A secure multi-party medical data collaborative analysis method and system

Abstract

The application discloses a kind of safe multi-party medical data collaborative analysis method and system, it is related to medical data analysis technical field.The method comprises: for each client, the sample data of the client is preprocessed to obtain preprocessed feature, and the preprocessed feature is learned to sparse feature;Information encryption interaction is carried out to several client and preset server end, and encrypted fusion data is obtained, wherein each client corresponds to different medical institutions;According to the encrypted fusion data, sparse integrated features are obtained, and the sparse integrated features are weighted according to noise;According to the weighted sparse integrated features, data analysis is carried out, and sample classification results are obtained.The application can realize the collaborative processing of medical data of different medical institutions under privacy protection by means of sparse feature representation, data encryption fusion and noise weighting, and effectively improve the accuracy of data analysis results.

Description

Technical Field

[0001] This invention relates to the field of medical data analysis technology, and in particular to a secure multi-party collaborative analysis method and system for medical data. Background Technology

[0002] With the development of intelligent sensing, signal processing, and artificial intelligence technologies, it is now possible to collect and process digitized medical data and perform data analysis using data-driven AI algorithms. However, medical data often contains personal privacy information such as identity information and medical records. When data samples from a single medical institution are insufficient for data analysis tasks, collaboration among multiple medical institutions is necessary. Direct fusion and analysis of sample data by different institutions can easily lead to the leakage of personal information; therefore, privacy protection is a crucial requirement in medical data analysis. Furthermore, data sample collection is susceptible to noise and other issues due to human factors or equipment malfunctions, which can affect the accuracy of data analysis results.

[0003] Therefore, existing technologies still need improvement and development. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a secure multi-party medical data collaborative analysis method and system to address the above-mentioned defects of the prior art. The aim is to solve the problems that in the prior art, when multiple medical institutions collaboratively process data samples, personal information is easily leaked, and the noise in the collected data samples affects the accuracy of the data analysis results.

[0005] The technical solution adopted by this invention to solve the problem is as follows:

[0006] In a first aspect, embodiments of the present invention provide a secure multi-party collaborative analysis method for medical data, wherein the method includes:

[0007] For each client, the sample data of that client is preprocessed to obtain preprocessed features, and sparse feature learning is performed on the preprocessed features.

[0008] The information is encrypted and interacted with several clients and a preset server to obtain encrypted fused data, wherein each client corresponds to a different medical institution;

[0009] Sparse ensemble features are obtained from the encrypted fusion data, and noise weighting is applied to the sparse ensemble features.

[0010] Data analysis is performed based on the weighted sparse ensemble features to obtain sample classification results.

[0011] In one implementation, the data preprocessing step includes:

[0012] Feature extraction is performed on the sample data;

[0013] The extracted data features are normalized to obtain the preprocessed features.

[0014] In one implementation, the sparse feature learning step includes:

[0015] The preprocessed features are input into the sparse feature learning module, wherein the sparse feature learning module is based on a randomly configured sparse autoencoder.

[0016] The sparse feature learning module calculates sparse features and outputs weights.

[0017] In one implementation, the step of encrypting the information exchange includes:

[0018] The number of clients is two. Three matrices are randomly generated by the server. One matrix is ​​sent to the first client, and the other two matrices are sent to the second client.

[0019] For the first client, the preprocessed features are encrypted according to the received matrix, and the encrypted preprocessed features are sent to the second client;

[0020] For the second client, the output weight is encrypted based on a received matrix. First calculation data is obtained based on the encrypted preprocessed features from the first client, the output weight, and another received matrix. The encrypted output weight and the first calculation data are then sent to the first client.

[0021] For the first client, second calculation data is obtained based on the first calculation data, the received matrix, and the encrypted output weight of the second client, and the second calculation data is sent to the server.

[0022] The encrypted and fused data is obtained by encrypting and fusing the data on the server side.

