WO2021189907A1 - 电子设备、疾病类型检测方法、装置及介质 - Google Patents

电子设备、疾病类型检测方法、装置及介质 Download PDF

Info

Publication number
WO2021189907A1
WO2021189907A1 PCT/CN2020/131987 CN2020131987W WO2021189907A1 WO 2021189907 A1 WO2021189907 A1 WO 2021189907A1 CN 2020131987 W CN2020131987 W CN 2020131987W WO 2021189907 A1 WO2021189907 A1 WO 2021189907A1
Authority
WO
WIPO (PCT)
Prior art keywords
gradient parameters
model gradient
model
disease type
type detection
Prior art date
Application number
PCT/CN2020/131987
Other languages
English (en)
French (fr)
Inventor
李泽远
王健宗
Original Assignee
平安科技(深圳)有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 平安科技(深圳)有限公司 filed Critical 平安科技(深圳)有限公司
Publication of WO2021189907A1 publication Critical patent/WO2021189907A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G16INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16HHEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/50ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for simulation or modelling of medical disorders
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/21Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F21/00Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
    • G06F21/60Protecting data
    • G06F21/602Providing cryptographic facilities or services
    • GPHYSICS
    • G16INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16HHEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H10/00ICT specially adapted for the handling or processing of patient-related medical or healthcare data
    • G16H10/60ICT specially adapted for the handling or processing of patient-related medical or healthcare data for patient-specific data, e.g. for electronic patient records

Definitions

  • This application relates to data processing technology, and in particular to an electronic device, disease type detection method, device, and computer-readable storage medium.
  • Heart disease has become a common disease in people's lives.
  • the specific subdivision of heart disease includes many types.
  • the realization of rapid and accurate detection of heart disease types can reduce the mortality of heart disease patients.
  • the detection of heart diseases mainly uses electrocardiogram detection methods to diagnose heart diseases.
  • the ECG data used to train the model is limited to a few hospitals, and the ECG data owned by each hospital still has data barriers. , So that a large amount of electrocardiogram data cannot be effectively used, so that it is impossible to accurately determine the type of heart disease based on the electrocardiogram.
  • the electronic device includes:
  • At least one processor At least one processor
  • a memory connected in communication with the at least one processor
  • the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor, so that the at least one processor can execute the following steps:
  • Access the monitoring port of the server and after successfully establishing a connection with the server, encrypt the model gradient parameters after training and upload them to the server;
  • the application also provides an electronic device, wherein the electronic device includes:
  • At least one processor At least one processor
  • a memory connected in communication with the at least one processor
  • the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor, so that the at least one processor executes the following steps:
  • the present application also provides a disease type detection device applied to a client, the device includes:
  • the encryption module is used to obtain the disease type detection model from the server, train the disease type detection model through the local image data training data, obtain the trained model gradient parameters, access the monitoring port of the server, and successfully establish a connection , Encrypt the model gradient parameters and upload them to the server;
  • the model update module is used to receive the updated model gradient parameters transmitted by the server, and obtain the standard disease type detection model according to the updated model gradient parameters;
  • the classification module is used to receive the image data to be detected, and sequentially pass the image data to be detected through the convolutional layer, the normalization layer, the linear rectification layer, the random inactivation layer, the fully connected layer and the standard disease type detection model. Logistic regression layer to get the detection result of disease type.
  • the present application also provides a disease type detection device applied to the server, the device including:
  • the decryption module is used to open K listening ports, where K is the number of clients, using the listening port to receive encrypted model gradient parameters sent by multiple clients, and decrypt the encrypted model gradient parameters, Obtain the model gradient parameters corresponding to each client;
  • the parameter update module is configured to perform a joint operation on the model gradient parameters corresponding to each client to obtain the updated model gradient parameters, and distribute the updated model gradient parameters to each client.
