Medical data reasoning prediction method based on homomorphic encryption and related equipment
By optimizing polynomial acceleration algorithms and homomorphic encryption processing on the convolutional neural network model, the problem of high computing and memory burden in telemedicine is solved, efficient medical data inference on edge devices is realized, and real-time telemedicine services are improved Sex and privacy protection.
Patent Information
- Application Number
- CN202411032723.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-30
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-07-30
AI Technical Summary
In telemedicine applications, homomorphic encrypted convolutional neural networks have too much computing and memory burden on the processing of medical data, making it difficult to achieve real-time diagnosis and treatment in scenarios with limited computing resources.
The convolutional neural network model is optimized through a polynomial acceleration algorithm, and the medical data is encrypted using homomorphic encryption technology, and the data structure is processed and optimized in parallel on the image processor to reduce computing complexity and memory requirements.
It improves the inference speed of telemedicine data, improves the real-time nature of telemedicine services, and ensures the privacy and security of patient data.
Smart Images

Figure CN119008029A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of smart medical technology, and in particular to a medical data reasoning and prediction method based on homomorphic encryption and related equipment. Background Art
[0002] With the advancement of IoT medical technology and cryptography, telemedicine services have become more popular, which brings new challenges to patient privacy and data security. Convolutional neural networks on the edge have become the basis of advanced machine learning due to their excellent performance and scalability, but there are privacy and reasoning efficiency issues when processing convolutional neural network reasoning directly in plaintext. Therefore, homomorphic encryption provides an alternative solution that can perform reasoning on encrypted data without exposing any original information. Convolutional neural networks using homomorphic encryption can perform calculations directly on encrypted data, bringing advantages such as single-round communication, bandwidth reduction, and non-interactive computing.
[0003] In telemedicine applications, the reliance of these homomorphic encryption-based models on specific medical datasets further hinders the development of standardized and universally applicable network architectures. Different medical institutions may collect and use different types of patient data, which poses a challenge to the universality of the model. In addition, the computational differences between different model layers in homomorphic encrypted convolutional neural networks also pose huge challenges, especially when performing homomorphic computations on encrypted medical data. The inherent characteristics of homomorphic encryption further exacerbate these challenges because it brings huge computational and memory burdens, which results in a large amount of computational resource allocation even for basic medical diagnosis or prediction operations, thus limiting the practical application of this method in telemedicine scenarios with limited edge devices or computing resources and tight time requirements. For example, smart devices deployed in patients' homes may not be able to bear the complex homomorphic computing overhead, thus affecting the real-time nature of remote diagnosis and treatment. Summary of the invention
[0004] The present invention provides a medical data reasoning and prediction method based on homomorphic encryption and related equipment, the purpose of which is to improve the real-time performance of telemedicine services by increasing the reasoning speed.
[0005] In order to achieve the above object, the present invention provides a medical data reasoning and prediction method based on homomorphic encryption, comprising:
[0006] Obtain the original telemedicine data of the target patient, and encrypt the original telemedicine data through homomorphic encryption to obtain the encrypted telemedicine data;
[0007] Input the encrypted telemedicine data into the inference prediction model for inference prediction to obtain the medical prediction result of the target patient; the medical prediction result includes one or more physiological indicator data;
[0008] The inference prediction model is obtained by training the convolutional neural network model optimized by the polynomial acceleration algorithm.
[0009] Specifically, the original telemedicine data of the target patient is obtained, and the original telemedicine data is encrypted through homomorphic encryption to obtain the encrypted telemedicine data, including:
[0010] Acquire the original telemedicine data of the target patient through the collection device;
[0011] Preprocessing the original telemedicine data to obtain preprocessed telemedicine data;
[0012] The preprocessed telemedicine data is encrypted through homomorphic encryption to obtain encrypted telemedicine data.
[0013] Furthermore, the encrypted telemedicine data is input into the inference prediction model for inference prediction, including:
[0014] Select to initialize the convolutional neural network model. The initialization convolutional neural network model includes the input layer, convolution layer, pooling layer and fully connected layer connected in sequence;
[0015] The initialized convolutional neural network model is optimized by a polynomial acceleration algorithm to obtain an optimized convolutional neural network model;
[0016] Train the optimized convolutional neural network model to obtain an inference prediction model;
[0017] The encrypted telemedicine data is input into the inference prediction model for inference prediction.