[0023] In one implementation, the step of obtaining sparse integration features based on the encrypted fusion data includes:

[0024] The encrypted fused data is input into the sparse feature integration module, wherein the sparse feature integration module is composed of several parallel randomly configured sparse autoencoders.

[0025] The sparse feature integration module outputs the sparse integrated features.

[0026] In one implementation, the noise weighting step includes:

[0027] The sparse integrated features are input into a noise weighting module, wherein the noise weighting module is constructed based on a scoring function, and the scoring function is determined by a membership function and a non-membership function;

[0028] The sample weights are obtained through the scoring function.

[0029] In one implementation, the step of performing data analysis based on the weighted sparse ensemble features to obtain sample classification results includes:

[0030] The sparse integrated features are input into an intuitionistic fuzzy weighted random configuration network, wherein the intuitionistic fuzzy weighted random configuration network includes an L2 regularization term;

[0031] Data analysis is performed using the intuitionistic fuzzy weighted random configuration network, and the sample classification results are output.

[0032] Secondly, embodiments of the present invention also provide a secure multi-party medical data collaborative analysis system, the system comprising:

[0033] The data preprocessing module performs data preprocessing on the sample data of each client to obtain preprocessed features;

[0034] A sparse feature learning module is used to perform sparse feature learning on the preprocessed features;

[0035] The information encryption and fusion module is used to perform encrypted information interaction between several clients and a preset server to obtain encrypted fused data, wherein each client corresponds to a different medical institution.

[0036] A sparse feature integration module is used to obtain sparse integrated features based on the encrypted fusion data;

[0037] A noise weighting module is used to perform noise weighting on the sparse integration features;

[0038] The data analysis module is used to perform data analysis based on the weighted sparse ensemble features to obtain sample classification results.

[0039] Thirdly, embodiments of the present invention also provide a terminal, the terminal including a memory and one or more processors; the memory stores one or more programs; the programs include instructions for executing the secure multi-party medical data collaborative analysis method as described above; the processor is used to execute the programs.

[0040] Fourthly, embodiments of the present invention also provide a computer-readable storage medium storing a plurality of instructions, which are adapted to be loaded and executed by a processor to implement the steps of any of the above-described secure multi-party medical data collaborative analysis methods.

[0041] The beneficial effects of this invention are as follows: For each client, the sample data of that client is preprocessed to obtain preprocessed features, and sparse feature learning is performed on the preprocessed features. Encrypted information interaction is performed between several clients and a preset server to obtain encrypted fused data, wherein each client corresponds to a different medical institution. Sparse ensemble features are obtained from the encrypted fused data, and noise weighting is applied to the sparse ensemble features. Data analysis is performed based on the weighted sparse ensemble features to obtain sample classification results. This invention, through sparse feature representation, encrypted data fusion, and noise weighting, enables collaborative processing of medical data from different medical institutions while protecting privacy, and effectively improves the accuracy of data analysis results. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 This is a flowchart illustrating the secure multi-party collaborative analysis method for medical data provided in this embodiment of the invention.

[0044] Figure 2 This is a diagram of the data preprocessing module provided in an embodiment of the present invention.

[0045] Figure 3 This is a diagram of the sparse feature learning module provided in an embodiment of the present invention.

[0046] Figure 4 This is a diagram of the information encryption fusion module provided in an embodiment of the present invention.

[0047] Figure 5 This is a diagram of the sparse feature integration module provided in the embodiments of the present invention.

[0048] Figure 6 This is a diagram of the noise weighting module provided in an embodiment of the present invention.

[0049] Figure 7 This is a diagram of the data analysis module provided in an embodiment of the present invention.

[0050] Figure 8This is a schematic diagram of the modules of the secure multi-party medical data collaborative analysis system provided in an embodiment of the present invention.

[0051] Figure 9 This is a schematic diagram of the terminal provided in the embodiment of the present invention. Detailed Implementation

[0052] This invention discloses a secure multi-party collaborative analysis method and system for medical data. To make the objectives, technical solutions, and effects of this invention clearer and more explicit, the invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only for explaining the invention and are not intended to limit the invention.

