guide how to apply desensitization rules to data fields. A low value indicates that although sensitive content exists, its overall sensitivity is not very high and moderate desensitization measures are required.

[0165] See also Figure 5 , the steps to obtain the dynamic desensitization policy configuration are:

[0166] S221: Based on the sensitivity level and desensitization rule mapping table, the formula is used.

[0167]

[0168] Calculate the quantitative evaluation value of the preliminary desensitization rule configuration , generate the initial configuration of desensitization rules, where Adjust the sensitivity of dynamic adjustment, is the frequency of visits;

[0169] S222: Based on the preliminary configuration of dynamic desensitization rules, the desensitization rules are periodically re-evaluated and updated according to the real-time data of user access patterns and data update frequency, using the formula,

[0170]

[0171] Calculate the updated desensitization policy configuration value , generate desensitization rule update configuration, where, is the adjustment factor, is the change in access frequency;

[0172] S223: Apply the desensitization rules to update the configuration and evaluate the effect using the formula:

[0173]

[0174] Calculate the quantitative evaluation value of desensitization effect , and dynamically adjust to output dynamic desensitization strategy configuration, where: It is the effect evaluation coefficient, which quantifies the effectiveness of strategy execution.

[0175] Set access frequency , mapping value , sensitivity adjustment factor .

[0176] The calculation process is as follows:

[0177]

[0178]

[0179]

[0180] Get the quantitative evaluation value of the preliminary desensitization rule configuration It indicates that under the current parameter settings, desensitization measures need to be maintained at a moderate intensity to balance data protection and accessibility.

[0181] Combined with the above calculation results , access frequency change , adjustment coefficient , substituting into the formula, we get:

[0182]

[0183]

[0184]

[0185] Updated desensitization policy configuration value (an increase from 0.5), indicating that the intensity of desensitization measures has been increased to cope with the increased frequency of data access and potential security risks.

[0186] Updated desensitization policy configuration value , effect evaluation coefficient .

[0187] Substituting into the formula, we get:

[0188]

[0189]

[0190] Output desensitization effect quantitative evaluation value The desensitization effect after implementation was quantified, indicating that the enhanced desensitization measures effectively improved the protection level of data.

[0191] See also Figure 6 , the steps to obtain the streaming desensitized data snapshot are:

[0192] S311: Apply dynamic desensitization strategy configuration, according to the time window and flow of medical data inflow, using the formula,

[0193]

[0194] Calculate buffer record processing speed , basic configuration of output buffer, where, represents the time window of data inflow, is the flow rate, which means the number of data records per unit time;

[0195] S312: Based on the basic configuration of the buffer, combined with the dynamic characteristics of real-time data flow, the formula is used.

[0196]

[0197] Calculate and output the adjusted data residence time ,in, Used to adjust the residence time changes caused by traffic changes to avoid processing delays caused by sudden traffic increases;

[0198] S313: Use the adjusted data residence time to desensitize the medical data using the formula,

[0199]

[0200] Calculating the sensitivity of immediate desensitization treatment , generate a streaming desensitized data snapshot, where, Used to adjust the sensitivity of desensitization intensity relative to residence time.

[0201] Time window Seconds, flow Records / second, substitute into the formula, get:

[0202]

[0203] The results show that the buffer has a processing capacity of 1,200 records per second, which means that the system can efficiently handle a flow of up to 72,000 data records per minute and maintain the continuity and stability of data processing.

[0204] Combined with the above calculation results ,flow .

[0205] Substituting into the formula, we get:

[0206]

[0207]

[0208]

[0209] The result shows that each record stays in the buffer for 0.001154 seconds. The adjusted stay time ensures that the buffer will not be overloaded during high traffic while maintaining the real-time nature of data processing.

[0210] , combined with the above calculation results seconds, we get:

[0211]

[0212] This indicates that the sensitivity of the immediate desensitization process is relatively high and can react in a short time. This indicates that the desensitization process still maintains a good balance after dynamic adjustment and adapts to the changing data flow conditions.