  • the present application also provides a computer-readable storage medium, including a storage data area and a storage program area, the storage data area stores created data, and the storage program area stores a computer program; wherein the computer program is executed when the processor is executed:
  • initial gradient parameters from the server, construct a disease type detection model based on the initial gradient parameters, train the disease type detection model through local image training data, obtain the trained model, and extract the trained model gradient parameters;
  • Access the monitoring port of the server after successfully establishing a connection, encrypt the model gradient parameters and upload them to the server;
  • the present application also provides a computer-readable storage medium, including a storage data area and a storage program area, the storage data area stores created data, and the storage program area stores a computer program; wherein the computer program is executed when the processor is executed:
  • This application also provides a disease type detection method, which includes:
  • Access the monitoring port of the server and after successfully establishing a connection with the server, encrypt the model gradient parameters after training and upload them to the server;
  • This application also provides a disease type detection method, which includes:
  • FIG. 1 is a schematic diagram of the internal structure of an electronic device provided by the first embodiment of the application
  • FIG. 2 is a schematic flowchart of a disease type detection method provided by a second embodiment of this application;
  • FIG. 3 is a schematic flowchart of a disease type detection method provided by a third embodiment of this application.
  • FIG. 4 is a schematic diagram of modules of a disease type detection device provided by a fourth embodiment of this application.
  • the embodiment of the present application provides an electronic device.
  • the electronic device may be, for example, at least one of a server and a terminal.
  • the server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc.
  • FIG. 1 it is a schematic diagram of the internal structure of an electronic device provided by an embodiment of this application.
  • the electronic device 1 may include a processor 10, a memory 11, and a bus, and may also include a computer program stored in the memory 11 and running on the processor 10, such as a disease type detection program 12.
  • the memory 11 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, mobile hard disk, multimedia card, card-type memory (such as SD or DX memory, etc.), magnetic memory, magnetic disk, CD etc.
  • the memory 11 may be an internal storage unit of the electronic device 1 in some embodiments, for example, a mobile hard disk of the electronic device 1.
  • the memory 11 may also be an external storage device of the electronic device 1, such as a plug-in mobile hard disk, a smart media card (SMC), and a secure digital (Secure Digital) equipped on the electronic device 1. , SD) card, flash card (Flash Card), etc.
  • the memory 11 may also include both an internal storage unit of the electronic device 1 and an external storage device.
  • the memory 11 can be used not only to store application software and various data installed in the electronic device 1, such as the code of the disease type detection program 12, etc., but also to temporarily store data that has been output or will be output.
  • the processor 10 may be composed of integrated circuits in some embodiments, for example, may be composed of a single packaged integrated circuit, or may be composed of multiple integrated circuits with the same function or different functions, including one or more Combinations of central processing unit (CPU), microprocessor, digital processing chip, graphics processor, and various control chips, etc.
  • the processor 10 is the control unit of the electronic device, which uses various interfaces and lines to connect the various components of the entire electronic device, and runs or executes programs or modules stored in the memory 11 (such as executing Disease type detection programs, etc.), and call data stored in the memory 11 to execute various functions of the electronic device 1 and process data.
  • the bus may be a peripheral component interconnect standard (PCI) bus or an extended industry standard architecture (EISA) bus, etc.
  • PCI peripheral component interconnect standard
  • EISA extended industry standard architecture
  • the bus can be divided into address bus, data bus, control bus and so on.
  • the bus is configured to implement connection and communication between the memory 11 and at least one processor 10 and the like.
  • FIG. 1 only shows an electronic device with components. Those skilled in the art can understand that the structure shown in FIG. 1 does not constitute a limitation on the electronic device 1, and may include fewer or more components than shown in the figure. Components, or a combination of certain components, or different component arrangements.
  • the electronic device 1 may also include a power source (such as a battery) for supplying power to various components.
  • the power source may be logically connected to the at least one processor 10 through a power management device, thereby controlling power
  • the device implements functions such as charge management, discharge management, and power consumption management.
  • the power supply may also include any components such as one or more DC or AC power supplies, recharging devices, power failure detection circuits, power converters or inverters, and power status indicators.
  • the electronic device 1 may also include various sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be repeated here.
  • the electronic device 1 may also include a network interface.
