CN116884556B - A secure sharing platform for medical data based on inline blockchain - Google Patents

Abstract

The invention discloses a medical data security sharing platform based on inline blockchain, which includes a data interface module, a sharing function module, a resource feedback module and a resource supply pool, wherein the data interface module is used to connect local user equipment, and It is configured with a device vulnerability detection function; the sharing function module accesses the detected device and performs risk detection on the data to be shared; it uploads data summary information from local users to the outer blockchain; it uploads data information from local users to the inner area. block chain; and when the requesting user initiates a sharing request on the outer block chain, the inner block chain is linked to start an automated sharing task based on the sharing request; the resource feedback module is used to realize data sharing transaction records and transaction traceability, and to realize data Comprehensive maintenance of the entire sharing process, construction of resource attack pools, and design of intelligent attack detection optimization methods achieve significant improvements in model accuracy, efficiency and other performance.

Description

A secure sharing platform for medical data based on inline blockchain

Technical field

The invention relates to the technical field of data security sharing, specifically a medical data security sharing platform based on inline blockchain.

Background technique

Medical research and healthcare require the use of large amounts of data, such as patient medical records, imaging data, genetic data, etc.; since these data are usually scattered in various medical institutions, research institutions and data warehouses, it is difficult to utilize them effectively; in order to To solve this problem, medical data sharing came into being; by sharing data, medical researchers and health care workers can better understand the occurrence and development of diseases, develop new diagnostic methods and treatment plans, thereby improving the efficiency of disease diagnosis and treatment. Accuracy and success rate; there are also some risks in medical data sharing. Sharing of sensitive personal information may lead to privacy leaks, such as medical records, genetic data, etc.; traditional data sharing platforms use encryption technology, access control and attack detection to ensure data security. and privacy. Traditional homomorphic encryption technology requires the platform to have high hardware performance. Access control cannot achieve traceability of the sharing process. Intelligent attack detection models still have problems such as low detection accuracy and low efficiency. To this end, design a traceable, A safe and efficient data sharing platform is a hot issue in current research.

Contents of the invention

The purpose of the present invention is to provide a secure sharing platform for medical data based on inline blockchain, which can effectively solve the above problems.

In order to solve the above technical problems, the present invention adopts the following technical solution: a medical data security sharing platform based on inline blockchain, the platform includes:

The data interface module is used to connect to local user equipment and is configured with device vulnerability detection function;

The shared function module accesses the detected device and executes:

Conduct risk detection on data to be shared;

Upload data summary information from local users to the outer blockchain;

Upload data information from local users to the inner blockchain;

And when the requesting user initiates a sharing request on the outer blockchain, the inner blockchain is linked to start an automated sharing task based on the sharing request;

The resource feedback module is used to realize data sharing transaction records and transaction traceability.

Preferably, when performing the device vulnerability detection function, the user device is tested for vulnerabilities by starting an intelligent attack detection model. If a vulnerability is found, an alarm is initiated, and the network device vulnerability attack detection result is defined as N_a, and the first layer alarm threshold is defined as N, like , an alarm is initiated.

Preferably, the risk detection is: start the intelligent attack detection model to detect the data to be uploaded, initiate an alarm if a risk is found, define the detection result of the data to be uploaded as U_d, define the second layer alarm threshold as U, if , an alarm is initiated.

Preferably, in the sharing function module, the outer blockchain is also used to record shared transaction information and shared result information, and supports access by all users on the chain; the inner blockchain only supports automated operations.

Preferably, the resource feedback module is also configured to require the requesting end user to upload local data and data operation results, and activate an intelligent attack detection model to implement attack detection on the uploaded results.

Further preferably, when implementing attack detection on the uploaded results, if a risk is found, an alarm will be initiated; each level of alarm requires the user to manually confirm whether to cancel the alarm and continue the task and define the detection result of the data to be uploaded as U_r, and define the second level alarm threshold as R, if , an alarm is initiated.

Preferably, the platform further includes a resource supply pool, which is used to store classified attack detection results and to optimize and train an intelligent attack detection model.