[0213] See also Figure 7 ,The steps for obtaining the dynamic desensitization monitoring records of the middle-layer medical data flow are:

[0214] S411: Collect and analyze streaming desensitized data snapshots, identify changes in the structure and access patterns of data streams, and use formulas to

[0215]

[0216] Calculate the average comprehensive index of each data point in the data stream , generate data flow structure and access pattern analysis records, where, Representative The access frequency of data points, Representative data points The weight of the position in the data stream, is the total number of data points;

[0217] S412: Based on the data stream structure and access pattern analysis records, real-time monitoring of the continuity and integrity of the desensitization processing in the data stream, using the formula,

[0218]

[0219] Calculate the average integrity score of the data block , output continuity and integrity monitoring records, where is the coefficient for adjusting the sensitivity of continuity monitoring, is the threshold factor for integrity detection, used to adjust the impact of integrity score, For the The integrity score of the data block, is the total number of data blocks;

[0220] S413: Analyze records and monitor records for continuity and integrity based on data flow structure and access patterns, identify abnormal patterns and potential data leakage risks during dynamic desensitization, and use formulas to

[0221]

[0222] Calculate the risk assessment value of the anonymized medical data stream , Generate dynamic desensitized monitoring records of medical data flows in the middle layer, where is the adjustment factor for risk assessment, is the abnormal pattern sensitivity factor, which determines the impact of abnormal data on the overall risk assessment. Indicates The abnormal indicators of the monitoring points, is the total number of monitoring points.

[0223] For example, there are 3 data points with access frequencies of 10, 15, and 5, and position weights of 0.5, 0.3, and 0.2, respectively. Then:

[0224]

[0225]

[0226] What you get It shows that after weighting the access frequency and location importance, the average comprehensive index of each data point in the data stream is 3.5. This helps to understand which data points are the most frequently interacted and important, thus providing a basis for optimizing the data stream structure and access strategy.

[0227] For example, there are 3 data blocks with integrity scores of 0.8, 0.9, and 0.7 respectively. and .

[0228] Substituting into the formula, we get:

[0229]

[0230]

[0231] It means that under the influence of the adjustment factor, the average integrity score of the overall data block is adjusted to 2.4. The high score indicates that the desensitization process has good continuity and data integrity.

[0232] For example, the abnormal indicators of three monitoring points are 0.3, 0.2, and 0.4. and ,have to:

[0233]

[0234]

[0235]

[0236] It indicates that when data anomalies are taken into account, the risk assessment value of the overall data flow is 1.287, indicating that there is a certain risk of data leakage.

[0237] The middle-layer data stream dynamic desensitization system includes:

[0238] The data capture and analysis module connects to the virtual database through ODBC and JDBC, synchronously inputs the medical data stream and calculates the information entropy, combines the density of sensitive words in the data field, calculates the sensitivity index of each field, and obtains the data field sensitivity index data;

[0239] The sensitivity level assessment module uses data field sensitivity index data, combined with historical access frequency and user role information, to conduct a preliminary assessment of the data field sensitivity level and calculate the field keyword density, and output a medical data sensitivity level association table;

[0240] The desensitization rule mapping module accesses and queries the medical data sensitivity level association table, identifies the sensitive content type in the data field, calculates the coverage ratio of the sensitive content, maps the sensitivity level with the corresponding desensitization rule, and generates a sensitivity level and desensitization rule mapping table;

[0241] The desensitization policy update module periodically re-evaluates and updates the desensitization rules based on the sensitivity level and desensitization rule mapping table, according to the real-time data of user access patterns and data update frequency, and evaluates the effects to output dynamic desensitization policy configuration;

[0242] The data stream desensitization module applies dynamic desensitization strategy configuration, calculates the buffer record processing speed according to the time window and flow of medical data inflow, and calculates the adjusted data residence time and desensitizes the medical data in combination with the dynamic characteristics of real-time data streams, generating a streaming desensitized data snapshot;

[0243] The monitoring and evaluation module collects and analyzes snapshots of streaming desensitized data, identifies changes in the structure and access patterns of data streams, performs continuity and integrity monitoring, calculates the risk assessment value of the desensitized medical data stream, and generates dynamic desensitized monitoring records for the middle-layer medical data stream.