[0053] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the term “comprising” as used in this specification means the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when we say an element is “connected” or “coupled” to another element, it can be directly connected or coupled to the other element, or there may be intermediate elements. Furthermore, “connected” or “coupled” as used herein can include wireless connections or wireless coupling. The term “and / or” as used herein includes all or any units and all combinations of one or more associated listed items.

[0054] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0055] To address the aforementioned deficiencies in existing technologies, this invention provides a secure multi-party collaborative analysis method for medical data. The method includes: for each client, performing data preprocessing on the client's sample data to obtain preprocessed features, and then performing sparse feature learning on the preprocessed features; performing encrypted information interaction between several clients and a preset server to obtain encrypted fused data, wherein each client corresponds to a different medical institution; obtaining sparse ensemble features based on the encrypted fused data, and then applying noise weighting to the sparse ensemble features; and performing data analysis based on the weighted sparse ensemble features to obtain sample classification results. This invention, through sparse feature representation, encrypted data fusion, and noise weighting, enables collaborative processing of medical data from different medical institutions while protecting privacy, and effectively improves the accuracy of data analysis results.

[0056] like Figure 1 As shown, the method includes:

[0057] Step S100: For each client, perform data preprocessing on the sample data of that client to obtain preprocessed features, and perform sparse feature learning on the preprocessed features.

[0058] Specifically, the sample data in this embodiment can be an individual's in vitro diagnostic information, including but not limited to identity information and medical data. Since this embodiment involves collaborative data processing from multiple clients, data preprocessing is required for each client before sparse feature learning is performed.

[0059] In one implementation, the data preprocessing steps include:

[0060] Feature extraction is performed on the sample data;

[0061] The extracted data features are normalized to obtain the preprocessed features.

[0062] Specifically, such as Figure 2 As shown, this embodiment pre-constructs a data preprocessing module. After the client's sample data is input into the data preprocessing module, feature extraction is first performed, followed by normalization processing to obtain the preprocessed features of the sample data. Taking the collaborative processing of data from two clients, medical institution A and medical institution B, as an example, data preprocessing is first performed on the sample data of medical institution A and medical institution B respectively to obtain normalized preprocessed features.

[0063] In one implementation, the steps of sparse feature learning include:

[0064] The preprocessed features are input into the sparse feature learning module, wherein the sparse feature learning module is based on a randomly configured sparse autoencoder.

[0065] The sparse feature learning module calculates sparse features and outputs weights.

[0066] Specifically, such as Figure 3 As shown, this embodiment pre-constructs a sparse feature learning module using a randomly configured sparse autoencoder. After obtaining the preprocessed features from each client, each client uses the randomly configured sparse autoencoder in the sparse feature learning module to learn features and obtain the corresponding output weights. The advantage of the randomly configured network is that it can quickly and effectively model the model. Furthermore, by introducing a supervision mechanism through incremental learning, assigning values ​​to random parameters using inequality constraints, adaptively selecting the range of values ​​for random parameters, and calculating the hidden layer output weights using the least squares method, the global approximation of the learning model can be guaranteed. With less human intervention, the learning accuracy and efficiency of the network can be effectively improved.

[0067] For example, the learning process is as follows: Medical institutions A and B are two clients. A and B respectively input their preprocessed features: X = {x1, x2, ..., x...} N}, i = 1, 2, ..., N, where N represents the number of samples;

[0068] Define the objective function as: f: Assuming L-1 hidden layer nodes have been generated, the current network output f L-1 (X) is:

[0069]

[0070] Where, β j The output weights of node j; g j w is the activation function. j and b j Let be the input weights and biases of node j, respectively; j = 1, 2, ..., L max ;

[0071] Calculate the network residual vector:

[0072] e L-1 =ff L-1 (X)=[e L-1，1 (X), e L-1，2 (X), ..., e L-1，d (X)], (2);

[0073] Determine if the termination condition is met. If not, add node L and configure the node parameters according to equation (3), selecting the node that meets the condition. The candidate node parameters with the maximum value are used as the final parameters of node L;

[0074]

[0075] h L =[g L (w L T x1+b L ), ..., g L (w L T x N +b L )] T (4);

[0076] Where q = 1, 2, ..., d, d represents the output dimension; h L The output representing node L is calculated using equation (4); w L and b L These represent the L parameters of the node to be determined; T max Represents the number of candidate parameters; r∈(0,1); {μ L} represents a sequence of non-negative real numbers, where μ L ≤1-r,lim L→+∞ μ L =0.