  • the network interface may include a wired interface and/or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is usually used in the electronic device 1 Establish a communication connection with other electronic devices.
  • the electronic device 1 may also include a user interface.
  • the user interface may be a display (Display) and an input unit (such as a keyboard (Keyboard)).
  • the user interface may also be a standard wired interface or a wireless interface.
  • the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode, organic light-emitting diode) touch device, etc.
  • the display can also be appropriately called a display screen or a display unit, which is used to display the information processed in the electronic device 1 and to display a visualized user interface.
  • the disease type detection program 12 stored in the memory 11 in the electronic device 1 is a combination of multiple instructions.
  • a disease type detection method can be implemented.
  • the disease type detection method can refer to the following description in the flowchart shown in FIG. 2.
  • FIG. 2 it is a schematic flowchart of a disease type detection method provided by the second embodiment of this application.
  • the disease type detection method described in the second embodiment of this embodiment is applied to the client, including:
  • the local image training data includes the patient's ECG image stored in the hospital database.
  • the local image training data is used to train the disease type detection model obtained from the server end to obtain the trained model gradient parameters.
  • the encrypting the model gradient parameters includes:
  • the public key is (n, g) and the private key is ( ⁇ , ⁇ );
  • Prime number refers to a natural number that has no other factors except 1 and itself among the natural numbers greater than 1, and the large prime number refers to the largest one or more of the natural numbers that satisfy the definition of a prime number.
  • the public key is transmitted to the server, and the model gradient parameter is encrypted by using the private key ( ⁇ , ⁇ ) to obtain the encrypted model gradient parameter.
  • training of the disease type detection model is performed on multiple clients, such as multiple hospitals, to obtain model gradient parameters corresponding to each client.
  • the multiple hospitals may include Hospital A, Hospital B, Hospital C, etc.
  • the model gradient parameter corresponding to Hospital A is w 1
  • the model gradient parameter corresponding to Hospital B is w 2
  • the model gradient parameter corresponding to Hospital C is w 3 and so on, and so on.
  • the embodiment of the application uses the private key to encrypt the model gradient parameters, which facilitates subsequent upload and update, and achieves the purpose of expanding data samples and protecting patient privacy.
  • the embodiment of the present application uses the http protocol to upload the encrypted model gradient parameters to the server through the listening port opened by the server.
  • the disease type detection model is updated by using the updated model gradient parameters transmitted by the server to obtain a standard disease type detection model.
  • the image data to be detected is received, and the image data to be detected is sequentially passed through the convolutional layer, the normalization layer, the linear rectification layer, the random inactivation layer, and the fully connected layer of the standard disease type detection model. And logistic regression layer to get the detection result of disease type.
  • the convolutional layer is used for feature extraction of the image data to be detected
  • the normalization layer is used to prevent gradient explosion and gradient disappearance
  • the linear rectification layer is used to improve the gradient descent and back propagation process
  • the random inactivation layer is used to realize the regularization of the standard disease type detection model and reduce its structural risk
  • the fully connected layer is used to assemble the local features extracted from the feature
  • the logistic regression The layer is used to make predictions.
  • the image data to be detected and the corresponding annotation data include the electrocardiogram image and its corresponding json file containing the annotation data of the cardiologist.
  • the electrocardiogram data of the patient is obtained, and the electrocardiogram data is input into the standard disease type detection model, through the convolutional layer, normalization layer, and linear rectification layer of the standard disease type detection model, Random inactivation layer, fully connected layer and logistic regression layer to get the diagnosis result.
  • the diagnosis result is a specific classification of heart diseases, including but not limited to atrial fibrillation and flutter, atrioventricular block, dual rhythm, ectopic atrial rhythm, handover rhythm, sinus rhythm, supraventricular Tachycardia.
  • the schematic flowchart shown in FIG. 3 describes the disease type detection method provided by the third embodiment of the present application.
  • the disease type detection method described in the third embodiment of this embodiment is applied to the server and includes:
  • the server opens K listening ports according to the number of clients, so as to perform data transmission with each client.