Preferably, the optimization training of the intelligent attack detection model is specifically as follows:

S1. Obtain the known attack characteristic data set T from the resource supply pool and copy it to form a copy set. ;

S2. According to the designed attack characteristic changing method, modify the copy set Perform operations until all features change to form a mutation set/> ;

S3. Collect the mutations Merge with known attack feature data set T to form a new attack feature set/> ;

S4. Use optimization methods to select parameters for the intelligent attack detection model to obtain the optimal parameter model;

S5. Utilize attack signature set to train the optimal parameter model.

Further preferably, the attack characteristic changing method is: using copy set , Chebyshev distance formula, calculate the distance threshold; copy the set/> , the distance threshold is substituted into the Chebyshev distance formula to obtain multiple sets of unknown attack features, and then the feature variant direction and variant range are calculated and the multiple sets of unknown attack features are screened based on the calculation results, and finally a mutation set of unknown attacks that meets the requirements is obtained/> .

Further preferably, an optimization method is used for parameter selection, and the optimal parameter model is obtained as follows:

S1. Optimize the objective function based on the model detection accuracy. When the detection rate is the highest, the corresponding parameter set is the optimal parameter set;

S2. Given the parameter variation range, divide the parameters into smaller intervals;

S3. Randomly select a set of parameter sets p for model training, and store the parameter set p and model accuracy. is the optimal value;

S4. Produce a random rotation direction vector i and the corresponding step size C(i). The parameter set p moves the step size C(i) in the rotation direction to generate a new parameter set. ;

S5, calculation model accuracy/> , compare/> and/> , storing the parameter set with higher accuracy as the latest optimal parameter set;

S6. Repeat steps S3-S5 until the accuracy no longer changes, and the current parameter set is the optimal parameter of the model.

The beneficial effects of the present invention are: in the present invention, different information uploaded by local users is classified and transmitted to the outer blockchain and the inner blockchain through the sharing function module, and the requesting end user initiates sharing on the outer blockchain When requested, based on the sharing request, the inner blockchain is linked to start an automated sharing task to ensure data sharing security and transaction traceability. Alarm functions are set up in the data interface module, sharing function module and resource feedback module, realizing a three-layer The attack alarm architecture monitors the security of the entire data sharing process. At the same time, it allows users to interact with the alarm system to improve the system's fault tolerance, thereby achieving comprehensive maintenance of the entire data sharing process.

In addition, in the present invention, through the setting of a resource attack pool, the intelligent attack detection model can be optimized and trained, thereby achieving a significant improvement in model accuracy, efficiency and other performance.

Description of the drawings

The drawings are used to provide a further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.

In the attached picture:

Figure 1 is an overall flow chart of the present invention;

Figure 2 is a schematic diagram of the shared function module in the present invention;

Figure 3 is a schematic diagram of the optimization process of the intelligent attack detection model in the present invention;

Figure 4 is a flow chart of the multi-layer alarm system architecture in the present invention.

Detailed ways

In order to make the purpose and advantages of the present invention more clear, the present invention will be described in detail below with reference to examples.

It should be understood that the following text is only used to describe an inline blockchain-based medical data security sharing platform or several specific implementations of the present invention, and does not strictly limit the scope of protection specifically requested by the present invention.

Embodiment, a medical data security sharing platform based on inline blockchain, as shown in Figure 1, the platform includes a data interface module, a shared function module, a resource feedback module and a resource supply pool;

The data interface module is used to connect to local user equipment and is configured with a device vulnerability detection function. It performs vulnerability detection on user equipment by starting an intelligent attack detection model. Among them, user equipment vulnerability detection features include a large number of TCP SYN connection requests in network traffic. and UDP packets;

Behavior of frequently accessing a specific URL and using unknown protocols in the device log; software information in the device;

The security strength of the weak password of the device; whether it is not patched;

The tested equipment enters the shared function module, as shown in Figure 2, which is a schematic diagram of the shared function module;

The shared function module accesses the detected device and executes:

First, perform risk detection on the data to be shared;

Among them, the detection characteristics of data to be uploaded by users are:

User login information, access time, IP address and data operation records in local device logs; whether it deviates from the average, standard deviation or is in the extreme percentile; whether the data shows an obvious peak or long-tail distribution; certain data points Whether it has extremely high or low values, rare or very unusual; whether the change of a certain data point is abnormal compared with surrounding data points;