[0244] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. The method for dynamic desensitization of intermediate layer data streams is characterized by: The following steps are involved: Access the virtual database with built-in ODBC and JDBC drivers, capture the input medical data flow, analyze the data content and labels, calculate the sensitivity index of each data field, and determine the data sensitivity level based on the calculation results combined with the historical frequency of medical data access and user roles. By calculating the correlation between the keyword density in the field and the access frequency, output the medical data sensitivity level association table; Based on the medical data sensitivity level association table, the sensitive content of the data field is analyzed, the sensitivity level is mapped with the desensitization rule, and the desensitization rule is dynamically adjusted according to the mapping result combined with the characteristics of the medical data and the user access frequency. The desensitization strategy is regularly updated according to the data type and user access mode, and the dynamic desensitization strategy configuration is output; Apply the dynamic desensitization strategy configuration to perform streaming data processing, perform buffer management according to the time window and flow of medical data inflow, adjust the residence time and processing order of data in the buffer, perform instant desensitization processing according to the new desensitization rules, and generate a streaming desensitized data snapshot; Based on the streaming desensitized data snapshot, analyze the structure and access mode changes of the data stream, monitor the continuity and data integrity of the desensitization process in real time, identify abnormal patterns and data leakage risks in the dynamic desensitization process, and output the dynamic desensitization monitoring record of the middle-layer medical data stream; The calculation steps of the sensitivity index of each data field are: Connect to the virtual database through ODBC and JDBC, synchronously input medical data streams, read and classify medical information and metadata tags, and generate medical data field sets; The medical data field set is subjected to content analysis using the formula: Calculating Information Entropy , evaluate the complexity and diversity of data fields, and generate data field information entropy index data, where, Representation field The probability of the target information appearing in is the diversity factor of information categories, increasing the weight adjustment of information categories, Represents a collection of medical data fields; Based on the information entropy index data of the data field and the density of sensitive words in the data field, the formula is adopted: Calculate the sensitivity index for each field , get the data field sensitivity index data, where, and is the adjustment factor, is the number of sensitive words, is the total number of words, represents the natural exponential function; The steps for obtaining the medical data sensitivity level association table are as follows: Using the data field sensitivity indicator data, combined with historical access frequency and user role information, the formula is adopted: Calculate a preliminary sensitivity level for each data field , and obtain the preliminary assessment results of the sensitivity level of the data field, where: and is a coefficient adjusted based on role and frequency, is the frequency of visits; Based on the preliminary assessment results of the sensitivity level of the data field, the formula is used: Calculate field keyword density ,in, is the total number of words, is the number of keywords, is a scaling factor used to adjust the impact of keyword density; Combined with the preliminary assessment results of the sensitivity level of the data field and the field keyword density, the formula is used: Calculate the final medical data sensitivity level , output the medical data sensitivity level association table, where, and It is the influence coefficient used to balance the sensitivity level and keyword density; The steps for obtaining the streaming desensitized data snapshot are: Apply the dynamic desensitization strategy configuration, according to the time window and flow of medical data inflow, use the formula, Calculate buffer record processing speed , basic configuration of output buffer, where, represents the time window of data inflow, is the flow rate, which means the number of data records per unit time; Based on the basic configuration of the buffer, combined with the dynamic characteristics of real-time data flow, the formula is adopted: Calculate and output the adjusted data residence time ,in, Used to adjust the residence time changes caused by traffic changes to avoid processing delays caused by sudden traffic increases; The medical data is desensitized using the adjusted data residence time, using the formula: Calculating the sensitivity of immediate desensitization treatment , generate a streaming desensitized data snapshot, where, for adjusting the sensitivity of desensitization intensity relative to dwell time; The steps for obtaining the dynamic desensitization monitoring record of the intermediate layer medical data flow are: Collect and analyze the streaming desensitized data snapshots, identify changes in the structure and access patterns of the data stream, and use the formula, Calculate the average comprehensive index of each data point in the data stream , generate data flow structure and access pattern analysis records, where, Representative The access frequency of data points, Representative data points The weight of the position in the data stream, is the total number of data points; Based on the data stream structure and access pattern analysis records, the continuity and integrity of the desensitization processing in the data stream are monitored in real time, using the formula, Calculate the average integrity score of the data block , output continuity and integrity monitoring records, where is the coefficient for adjusting the sensitivity of continuity monitoring, is the threshold factor for integrity detection, used to adjust the impact of integrity score, For the The integrity score of the data block, is the total number of data blocks; According to the data flow structure and access pattern analysis records and continuity and integrity monitoring records, abnormal patterns and potential data leakage risks in the dynamic desensitization process are identified, and the formula is used. Calculate the risk assessment value of the anonymized medical data stream , Generate dynamic desensitized monitoring records of medical data flows in the middle layer, where is the adjustment factor for risk assessment, is the abnormal pattern sensitivity factor, which determines the impact of abnormal data on the overall risk assessment. Indicates The abnormal indicators of the monitoring points, is the total number of monitoring points.