[0077] Based on the objective function of equation (5), the output weight β of the randomly configured sparse autoencoder is determined and solved by the alternating direction multiplier method;

[0078]

[0079] Among them, β=[β1, β2,…,β L H = [h1, h2, ..., h] L ];

[0080] Based on equations (6) and (7) respectively, obtain the features extracted and the output f of the randomly configured sparse autoencoder:

[0081] X = Xβ, (6);

[0082] f = Xβ + (7);

[0083] Where, β + Represents the Moore-Penrose generalized inverse of β.

[0084] Step S200: Perform encrypted information interaction between several clients and a preset server to obtain encrypted fused data, wherein each client corresponds to a different medical institution.

[0085] Specifically, such as Figure 4As shown, this embodiment pre-defines a third-party server. This server interacts with each client via encrypted communication to obtain encrypted fused data. This encrypted fused data integrates data from each client, and because the original data is encrypted, it cannot be directly accessed. Therefore, collaborative data analysis between different medical institutions can be achieved while protecting privacy.

[0086] In one implementation, the steps of the encrypted information exchange include:

[0087] The number of clients is two. Three matrices are randomly generated by the server. One matrix is ​​sent to the first client, and the other two matrices are sent to the second client.

[0088] For the first client, the preprocessed features are encrypted according to the received matrix, and the encrypted preprocessed features are sent to the second client;

[0089] For the second client, the output weight is encrypted based on a received matrix. First calculation data is obtained based on the encrypted preprocessed features from the first client, the output weight, and another received matrix. The encrypted output weight and the first calculation data are then sent to the first client.

[0090] For the first client, second calculation data is obtained based on the first calculation data, the received matrix, and the encrypted output weight of the second client, and the second calculation data is sent to the server.

[0091] The encrypted and fused data is obtained by encrypting and fusing the data on the server side.

[0092] Specifically, this embodiment takes two clients, medical institution A (hereinafter referred to as client A) and medical institution B (hereinafter referred to as client B), as examples. After the sparse feature learning module, client A and medical institution B respectively obtain β. A and β B ;

[0093] Matrix R is randomly generated by a third-party server. A R R and R b and R A Transmitted to client A, R B and R b Transmitted to client B;

[0094] For client A, for X A Encrypt: And encrypt Transmitted to client B;

[0095] For client B, encrypt the output matrix. calculate (E1 is the first calculated data), and E1 is sent to client A;

[0096] For client A, calculate E2 (i.e., the second calculated data) and send it to the server. The formula for calculating E2 is:

[0097] X can be obtained from the server. A β B =E2-R b +R A R B Data encryption and fusion are performed to obtain

[0098] Step S300: Obtain sparse integration features based on the encrypted fusion data, and apply noise weighting to the sparse integration features.

[0099] Specifically, this embodiment first obtains sparse integration features for encrypted fused data, and then reduces the impact of noise samples on the data analysis results through noise weighting.

[0100] In one implementation, the step of obtaining sparse integration features based on the encrypted fusion data includes:

[0101] The encrypted fused data is input into the sparse feature integration module, wherein the sparse feature integration module is composed of several parallel randomly configured sparse autoencoders.

[0102] The sparse feature integration module outputs the sparse integrated features.

[0103] Specifically, such as Figure 5 As shown, this embodiment pre-constructs a sparse feature integration module using multiple randomly configured sparse autoencoders in parallel. In practical applications, the encrypted and fused data is processed using the sparse feature integration module to obtain sparse integrated features through multiple randomly configured sparse autoencoders in parallel.