[0018] Furthermore, the initialization of the convolutional neural network model is optimized through the polynomial acceleration algorithm, including:
[0019] Optimize the input layer of the initialized convolutional neural network model so that the input layer processes the input encrypted telemedicine data in parallel;
[0020] By assigning each homomorphic encryption operation type its own implementation strategy, the data processing process for initializing the convolutional neural network model is optimized;
[0021] The model parameters are stored in the registers of the image processor and the shared memory of the image processor is used for data exchange and transmission to optimize the data structure for initializing the convolutional neural network model.
[0022] Furthermore, the input layer of the initialized convolutional neural network model is optimized, including:
[0023] Thread blocks are allocated to multiple encrypted telemedicine data in the input layer of the initialized convolutional neural network model, and the thread blocks are processed in parallel.
[0024] Further, the implementation strategies include spatial operations and non-spatial operations;
[0025] For spatial operations, it is used to extract local feature maps and edge feature maps in the spatial domain of encrypted telemedicine data;
[0026] For non-spatial operations, it is used to perform nonlinear transformation on local feature maps and edge feature maps to extract one-dimensional feature vectors.
[0027] Specifically, local feature maps and edge feature maps are extracted in the spatial domain of the encrypted telemedicine data, including:
[0028] The telemedicine data after parallel processing is input into the convolution layer. The convolution layer extracts features by sliding the two-dimensional convolution kernel on the spatial domain of the encrypted telemedicine data to obtain a local feature map.
[0029] The local feature map is input into the pooling layer, and the pooling layer downsamples the local feature map in the spatial domain of the encrypted telemedicine data to obtain an edge feature map.
[0030] The present invention also provides a medical data reasoning and prediction device based on homomorphic encryption, comprising:
[0031] An acquisition module is used to acquire the original telemedicine data of the target patient and encrypt the original telemedicine data through homomorphic encryption to obtain the encrypted telemedicine data;
[0032] An inference prediction module is used to input the encrypted telemedicine data into an inference prediction model for inference prediction to obtain a medical prediction result of a target patient; the medical prediction result includes one or more physiological index data;
[0033] The inference prediction model is obtained by training the convolutional neural network model optimized by the polynomial acceleration algorithm.
[0034] The present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, a medical data reasoning and prediction method based on homomorphic encryption is implemented.
[0035] The present invention also provides a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, a medical data reasoning and prediction method based on homomorphic encryption is implemented.
[0036] The above scheme of the present invention has the following beneficial effects:
[0037] The present invention encrypts the original telemedicine data of the target patient acquired through homomorphic encryption to obtain encrypted telemedicine data; the encrypted telemedicine data is input into an inference prediction model for inference prediction to obtain a medical prediction result of the target patient; the inference prediction model is obtained by training a convolutional neural network model optimized by a polynomial acceleration algorithm; compared with the prior art, the present invention uses homomorphic encryption technology to ensure that the original telemedicine data of the target patient is always in an encrypted state, effectively protecting personal privacy, and at the same time uses a polynomial acceleration algorithm to optimize the convolutional neural network model, thereby improving the model's inference speed for telemedicine data, thereby improving the real-time performance of telemedicine services.
[0038] Other beneficial effects of the present invention will be described in detail in the subsequent specific implementation section. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] Figure 1 A schematic diagram of a flow chart of an embodiment of the present invention;
[0040] Figure 2 A schematic diagram of the structure of a medical data inference and prediction device provided by an embodiment of the present invention;
[0041] Figure 3 A schematic diagram of the structure of a terminal device provided in an embodiment of the present invention. DETAILED DESCRIPTION
[0042] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, the following will be described in detail with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0043] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship 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, the terms "first", "second", and "third" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.
[0044] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installation", "connection" and "connection" should be understood in a broad sense, for example, it can be a locking connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0045] In addition, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.
[0046] In view of the existing problems, the present invention provides a medical data reasoning and prediction method based on homomorphic encryption and related equipment.
[0047] like Figure 1 As shown, an embodiment of the present invention provides a medical data reasoning and prediction method based on homomorphic encryption, comprising:
[0048] Obtain the original telemedicine data of the target patient, and encrypt the original telemedicine data through homomorphic encryption to obtain the encrypted telemedicine data;
[0049] Input the encrypted telemedicine data into the inference prediction model for inference prediction to obtain the medical prediction result of the target patient; the medical prediction result includes one or more physiological index data, such as heart rate, respiratory rate, electromyography, etc.;
[0050] The inference prediction model is obtained by training the convolutional neural network model optimized by the polynomial acceleration algorithm.