  • the public key (n, g) is used to decrypt the encrypted model gradient parameters to obtain the model gradient parameters corresponding to each client.
  • the decrypting the encrypted model gradient parameters includes:
  • m is the model gradient parameter after decryption
  • c is the model gradient parameter after encryption
  • mod is the modulus operator
  • n p ⁇ q, where p and q are satisfying pq and (p-1)( The greatest common multiple of q-1) is a large prime number of 1
  • is the Carmichael function
  • is the preset parameter.
  • each client includes but is not limited to A hospital, B hospital, and C hospital, and each client decrypts the encrypted model gradient parameter to obtain the model gradient parameter corresponding to each client.
  • the performing a joint operation on the model gradient parameters corresponding to each client to obtain the updated model gradient parameters includes:
  • f(w) is the updated model gradient parameter
  • f i (w) is the model gradient parameter
  • F k (w) represents the intermediate parameter
  • K is the number of clients
  • P k represents the stored in the kth client
  • the training data in the terminal n k is the number of training data.
  • a secure connection is made between the client and the server, and the client and the server that have successfully established a connection are obtained, and the updated model gradient parameters are distributed to each client .
  • FIG. 4 it is a schematic diagram of the modules of the disease type detection device provided by the fourth embodiment of the present application.
  • the disease type detection device described in this application can be divided into a first disease type detection device 100 and a second disease type detection device 200.
  • the first disease type detection device 100 can be installed in the client and the second disease type detection device 200 can be installed in the server.
  • the first disease type detection device 100 may include an encryption module 101, a model update module 102, and a classification module 103; and the second disease type detection device 200 may include a decryption module 201 and a parameter update module 202.
  • the module described in this application can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can complete fixed functions, and are stored in the memory of the electronic device.
  • the functions of the first disease type detection device 100 and the modules of the first disease type detection device 200 are as follows:
  • the encryption module 101 is configured to obtain initial gradient parameters from the server, construct a disease type detection model according to the initial gradient parameters, and train the disease type detection model through local image training data to obtain the disease type detection After training the model gradient parameters of the model, access the monitoring port of the server, and after successfully establishing a connection, encrypt the trained model gradient parameters and upload them to the server;
  • the decryption module 201 is configured to open K listening ports, where K is the number of clients, and the listening ports are used to receive encrypted model gradient parameters sent by multiple clients, and to compare the encrypted model gradient parameters Perform decryption to obtain the model gradient parameters corresponding to each client;
  • the parameter update module 202 is configured to perform a joint operation on the model gradient parameters corresponding to each client to obtain the updated model gradient parameters, and distribute the updated model gradient parameters to each client.
  • the model update module 102 is configured to receive updated model gradient parameters transmitted by the server, and obtain a standard disease type detection model according to the updated model gradient parameters;
  • the classification module 103 is configured to receive image data to be detected, and sequentially pass the image data to be detected through the convolutional layer, the normalization layer, the linear rectification layer, the random inactivation layer, and the standard disease type detection model. Connect the layer and the logistic regression layer to get the disease type detection result.
  • modules of the first disease type detection device 100 perform the following operations in the client:
  • Step 1 The encryption module 101 obtains initial gradient parameters from the server, constructs a disease type detection model based on the initial gradient parameters, and trains the disease type detection model through local image training data to obtain the disease type detection Model gradient parameters after model training; in the embodiment of the present application, the local image training data includes the patient's ECG image stored in the hospital database.
  • the local image training data is used to train the disease type detection model obtained from the server end to obtain the trained model gradient parameters.
  • Step 2 The encryption module 101 accesses the listening port of the server, and after the connection is successfully established,
  • the model gradient parameters are encrypted and uploaded to the server.
  • the encryption module 101 encrypting the model gradient parameters includes:
  • the public key is (n, g) and the private key is ( ⁇ , ⁇ );
  • Prime number refers to a natural number that has no other factors except 1 and itself among the natural numbers greater than 1, and the large prime number refers to the largest one or more of the natural numbers that satisfy the definition of a prime number.