2. The method for dynamic desensitization of intermediate layer data stream according to claim 1, characterized in that: The steps of mapping the sensitivity level and the desensitization rule are as follows: Based on the medical data sensitivity level association table, access and query are performed to extract the sensitivity level of each data field, and obtain a list of each data field and its sensitivity level; Analyze each of the data fields and their sensitivity level lists, identify the sensitive content types in the data fields, including personal information and medical records, and use the formula, Calculate the coverage ratio of sensitive content , generate sensitive content analysis results, where is the number of sensitive content items, is the total number of field items; According to the sensitive content analysis results, the sensitivity level is mapped to the corresponding desensitization rules, using the formula: Calculate weighted index of mapped values , generate a sensitivity level and desensitization rule mapping table, where, For the The weight of desensitization rules corresponding to sensitive content. Represents the total number of classes of sensitive content.

3. The method for dynamic desensitization of intermediate layer data stream according to claim 2, characterized in that: The steps for obtaining the dynamic desensitization strategy configuration are: Based on the sensitivity level and desensitization rule mapping table, the formula is adopted: Calculate the quantitative evaluation value of the preliminary desensitization rule configuration , generate the initial configuration of desensitization rules, where Adjust the sensitivity of dynamic adjustment, is the frequency of visits; Based on the preliminary configuration of the dynamic desensitization rules, the desensitization rules are periodically re-evaluated and updated according to the real-time data of user access patterns and data update frequency, using the formula, Calculate the updated desensitization policy configuration value , generate desensitization rule update configuration, where, is the adjustment factor, is the change in access frequency; Apply the desensitization rules to update the configuration and evaluate the effect using the formula: Calculate the quantitative evaluation value of desensitization effect , and dynamically adjust to output dynamic desensitization strategy configuration, where: It is the effect evaluation coefficient, which quantifies the effectiveness of strategy execution.

4. The middle layer data flow dynamic desensitization system is characterized by: According to the method for dynamic desensitization of intermediate layer data streams according to any one of claims 1 to 3, the system comprises: The data capture and analysis module connects to the virtual database through ODBC and JDBC, synchronously inputs the medical data stream and calculates the information entropy, combines the density of sensitive words in the data field, calculates the sensitivity index of each field, and obtains the data field sensitivity index data; The sensitivity level assessment module uses the data field sensitivity index data, combined with historical access frequency and user role information, to conduct a preliminary assessment of the data field sensitivity level and calculate the field keyword density, and output a medical data sensitivity level association table; The desensitization rule mapping module accesses and queries the medical data sensitivity level association table, identifies the sensitive content type in the data field, calculates the coverage ratio of the sensitive content, maps the sensitivity level with the corresponding desensitization rule, and generates a sensitivity level and desensitization rule mapping table; The desensitization strategy update module periodically re-evaluates and updates the desensitization rules based on the sensitivity level and desensitization rule mapping table, according to the real-time data of user access mode and data update frequency, and performs effect evaluation to output dynamic desensitization strategy configuration; The data stream desensitization module applies the dynamic desensitization strategy configuration, calculates the buffer record processing speed according to the time window and flow of the medical data inflow, calculates the adjusted data residence time and desensitizes the medical data in combination with the dynamic characteristics of the real-time data stream, and generates a streaming desensitized data snapshot; The monitoring and evaluation module collects and analyzes the streaming desensitized data snapshots, identifies changes in the structure and access patterns of the data stream, performs continuity and integrity monitoring, calculates the risk assessment value of the desensitized medical data stream, and generates dynamic desensitization monitoring records for the intermediate layer medical data stream.

Images