[0051] Specifically, the original telemedicine data of the target patient is obtained, and the original telemedicine data is encrypted through homomorphic encryption to obtain the encrypted telemedicine data, including:
[0052] Acquire the original telemedicine data of the target patient through the collection device;
[0053] Preprocessing the original telemedicine data to obtain preprocessed telemedicine data;
[0054] The preprocessed telemedicine data is encrypted through homomorphic encryption to obtain encrypted telemedicine data.
[0055] In the embodiment of the present invention, the acquisition device may be a device with a camera function such as a mobile phone or a PDA of the target patient. The target patient may shoot or upload the original medical data through the mobile phone or PDA to obtain the original telemedicine data. The original telemedicine data includes but is not limited to the original medical examination sheets such as electrocardiogram and electroencephalogram.
[0056] The original telemedicine data is preprocessed through a series of methods such as data cleaning, denoising, and standardization to obtain preprocessed telemedicine data;
[0057] The pre-processed telemedicine data is encrypted using homomorphic encryption technology in the collection device to obtain encrypted telemedicine data.
[0058] In an embodiment of the present invention, a monitoring service provider uses a homomorphic encryption algorithm to generate a key for a collection device, and the collection device encrypts the pre-processed telemedicine data using the key to obtain encrypted telemedicine data.
[0059] Specifically, the encrypted telemedicine data is input into the inference prediction model for inference prediction, including:
[0060] Selecting an initialized convolutional neural network model, the initialized convolutional neural network model includes an input layer for receiving data, a convolution layer for extracting local features of the data, a pooling layer for extracting edge features of the data, and a fully connected layer for fusing features, which are connected in sequence;
[0061] The initialized convolutional neural network model is optimized by a polynomial acceleration algorithm to obtain an optimized convolutional neural network model;
[0062] Train the optimized convolutional neural network model to obtain an inference prediction model;
[0063] The encrypted telemedicine data is input into the inference prediction model for inference prediction.
[0064] In the embodiment of the present invention, the initialized convolutional neural network model is a conventional convolutional neural network model. The present invention does not involve improving the structure of the model, so its structure will not be described in detail. The conventional training method is also adopted for training the optimized convolutional neural network model, so its training process will not be described in detail.
[0065] Specifically, the initialization convolutional neural network model is optimized through the polynomial acceleration algorithm, including:
[0066] Optimize the input layer of the initialized convolutional neural network model so that the input layer processes the input encrypted telemedicine data in parallel;
[0067] By assigning each homomorphic encryption operation type its own implementation strategy, the data processing process for initializing the convolutional neural network model can be optimized.
[0068] The model parameters are stored in the registers of the image processor and the shared memory of the image processor is used for data exchange and transmission to optimize the data structure of the initialized convolutional neural network model.
[0069] Specifically, the intermediate data of the convolution operation will first be stored in the shared memory of the graphics processing unit (GPU). The model parameters stored in the GPU shared memory will be read into the registers, and operations such as number theoretic transformation (NTT) will be performed in the registers. After completion, the data results will be transferred to the GPU shared memory for subsequent operations. At the same time, the GPU shared memory will read other model parameters to be transformed into the GPU shared memory again to complete the data "exchange", thereby entering the next NNT and other data operation cycles.
[0070] Specifically, the input layer of the initialized convolutional neural network model is optimized, including:
[0071] Thread blocks are allocated to multiple encrypted telemedicine data in the input layer of the initialized convolutional neural network model, and the thread blocks are processed in parallel.
[0072] In the embodiment of the present invention, each thread block contains 128 threads for realizing highly parallel processing of fine-grained data. Meanwhile, batch processing of thread blocks is adopted to further improve the parallelism.
[0073] In an embodiment of the present invention, homomorphic encryption operation types include element addition (EleAdd), element multiplication (EleMult), convolution (Conv) and number theory transformation operations.
[0074] Specifically, implementation strategies include spatial operations and non-spatial operations;
[0075] For spatial operations, it is used to extract local feature maps and edge feature maps in the spatial domain of encrypted telemedicine data;
[0076] For non-spatial operations, it is used to perform nonlinear transformation on local feature maps and edge feature maps to extract one-dimensional feature vectors.