Then, local users upload data summary information to the outer blockchain. In addition to data summary information, the outer blockchain is also used to record shared transaction information and shared result information, supporting access by all users on the chain;

Upload data information from local users to the inner blockchain. The inner blockchain only supports automated operations and does not support access;

And when the requesting user initiates a sharing request on the outer blockchain, based on the sharing request, the inner blockchain is linked to start automated sharing tasks, such as transmission, mining, storage, etc.;

The resource feedback module is used to realize data sharing transaction records and transaction traceability. It also requires the requesting end user to upload local data and data operation results, and activates the intelligent attack detection model to implement attack detection on the uploaded results;

The resource supply pool is used to store classified attack detection results and optimize and train intelligent attack detection models based on calibrated attack characteristics.

As shown in Figure 3, it is a schematic diagram of the optimization process of the intelligent attack detection model. The optimization training of the intelligent attack detection model is specifically as follows:

S1. Obtain the known attack characteristic data set T from the resource supply pool and copy it to form a copy set. ;

S2. According to the designed attack characteristic changing method, modify the copy set Perform operations until all features change to form a mutation set/> ;

S3. Collect the mutations Merge with known attack feature data set T to form a new attack feature set/> ;

S4. Use optimization methods to select parameters for the intelligent attack detection model to obtain the optimal parameter model;

S5. Utilize attack signature set To train the optimal parameter model;

Among them, the attack characteristics change method is: using copy set , Chebyshev distance formula, calculate the distance threshold; copy the set/> , the distance threshold is substituted into the Chebyshev distance formula to obtain multiple sets of unknown attack features, and then the feature variant direction and variant range are calculated and the multiple sets of unknown attack features are screened based on the calculation results, and finally a mutation set of unknown attacks that meets the requirements is obtained/> ;

For example, assume that the known attack signature is , the characteristic after mutation is , the Chebyshev distance formula is:

Among them, a and b are the features in set A and set B,/> is the corresponding vector coordinate; then, calculate the minimum distance threshold in set A/> and maximum distance threshold/> ,/> Determine the characteristics and expand the scope,/> Determine the direction of feature expansion; randomly select features in set A/> Substitute into the formula to calculate the corresponding expanded features/> ;Loop the calculation process until all known features have been enlarged.

Among them, the optimization method is:

S1. Optimize the objective function based on the model detection accuracy. When the detection rate is the highest, the corresponding parameter set is the optimal parameter set;

S2. Given the parameter variation range, divide the parameters into smaller intervals;

S3. Randomly select a set of parameter sets p for model training, and store the parameter set p and model accuracy. is the optimal value;

S4. Produce a random rotation direction vector i and the corresponding step size C(i). The parameter set p moves the step size C(i) in the rotation direction to generate a new parameter set. ;

S5, calculation model accuracy/> , compare/> and/> , storing the parameter set with higher accuracy as the latest optimal parameter set;

S6. Repeat steps S3-S5 until the accuracy no longer changes, and the current parameter set is the optimal parameter of the model.

In addition, in another embodiment of the security sharing platform of the present invention, as shown in Figure 4, which is a multi-layer early warning system architecture flow chart, a three-layer attack alarm architecture is constructed within the platform, specifically as follows:

The first layer alarm is set at the data interface module. The user equipment is connected to the system, and vulnerability detection is performed on the user equipment. If a vulnerability is found, an alarm is initiated, and the network equipment vulnerability attack detection result is defined as N_a, and the first layer alarm threshold is defined. is N, if , then an alarm is initiated;

The second-level alarm is set at the shared function module. It detects the data to be uploaded. If a risk is found, an alarm is initiated. The detection result of the data to be uploaded is defined as U_d, and the second-level alarm threshold is defined as U. If , an alarm is initiated.

The third layer alarm is set at the resource feedback module. It detects the results to be uploaded. If a risk is found, an alarm is initiated. The detection result of the data to be uploaded is defined as U_r. The second layer alarm threshold is defined as R. If , then an alarm is initiated;

Among them, each level of alarm requires the user to manually confirm whether to cancel the alarm and continue the task, realizing security monitoring of the entire data sharing process; at the same time, users are allowed to interact with the early warning system to improve the system's fault tolerance.