  • the public key is transmitted to the server, and the model gradient parameter is encrypted by using the private key ( ⁇ , ⁇ ) to obtain the encrypted model gradient parameter.
  • training of the disease type detection model is performed on multiple clients, such as multiple hospitals, to obtain model gradient parameters corresponding to each client.
  • the multiple hospitals may include Hospital A, Hospital B, Hospital C, etc.
  • the model gradient parameter corresponding to Hospital A is w 1
  • the model gradient parameter corresponding to Hospital B is w 2
  • the model gradient parameter corresponding to Hospital C is w 3 and so on, and so on.
  • the embodiment of the application uses the private key to encrypt the model gradient parameters, which facilitates subsequent upload and update, and achieves the purpose of expanding data samples and protecting patient privacy.
  • the embodiment of the present application uses the http protocol to upload the encrypted model gradient parameters to the server through the listening port opened by the server.
  • Step 3 The model update module 102 receives the updated gradient parameters transmitted by the server, and obtains a standard disease type detection model according to the updated model gradient parameters.
  • the model update module 102 described in the embodiment of the present application updates the disease type detection model by using the updated model gradient parameters transmitted by the server to obtain a standard disease type detection model.
  • Step 4 The classification module 103 receives the image data to be detected, and sequentially passes the image data to be detected through the convolutional layer, the normalization layer, the linear rectification layer, the random inactivation layer, and the standard disease type detection model. Connect the layer and the logistic regression layer to get the disease type detection result.
  • the convolutional layer is used for feature extraction of the image data to be detected
  • the normalization layer is used to prevent gradient explosion and gradient disappearance
  • the linear rectification layer is used to improve the gradient descent and back propagation process
  • the random inactivation layer is used to realize the regularization of the standard disease type detection model and reduce its structural risk
  • the fully connected layer is used to assemble the local features extracted from the feature
  • the logistic regression The layer is used to make predictions.
  • the classification module 103 receives the image data to be detected, and sequentially passes the image data to be detected through the convolutional layer, the normalization layer, the linear rectification layer, and the random inactivation of the standard disease type detection model. Layer, fully connected layer and logistic regression layer to get the disease type detection results.
  • the image data to be detected and the corresponding annotation data include the electrocardiogram image and its corresponding json file containing the annotation data of the cardiologist.
  • the electrocardiogram data of the patient is obtained, and the electrocardiogram data is input into the standard disease type detection model, through the convolutional layer, normalization layer, and linear rectification layer of the standard disease type detection model, Random inactivation layer, fully connected layer and logistic regression layer to get the diagnosis result.
  • the diagnosis result is a specific classification of heart diseases, including but not limited to atrial fibrillation and flutter, atrioventricular block, dual rhythm, ectopic atrial rhythm, handover rhythm, sinus rhythm, supraventricular Tachycardia.
  • modules of the second disease type detection device 200 perform the following operations in the server:
  • Step 1 The decryption module 201 opens K listening ports, where K is the number of clients.
  • the server opens K listening ports according to the number of clients, so as to perform data transmission with each client.
  • Step 2 The decryption module 201 uses the listening port to receive encrypted model gradient parameters sent by multiple clients.
  • Step 3 The decryption module 201 decrypts the encrypted model gradient parameters to obtain the model gradient parameters corresponding to each client.
  • the decryption module 201 in the embodiment of the present application uses the public key (n, g) to decrypt the encrypted model gradient parameters to obtain the model gradient parameters corresponding to each client.
  • the decrypting the encrypted model gradient parameters includes:
  • m is the model gradient parameter after decryption
  • c is the model gradient parameter after encryption
  • mod is the modulus operator
  • n p ⁇ q, where p and q are satisfying pq and (p-1)( The greatest common multiple of q-1) is a large prime number of 1
  • is the Carmichael function
  • is the preset parameter.
  • each client includes but is not limited to A hospital, B hospital, and C hospital, and each client decrypts the encrypted model gradient parameter to obtain the model gradient parameter corresponding to each client.