[0077] Specifically, local feature maps and edge feature maps are extracted in the spatial domain of the encrypted telemedicine data, including:
[0078] The telemedicine data after parallel processing is input into the convolution layer. The convolution layer extracts features by sliding the two-dimensional convolution kernel on the spatial domain of the encrypted telemedicine data to obtain a local feature map.
[0079] The local feature map is input into the pooling layer, and the pooling layer downsamples the local feature map in the spatial domain of the encrypted telemedicine data to obtain an edge feature map.
[0080] In the embodiment of the present invention, spatial operation refers to the operation involving two-dimensional spatial processing of images or signals in the deep learning model. This includes convolution operation, which uses a two-dimensional convolution kernel to slide on the spatial domain of the input data to extract local features; it also includes pooling operation, which downsamples the feature map in the spatial domain to extract more abstract edge features. These operations retain the spatial structure of the input data, enabling the model to learn and understand the spatial hierarchical relationship in the input; for spatial operations, firstly, a parallel computing method is used in the input layer to package and batch process the input encrypted telemedicine data to optimize the calculation of weight multiplication and bias addition; next, the weights of the convolution layer are encoded as plaintext, and convolution operations are performed in the encrypted domain. By multiplying and aggregating the corresponding plaintext and encrypted telemedicine data, the convolution result of each channel can be obtained; the embodiment of the present invention optimizes the pooling layer by interleaved batch processing, which is different from the traditional method of sequentially storing each receptive field. The process of this method is as follows:
[0081] First, extract all receptive fields of size F×F in the encrypted telemedicine data and map them to F 2 Message;
[0082] Next, all receptive fields are permuted, which changes the position of each receptive field but does not affect the convolution operation itself, because the calculation is performed in a batch-processing single instruction multiple data stream (SIMD, Single Instruction Multiple Data) manner;
[0083] Then, four homomorphic rotations and additions are performed on the permuted receptive field to calculate the average of the 16 values required by the pooling layer.
[0084] This staggered batch processing method allows the pooling layer to be executed directly after the batch convolution operation without intermediate storage and data transmission.
[0085] In the embodiment of the present invention, for non-spatial operations, that is, in the fully connected layer, each neuron is connected to all neurons in the previous layer, regardless of their spatial position relationship and their spatial structure, and an activation function is used to independently perform a nonlinear transformation on each input value in order to optimize the activation function. The steps are as follows:
[0086] Choose the activation function: Use the ReLU activation function because it has advantages such as high computational efficiency and easy implementation;
[0087] Approximate optimization: Since the use of the ReLU activation function in a homomorphic encryption environment will bring about complexity problems, the embodiment of the present invention proposes to use a simpler quadratic function as the ReLU activation function, that is, ReLU(x)≈x2 ;
[0088] Computational efficiency optimization: Quadratic function approximation can be achieved through a single homomorphic multiplication, which greatly improves computational efficiency, which is very critical for homomorphic encryption scenarios;
[0089] Trade-offs: The embodiment of the present invention balances approximation accuracy and computational efficiency, and ultimately selects a quadratic function with higher computational efficiency as the ReLU activation function. Although it lacks the sparse characteristics of the ordinary ReLU activation function, in a homomorphic encryption environment, the calculation mode is not affected by the plaintext value, so the sparse characteristic advantage of the ordinary ReLU activation function will not bring significant performance improvement.
[0090] Specifically, by assigning each homomorphic encryption operation type its own implementation strategy, the data processing process for initializing the convolutional neural network model is optimized. The optimization process is as follows:
[0091] The polynomial acceleration algorithm decomposes complex homomorphic encryption operations into basic residual digital system operations through hierarchical reconstruction, simplifying the overall calculation process; the specific process of hierarchical reconstruction includes:
[0092] Thread allocation: The embodiment of the present invention allocates a dedicated thread block to each residual, each thread block is configured with 128 threads, and these thread blocks are batch loaded into each kernel to achieve concurrent processing of residuals;
[0093] Element-by-element kernel operations: For element-by-element operations such as element addition (EleAdd), element multiplication (EleMult), and convolution (Conv), each thread is responsible for processing a residual coefficient, performing weighting, comparison, and tuning operations to obtain different runs (convergence cycle times) corresponding to different residual coefficient levels;
[0094] Implementation of number theory transformation: Use a layered approach to implement number theory operations, follow the latest methodology, and implement 14 Optimization was performed involving 14 processing levels;
[0095] For complex number-theoretic transformations, they are divided into two cores for separate processing. The first core processes the first 8 levels of number-theoretic transformations, and the second core processes the last 6 levels of number-theoretic transformations. In each core, each thread manages 8 coefficients, greatly reducing the computational complexity.