[0104] In one implementation, the noise weighting step includes:

[0105] The sparse integrated features are input into a noise weighting module, wherein the noise weighting module is constructed based on a scoring function, and the scoring function is determined by a membership function and a non-membership function;

[0106] The sample weights are obtained through the scoring function.

[0107] Specifically, such as Figure 6As shown, the noise weighting module in this embodiment mainly consists of a scoring function, which is determined using membership and non-membership functions. In practical applications, to reduce the impact of outliers in the data samples on the data analysis results, sparse ensemble features need to be input into the noise weighting module for processing before data analysis. The noise weighting module uses intuitionistic fuzzy computation to obtain the membership and non-membership functions of different data samples, and defines a scoring function to obtain the sample weights in order to analyze outliers in the dataset.

[0108] For example, the membership function in the noise-weighted module is used to calculate the distance between each sample and its corresponding class centroid in the high-dimensional feature space. For each training sample x i The membership function is defined as:

[0109]

[0110] Among them, D + (D - ) and r + (r - ) are the center and radius of the positive and negative classes, respectively, and ξ is an adjustable parameter.

[0111] Non-membership is used to calculate the ratio between the number of heterogeneous points and the number of points in their neighborhood. The non-membership function is defined as:

[0112] α * (x i )=(1-α(x i ))Φ(x i ), (9);

[0113] in:

[0114] 0≤α(x i )+α * (x i )≤1, (10);

[0115]

[0116] Where η is an adjustable parameter, and |·| is the cardinality of the set.

[0117] The scoring function is determined by the membership function and the non-membership function. The scoring function is defined as follows:

[0118]

[0119] Step S400: Perform data analysis based on the weighted sparse ensemble features to obtain sample classification results.

[0120] Specifically, this embodiment uses noise-weighted sparse ensemble features for data analysis, which can yield more accurate sample classification results. The sample classification results can reflect which group category a sample belongs to; specifically, they can be the calculated probability or reference value of a sample belonging to different group categories, used to assist professionals in making further medical judgments.

[0121] In one implementation, the step of performing data analysis based on the weighted sparse ensemble features to obtain sample classification results includes:

[0122] The sparse integrated features are input into an intuitionistic fuzzy weighted random configuration network, wherein the intuitionistic fuzzy weighted random configuration network includes an L2 regularization term;

[0123] Data analysis is performed using the intuitionistic fuzzy weighted random configuration network, and the sample classification results are output.

[0124] Specifically, such as Figure 7 As shown, this embodiment pre-constructs an Intuitive Fuzzy Weighted Randomized Network (IFU) to perform data analysis on the sparse ensemble features of the input and output the corresponding sample classification results. The IFU can assign a score value to each sample to handle outliers in the dataset, reducing the impact of noisy data samples on the classification results. Furthermore, to prevent ill-conditioned solutions, this embodiment also adds an L2 regularization term to the IFU.

[0125] For example, for an intuitionistic fuzzy weighted random allocation network, the objective function is defined as follows:

[0126]

[0127] The optimization problem in equation (13) is a convex quadratic optimization problem, and the solution to this problem is shown below:

[0128]

[0129] Among them, S=diag(Θ1, Θ2,..., Θ N ), s i =Θ i .

[0130] The intuitionistic fuzzy weighted randomized network with added L2 regularization uses the following inequality constraints for node parameter configuration:

[0131]

[0132]

[0133] in,

[0134] Based on the above embodiments, the present invention also provides a secure multi-party medical data collaborative analysis system, such as... Figure 8 As shown, the system includes:

[0135] Data preprocessing module 01 performs data preprocessing on the sample data of each client to obtain preprocessed features;

[0136] Sparse feature learning module 02 is used to perform sparse feature learning on the preprocessed features;

[0137] The information encryption fusion module 03 is used to perform information encryption interaction between several clients and a preset server to obtain encrypted fusion data, wherein each client corresponds to a different medical institution.