  • Step 4 The parameter update module 202 performs a joint operation on the model gradient parameters corresponding to each client to obtain the updated model gradient parameters.
  • the parameter update module 202 performs a joint operation on the model gradient parameters corresponding to each client to obtain the updated model gradient parameters, including:
  • f(w) is the updated model gradient parameter
  • f i (w) is the model gradient parameter
  • F k (w) represents the intermediate parameter
  • K is the number of clients
  • P k represents the stored in the kth client
  • the training data in the terminal n k is the number of training data.
  • Step 5 The parameter update module 202 distributes the updated model gradient parameters to each client.
  • the parameter update module 202 securely connects the client and the server to obtain the client and the server that have successfully established a connection, and converts the updated model gradient parameter Distribute to each client.
  • the module/unit of the disease type detection device can be stored in a computer readable storage medium, and the computer readable storage medium It can be volatile or non-volatile.
  • the computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory) .
  • the computer usable storage medium may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required by at least one function, etc.; the storage data area may store a block chain node Use the created data, etc.
  • the storage program area of the computer-readable storage medium stores a computer program; wherein, when the computer program is executed by a processor, the following steps are implemented:
  • Access the monitoring port of the server and after successfully establishing a connection with the server, encrypt the model gradient parameters after training and upload them to the server;
  • the storage program area of the computer-readable storage medium stores a computer program; wherein, when the computer program is executed by a processor, the following steps are implemented:
  • modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the objectives of the solutions of the embodiments.
  • the functional modules in the various embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit.
  • the above-mentioned integrated unit may be implemented in the form of hardware, or may be implemented in the form of hardware plus software functional modules.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Public Health (AREA)
  • Medical Informatics (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Epidemiology (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Biomedical Technology (AREA)
  • Bioethics (AREA)