[0096] Specifically, the model parameters are stored in the registers of the image processor and the shared memory of the image processor is used for data exchange and transmission to optimize the data structure of the initialized convolutional neural network model. The optimization process is as follows:
[0097] The eight coefficients managed by each thread are stored in the register of the image processor, which improves the memory access efficiency. Then the transformation is performed using the shared memory of the image processor. After the transformation is completed, the next round of transformation can be carried out, which reduces the access to the global memory.
[0098] In summary, the embodiment of the present invention encrypts the original telemedicine data of the target patient obtained by homomorphic encryption to obtain encrypted telemedicine data; the encrypted telemedicine data is input into the inference prediction model for inference prediction to obtain the medical prediction result of the target patient; the medical prediction result includes one or more physiological indicator data; the inference prediction model is obtained by training the convolutional neural network model optimized by the polynomial acceleration algorithm; compared with the prior art, the present invention uses homomorphic encryption technology to ensure that the original telemedicine data of the target patient is always in an encrypted state, effectively protecting personal privacy, and at the same time uses the polynomial acceleration algorithm to optimize the convolutional neural network model, simplifying the calculation process of complex operations, and improving the model's inference speed for telemedicine data, thereby improving the real-time performance of telemedicine services.
[0099] like Figure 2 As shown, the embodiment of the present invention further provides a medical data reasoning prediction device 100 based on homomorphic encryption, and the medical data reasoning prediction device 100 includes:
[0100] The acquisition module 101 is used to acquire the original telemedicine data of the target patient and encrypt the original telemedicine data through homomorphic encryption to obtain the encrypted telemedicine data;
[0101] The inference prediction module 102 is used to input the encrypted remote medical data into the inference prediction model for inference prediction to obtain the medical prediction result of the target patient; the medical prediction result includes one or more physiological index data;
[0102] The inference prediction model is obtained by training the convolutional neural network model optimized by the polynomial acceleration algorithm.
[0103] In some embodiments, the acquisition module 101 is specifically used to acquire the original telemedical data of the target patient through an acquisition device; preprocess the original telemedical data to obtain preprocessed telemedical data; and encrypt the preprocessed telemedical data through homomorphic encryption to obtain encrypted telemedical data.
[0104] The acquisition module 101 includes:
[0105] The acquisition submodule is used to acquire the original telemedicine data of the target patient through the acquisition device;
[0106] A preprocessing submodule is used to preprocess the original telemedicine data to obtain preprocessed telemedicine data;
[0107] The encryption submodule is used to encrypt the pre-processed telemedicine data through homomorphic encryption to obtain encrypted telemedicine data.
[0108] In some embodiments, the above-mentioned reasoning and prediction module 102 is specifically used to select an initialized convolutional neural network model, which includes an input layer, a convolution layer, a pooling layer, and a fully connected layer; optimize the initialized convolutional neural network model through a polynomial acceleration algorithm to obtain an optimized convolutional neural network model; train the optimized convolutional neural network model to obtain an reasoning and prediction model; input the encrypted telemedicine data into the reasoning and prediction model for reasoning and prediction to obtain the disease risk index of the target patient.
[0109] The above-mentioned inference prediction submodule 102 includes:
[0110] A selection submodule is used to select and initialize a convolutional neural network model. The initialization convolutional neural network model includes an input layer, a convolutional layer, a pooling layer, and a fully connected layer connected in sequence;
[0111] An optimization submodule is used to optimize the initialized convolutional neural network model through a polynomial acceleration algorithm to obtain an optimized convolutional neural network model;
[0112] The training submodule is used to train the optimized convolutional neural network model to obtain an inference prediction model;
[0113] The inference and prediction submodule is used to input the encrypted telemedicine data into the inference and prediction model for inference and prediction to obtain the medical prediction results of the target patient; the medical prediction results include one or more physiological indicator data, such as heart rate, respiratory rate, electromyography, etc.