[0138] Sparse feature integration module 04 is used to obtain sparse integrated features based on the encrypted fusion data;

[0139] Noise weighting module 05 is used to perform noise weighting on the sparse integration features;

[0140] Data analysis module 06 is used to perform data analysis based on the weighted sparse ensemble features to obtain sample classification results.

[0141] Furthermore, the data preprocessing module is specifically used for feature extraction and data normalization; the sparse feature learning module is composed of randomly configured sparse autoencoders, specifically used to calculate sparse features and output weights; the specific implementation method of the information encryption fusion module can be found in the corresponding method steps; the sparse feature integration module is composed of multiple parallel randomly configured sparse autoencoders; the noise weighting module is composed of a scoring function determined based on membership and non-membership functions; the data analysis module is composed of an intuitionistic fuzzy weighted random configuration network, and the output sample classification results can reflect which group category the sample belongs to, specifically the calculated probability or reference value of the sample belonging to different group categories, used to assist professionals in making further medical judgments.

[0142] For example, based on some clinical data collected from a hospital, and using pulse signals, a preset group category C was selected for verification. 200 samples from group C and 200 samples from non-group C were selected, and 300 samples were randomly selected as training data and 100 samples as test data.

[0143] Medical institutions A and B each used 150 data samples for training. Feature extraction was performed using short-time Fourier transform through a data preprocessing module, and the features were then normalized. Subsequently, the project was implemented step-by-step, following modules for sparse feature learning, information encryption and fusion (including a server-side module), sparse feature integration, noise weighting, and data analysis. This resulted in a 94.30% accuracy rate in sample classification, demonstrating that the invention can achieve collaborative data analysis between different medical institutions while protecting privacy.

[0144] Based on the above embodiments, the present invention also provides a terminal, the principle block diagram of which can be as follows: Figure 9 As shown, the terminal includes a processor, memory, network interface, and display screen connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements a secure multi-party collaborative analysis method for medical data. The display screen can be an LCD screen or an e-ink screen.

[0145] Those skilled in the art will understand that Figure 9 The schematic diagram shown is merely a partial structural diagram related to the present invention and does not constitute a limitation on the terminal to which the present invention is applied. A specific terminal may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0146] In one implementation, the terminal's memory stores one or more programs, and these programs are configured to be executed by one or more processors, and the programs contain instructions for performing a secure multi-party collaborative analysis method for medical data.

[0147] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided by this invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0148] In summary, this invention discloses a secure multi-party collaborative analysis method and system for medical data. The method includes: for each client, performing data preprocessing on the client's sample data to obtain preprocessed features, and then performing sparse feature learning on the preprocessed features; performing encrypted information interaction between several clients and a preset server to obtain encrypted fused data, wherein each client corresponds to a different medical institution; obtaining sparse ensemble features based on the encrypted fused data, and then applying noise weighting to the sparse ensemble features; and performing data analysis based on the weighted sparse ensemble features to obtain sample classification results. This invention, through sparse feature representation, encrypted data fusion, and noise weighting, can achieve collaborative processing of medical data from different medical institutions while protecting privacy, and effectively improves the accuracy of data analysis results.

[0149] It should be understood that the application of the present invention is not limited to the examples above. Those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A secure multi-party collaborative analysis method for medical data, characterized in that, The method includes: For each client, the sample data of that client is preprocessed to obtain preprocessed features, and sparse feature learning is performed on the preprocessed features; the steps of sparse feature learning include: inputting the preprocessed features into a sparse feature learning module, wherein the sparse feature learning module is based on a randomly configured sparse autoencoder; calculating sparse features and outputting weights through the sparse feature learning module; Information encryption interaction is performed between several clients and a preset server to obtain encrypted fused data, wherein each client corresponds to a different medical institution; the steps of the information encryption interaction include: two clients are involved; the server randomly generates three matrices, sending one matrix to a first client and the other two matrices to a second client; for the first client, the preprocessed features are encrypted based on the received matrices, and the encrypted preprocessed features are sent to the second client; for the second client, the output weights are encrypted based on one received matrix; first calculated data is obtained based on the encrypted preprocessed features from the first client, the output weights, and the other received matrix, and the encrypted output weights and the first calculated data are sent to the first client; for the first client, second calculated data is obtained based on the first calculated data, the received matrix, and the encrypted output weights from the second client, and the second calculated data is sent to the server; the server performs data encryption and fusion to obtain the encrypted fused data. Sparse ensemble features are obtained from the encrypted fusion data, and noise weighting is applied to the sparse ensemble features. Data analysis is performed based on the weighted sparse ensemble features to obtain sample classification results.