The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. For those of ordinary skill in the art, after learning the contents described in the present invention, they can proceed without departing from the principles of the present invention. Under the premise, several equivalent transformations and substitutions can also be made, and these equivalent transformations and substitutions should also be regarded as belonging to the protection scope of the present invention.

Claims (7)

1. A secure sharing platform for medical data based on inline blockchain, characterized in that the platform includes: The data interface module is used to connect to local user equipment and is configured with device vulnerability detection function; The shared function module accesses the detected device and executes: Conduct risk detection on data to be shared; Upload data summary information from local users to the outer blockchain; Upload data from local users to the inner blockchain; and When the requesting user initiates a sharing request on the outer blockchain, the inner blockchain is linked to start an automated sharing task based on the sharing request; Resource feedback module, used to realize data sharing transaction records and transaction traceability; The resource feedback module also requires the requesting end user to upload local data and data operation results, and activates an intelligent attack detection model to implement attack detection on the uploaded results; The platform also includes a resource supply pool, which is used to store classified attack detection results and to optimize and train intelligent attack detection models; The optimization training of the intelligent attack detection model is specifically as follows: S1. Obtain the known attack characteristic data set T from the resource supply pool and copy it to form a copy set. ; S2. According to the designed attack characteristics change method, the set Perform operations until all features change to form a mutation set/> ; S3. Collect the mutations Merge with the known attack feature set T to form a new attack feature set/> ; S4. Use optimization methods to select parameters for the intelligent attack detection model to obtain the optimal parameter model; S5. Utilize attack signature set to train the optimal parameter model. 2. A medical data security sharing platform based on inline blockchain according to claim 1, characterized in that: when performing the device vulnerability detection function, vulnerability detection is performed on the user equipment by starting an intelligent attack detection model. If a vulnerability is found If the vulnerability is detected, an alarm is initiated, and the network device vulnerability attack detection result is defined as N_a, and the first layer alarm threshold is defined as N. If , an alarm is initiated. 3. A medical data security sharing platform based on inline blockchain according to claim 1, characterized in that: the risk detection is: initiating an intelligent attack detection model to detect the data to be uploaded, and if a risk is found, initiate alarm, and define the detection result of the data to be uploaded as U_d, and define the second-level alarm threshold as U. If , an alarm is initiated. 4. A medical data security sharing platform based on inline blockchain according to claim 1, characterized in that: in the sharing function module, the outer blockchain is also used to record shared transaction information and sharing The result information is accessible to all users on the chain; the inner blockchain only supports automated operations. 5. A medical data security sharing platform based on inline blockchain according to claim 1, characterized in that: when detecting attacks on uploaded results, an alarm will be initiated if a risk is found; each level of alarm requires manual operation by the user. Confirm whether to cancel the alarm and continue the task, and define the detection result of the data to be uploaded as U_r, and define the second-level alarm threshold as R. If , an alarm is initiated. 6. A medical data security sharing platform based on inline blockchain according to claim 1, characterized in that: the attack feature changing method is: known attack feature copy set , Chebyshev distance formula, calculate the distance threshold; copy the attack feature set/> , the distance threshold is substituted into the Chebyshev distance formula to obtain multiple sets of unknown attack features, and then the feature variant direction and variant range are calculated and the multiple sets of unknown attack features are screened based on the calculation results, and finally a set of unknown attack variant features that meet the requirements are obtained/> . 7. A medical data security sharing platform based on inline blockchain according to claim 1, characterized in that: wherein, an optimization method is used for parameter selection, and the optimal parameter model is obtained as follows: S1. Optimize the objective function based on the model detection accuracy. When the accuracy is the highest, the corresponding parameter set is the optimal parameter set; S2. Given the parameter variation range, divide the parameters into smaller intervals; S3. Randomly select a set of parameters p for model training, and store the parameters p and model accuracy. is the optimal value; S4. Produce a random rotation direction vector i and the corresponding step size C(i). The parameter set p moves the step size C(i) in the rotation direction to generate a new parameter set. ; S5, calculation model accuracy/> , compare/> and/> , storing the parameter set with higher accuracy as the latest optimal parameter set; S6. Repeat steps S3-S5 until the accuracy no longer changes, and the current parameter set is the optimal parameter of the model.