[0114] In some embodiments, the optimization submodule is specifically used to optimize the input layer of the initialized convolutional neural network model so that the input layer processes the input encrypted telemedicine data in parallel; by assigning respective implementation strategies to each homomorphic encryption operation type, the data processing process of the initialized convolutional neural network model is optimized; the model parameters are stored through the registers of the image processor and the shared memory of the image processor is used for data exchange and transmission, so as to optimize the data structure of the initialized convolutional neural network model.
[0115] Among them, the above optimization submodules include:
[0116] A first optimization unit is used to optimize the input layer of the initialized convolutional neural network model so that the input layer processes the input encrypted telemedicine data in parallel;
[0117] A second optimization unit is used to optimize the data processing process of initializing the convolutional neural network model by assigning a respective implementation strategy to each homomorphic encryption operation type;
[0118] The third optimization unit stores the model parameters through the registers of the image processor and uses the shared memory of the image processor to exchange and transmit data, so as to optimize the data structure of the initialized convolutional neural network model.
[0119] In some embodiments, the first optimization unit is specifically used to allocate thread blocks to multiple encrypted telemedicine data in the input layer of the initialized convolutional neural network model, and perform parallel processing on the thread blocks.
[0120] In some embodiments, the implementation strategy includes spatial operations and non-spatial operations;
[0121] For spatial operations, it is used to extract local feature maps and edge feature maps in the spatial domain of encrypted telemedicine data;
[0122] For non-spatial operations, it is used to perform nonlinear transformation on local feature maps and edge feature maps to extract one-dimensional feature vectors.
[0123] In some embodiments, extracting a local feature map and an edge feature map in the spatial domain of the encrypted telemedicine data includes:
[0124] The telemedicine data after parallel processing is input into the convolution layer. The convolution layer extracts features by sliding the two-dimensional convolution kernel on the spatial domain of the encrypted telemedicine data to obtain a local feature map.
[0125] The local feature map is input into the pooling layer, and the pooling layer downsamples the local feature map in the spatial domain of the encrypted telemedicine data to obtain an edge feature map.
[0126] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the protection scope of the embodiments of the present application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.
[0127] An embodiment of the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the medical data reasoning and prediction method based on homomorphic encryption described in any of the above embodiments.
[0128] If the integrated module is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the process in the above method embodiment, which can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may at least include: any entity or device that can carry the computer program code to the construction device / terminal device, recording medium, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. For example, U disk, mobile hard disk, disk or optical disk, etc. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electric carrier signals and telecommunication signals.
[0129] like Figure 3 As shown, the embodiment of the present invention further provides a terminal device D10 including: at least one processor D100 ( Figure 3Only one processor is shown in the figure), a memory D101, and a computer program D102 stored in the memory D101 and executable on the at least one processor D100, wherein the processor D100 implements the steps in any of the above-mentioned method embodiments when executing the computer program D102. Alternatively, the processor D100 implements the functions of each module / unit in the above-mentioned device embodiments when executing the computer program D102.
[0130] The processor may be a central processing unit (CPU), or other general-purpose processors, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.
[0131] In some embodiments, the memory may be an internal storage unit of the terminal device, such as a hard disk or memory of the terminal device. In other embodiments, the memory may also be an external storage device of the terminal device, such as a plug-in hard disk, a smart memory card (SMC, Smart Media Card), a secure digital (SD, Secure Digital) card, a flash card (Flash Card), etc. equipped on the terminal device. Further, the memory may also include both an internal storage unit of the terminal device and an external storage device. The memory is used to store an operating system, an application program, a boot loader (Boot Loader), data and other programs, such as the program code of the computer program, etc. The memory may also be used to temporarily store data that has been output or is to be output.
[0132] It should be noted that the information interaction, execution process, etc. between the above-mentioned devices / units are based on the same concept as the method embodiment of the embodiment of the present application. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.
[0133] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the protection scope of the embodiments of the present application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.
[0134] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. A medical data reasoning and prediction method based on homomorphic encryption, characterized in that: include: Obtaining original telemedicine data of a target patient, and encrypting the original telemedicine data by homomorphic encryption to obtain encrypted telemedicine data; Inputting the encrypted telemedicine data into an inference prediction model for inference prediction to obtain a medical prediction result of a target patient; the medical prediction result includes one or more physiological indicator data; The inference prediction model is obtained by training a convolutional neural network model optimized by a polynomial acceleration algorithm.