2. The secure multi-party medical data collaborative analysis method according to claim 1, characterized in that, The data preprocessing steps include: Feature extraction is performed on the sample data; The extracted data features are normalized to obtain the preprocessed features.

3. The secure multi-party medical data collaborative analysis method according to claim 1, characterized in that, The step of obtaining sparse integration features based on the encrypted fusion data includes: The encrypted fused data is input into the sparse feature integration module, wherein the sparse feature integration module is composed of several parallel randomly configured sparse autoencoders. The sparse feature integration module outputs the sparse integrated features.

4. The secure multi-party medical data collaborative analysis method according to claim 1, characterized in that, The noise weighting step includes: The sparse integrated features are input into a noise weighting module, wherein the noise weighting module is constructed based on a scoring function, and the scoring function is determined by a membership function and a non-membership function; The sample weights are obtained through the scoring function.

5. The secure multi-party medical data collaborative analysis method according to claim 1, characterized in that, The step of performing data analysis based on the weighted sparse ensemble features to obtain sample classification results includes: The sparse ensemble features are input into an intuitionistic fuzzy weighted random configuration network, wherein the intuitionistic fuzzy weighted random configuration network includes... L 2. Regularization terms; Data analysis is performed using the intuitionistic fuzzy weighted random configuration network, and the sample classification results are output.

6. A secure multi-party medical data collaborative analysis system, characterized in that, The system includes: The data preprocessing module is used to preprocess the sample data of each client to obtain preprocessed features. A sparse feature learning module is used to perform sparse feature learning on the preprocessed features; the steps of the sparse feature learning include: inputting the preprocessed features into the sparse feature learning module, wherein the sparse feature learning module is constructed based on a randomly configured sparse autoencoder; calculating sparse features and outputting weights through the sparse feature learning module; An information encryption and fusion module is used to perform encrypted information interaction between several clients and a preset server to obtain encrypted fused data. Each client corresponds to a different medical institution. The steps of the encrypted information interaction include: For each of the several clients (two in number), the server randomly generates three matrices, sending one matrix to a first client and the other two matrices to a second client; for the first client, encrypting the preprocessed features based on the received matrices and sending the encrypted preprocessed features to the second client; for the second client, encrypting the output weights based on one received matrix, obtaining first calculated data based on the encrypted preprocessed features from the first client, the output weights, and the other received matrix, and sending the encrypted output weights and the first calculated data to the first client; for the first client, obtaining second calculated data based on the first calculated data, the received matrix, and the encrypted output weights from the second client, and sending the second calculated data to the server; and finally, the server performs data encryption and fusion to obtain the encrypted fused data. A sparse feature integration module is used to obtain sparse integrated features based on the encrypted fusion data; A noise weighting module is used to perform noise weighting on the sparse integration features; The data analysis module is used to perform data analysis based on the weighted sparse ensemble features to obtain sample classification results.

7. A terminal, characterized in that, The terminal includes a memory and one or more processors; the memory stores one or more programs; the programs contain instructions for executing the secure multi-party medical data collaborative analysis method as described in any one of claims 1-5; the processors are used to execute the programs.

8. A computer-readable storage medium storing a plurality of instructions thereon, characterized in that, The instructions are applicable to be loaded and executed by a processor to implement the steps of the secure multi-party medical data collaborative analysis method according to any one of claims 1-5.

Images