2. The medical data reasoning and prediction method based on homomorphic encryption according to claim 1 is characterized in that: The original telemedicine data of the target patient is obtained, and the original telemedicine data is encrypted by homomorphic encryption to obtain the encrypted telemedicine data, including: Acquire the original telemedicine data of the target patient through the collection device; Preprocessing the original telemedicine data to obtain preprocessed telemedicine data; The preprocessed telemedicine data is encrypted through homomorphic encryption to obtain encrypted telemedicine data.
3. The medical data reasoning and prediction method based on homomorphic encryption according to claim 2 is characterized in that: Inputting the encrypted telemedicine data into the inference prediction model for inference prediction, including: Selecting an initialized convolutional neural network model, wherein the initialized convolutional neural network model includes sequentially connecting an input layer, a convolutional layer, a pooling layer, and a fully connected layer; Optimizing the initialized convolutional neural network model by a polynomial acceleration algorithm to obtain an optimized convolutional neural network model; Training the optimized convolutional neural network model to obtain an inference prediction model; The encrypted telemedicine data is input into the inference prediction model for inference prediction.
4. The medical data reasoning and prediction method based on homomorphic encryption according to claim 3 is characterized in that: The initialized convolutional neural network model is optimized by a polynomial acceleration algorithm, including: Optimizing the input layer of the initialized convolutional neural network model so that the input layer processes the input encrypted telemedicine data in parallel; By assigning a respective implementation strategy to each homomorphic encryption operation type, the data processing process of the initialized convolutional neural network model is optimized; The model parameters are stored through registers of the image processor and the shared memory of the image processor is used for data exchange and transmission to optimize the data structure of the initialized convolutional neural network model.
5. The medical data reasoning and prediction method based on homomorphic encryption according to claim 4 is characterized in that: Optimizing the input layer of the initialized convolutional neural network model, including: In the input layer of the initialized convolutional neural network model, thread blocks are allocated to multiple encrypted telemedicine data, and the thread blocks are processed in parallel.
6. The medical data reasoning and prediction method based on homomorphic encryption according to claim 4 is characterized in that: The implementation strategy includes spatial operations and non-spatial operations; The spatial operation is used to extract a local feature map and an edge feature map in the spatial domain of the encrypted telemedicine data; The non-spatial operation is used to perform nonlinear transformation on the local feature map and the edge feature map to extract a one-dimensional feature vector.
7. The medical data reasoning and prediction method based on homomorphic encryption according to claim 6 is characterized in that: Extract local feature maps and edge feature maps in the spatial domain of the encrypted telemedicine data, including: The telemedicine data after parallel processing is input into the convolution layer, and the convolution layer extracts features by sliding a two-dimensional convolution kernel on the spatial domain of the encrypted telemedicine data to obtain a local feature map; The local feature map is input into the pooling layer, and the pooling layer downsamples the local feature map in the spatial domain of the encrypted telemedicine data to obtain an edge feature map.
8. A medical data reasoning and prediction device based on homomorphic encryption, characterized in that: include: An acquisition module is used to acquire the original telemedicine data of the target patient and encrypt the original telemedicine data through homomorphic encryption to obtain the encrypted telemedicine data; An inference prediction module, used for inputting the encrypted telemedicine data into an inference prediction model for inference prediction to obtain a medical prediction result of a target patient; the medical prediction result includes one or more physiological index data; The inference prediction model is obtained by training a convolutional neural network model optimized by a polynomial acceleration algorithm.
9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the medical data reasoning and prediction method based on homomorphic encryption as described in any one of claims 1 to 7 is implemented.
10. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the medical data reasoning and prediction method based on homomorphic encryption as described in any one of claims 1 to 7 is implemented.
Citation Information
Patent Citations
Context-aware medical image segmentation method based on lesion boundary information response
CN112561937A
Electric power data prediction method and device and electronic equipment
CN116383666A
Encryption reasoning method and system based on homomorphic encryption technology
CN116865953A
FPGA-based homomorphic encryption CNN acceleration method
CN118296661A
System and methods for validating and performing operations on homomorphically encrypted data
US20170293913A1