  • Evolutionary Biology (AREA)
  • Computer Hardware Design (AREA)
  • Computer Security & Cryptography (AREA)
  • Software Systems (AREA)
  • Evolutionary Computation (AREA)
  • Databases & Information Systems (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Pathology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Medical Treatment And Welfare Office Work (AREA)

Abstract

本申请涉及数据处理技术,揭露一种电子设备、装置、存储介质及疾病类型检测方法。所述方法包括:客户端将利用本地数据训练后的模型梯度参数加密后上传至服务端,服务端对所述加密后的模型梯度参数进行解密并进行联合运算,得到更新后的模型梯度数据,客户端接收到服务端返回更新后的模型梯度参数并根据所述更新后的模型梯度参数得到标准疾病类型检测模型,利用标准疾病类型检测模型对待检测图像数据进行检测,得到疾病类型检测结果。本申请还涉及区块链技术,所述更新后的模型梯度参数可以存储在区块链节点中。本申请可以提高疾病类型检测模型的准确性。

Description

电子设备、疾病类型检测方法、装置及介质
本申请要求于2020年10月20日提交中国专利局、申请号为CN202011127805.4、名称为“电子设备、疾病类型检测方法、装置及介质”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
技术领域
本申请涉及数据处理技术,尤其涉及一种电子设备、疾病类型检测方法、装置及计算机可读存储介质。
背景技术
疾病的检测与判断对于降低病人的死亡率尤其重要。例如,当今,心脏病成为人们生活中一种常见疾病,心脏病具体细分包括众多种类,实现对心脏病种类快速且准确的检测可以降低心脏病患者死亡率。目前检测心脏病种类主要采用心电图检测手段,对心脏疾病进行诊断。
由于医疗数据的隐私性,发明人意识到目前用于对心电图进行检测的模型准确率不够高,用于训练模型的心电图数据局限于几个医院中,各个医院所拥有的心电图数据仍存在数据壁垒,使得大量的心电图数据无法有效利用,从而无法准确根据心电图判断出心脏疾病种类。
发明内容
本申请提供的一种电子设备,所述电子设备包括:
至少一个处理器;以及
与所述至少一个处理器通信连接的存储器;
其中,所述存储器存储有可被所述至少一个处理器执行的计算机程序,所述计算机程序被所述至少一个处理器执行,以使所述至少一个处理器能够执行如下步骤:
从服务端获取初始梯度参数,根据所述初始梯度参数构建疾病类型检测模型,通过本地的图像训练数据对所述疾病类型检测模型进行训练,得到所述疾病类型检测模型训练后的模型梯度参数;
访问服务端的监听端口,与所述服务端成功建立连接后,将训练后的所述模型梯度参数加密运算后上传至服务端;
接收服务端传送的更新后的模型梯度参数,根据所述更新后的模型梯度参数得到标准疾病类型检测模型;
接收待检测图像数据,将所述待检测图像数据依次通过所述标准疾病类型检测模型的卷积层,归一化层,线性整流层,随机失活层,全连接层及逻辑回归层,得到疾病类型检测结果。
本申请还提供一种电子设备,其中,所述电子设备包括:
至少一个处理器;以及
与所述至少一个处理器通信连接的存储器;
其中,所述存储器存储有可被所述至少一个处理器执行的计算机程序,所述计算机程序被所述至少一个处理器执行,以使所述至少一个处理器执行如下步骤:
开启K个监听端口,其中,K为客户端的数量;
利用所述监听端口接收多个客户端发送的加密后的模型梯度参数;
对所述加密后的模型梯度参数进行解密,得到每个客户端对应的模型梯度参数;
对所述每个客户端对应的模型梯度参数执行联合运算,得到更新后的模型梯度参数;
将所述更新后的模型梯度参数分发给每个客户端。
本申请还提供一种应用于客户端的疾病类型检测装置,所述装置包括:
加密模块,用于从服务端端获取疾病类型检测模型,通过本地的图像数据训练数据对所述疾病类型检测模型进行训练,得到训练后的模型梯度参数,访问服务端的监听端口,成功建立连接后,对所述模型梯度参数进行加密后上传至服务端;
模型更新模块,用于接收服务端传送的更新后的模型梯度参数,根据所述更新后的模型梯度参数得到标准疾病类型检测模型;
分类模块,用于接收待检测图像数据,将所述待检测图像数据依次通过所述标准疾病类型检测模型的卷积层,归一化层,线性整流层,随机失活层,全连接层及逻辑回归层,得到疾病类型检测结果。
本申请还提供一种应用于服务端的疾病类型检测装置,所述装置包括:
解密模块,用于开启K个监听端口,其中,K为客户端的数量,利用所述监听端口接收多个客户端发送的加密后的模型梯度参数,对所述加密后的模型梯度参数进行解密,得到每个客户端对应的模型梯度参数;
参数更新模块,用于对所述每个客户端对应的模型梯度参数执行联合运算,得到更新后的模型梯度参数,将所述更新后的模型梯度参数分发给每个客户端。
本申请还提供一种计算机可读存储介质,包括存储数据区和存储程序区,存储数据区存储创建的数据,存储程序区存储有计算机程序;其中,所述计算机程序被处理器执行时实现:
从服务端获取初始梯度参数,根据所述初始梯度参数构建疾病类型检测模型,通过本地的图像训练数据对所述疾病类型检测模型进行训练,得到训练后的模型并提取训练后的模型梯度参数;
访问服务端的监听端口,成功建立连接后,对所述模型梯度参数加密运算后上传至服务端;
接收服务端传送的更新后的模型梯度参数,根据所述更新后的模型梯度参数得到标准疾病类型检测模型;
接收待检测图像数据,将所述待检测图像数据依次通过所述标准疾病类型检测模型的卷积层,归一化层,线性整流层,随机失活层,全连接层及逻辑回归层,得到疾病类型检测结果。
本申请还提供一种计算机可读存储介质,包括存储数据区和存储程序区,存储数据区存储创建的数据,存储程序区存储有计算机程序;其中,所述计算机程序被处理器执行时实现:
开启K个监听端口,其中,K为客户端的数量;
利用所述监听端口接收多个客户端发送的加密后的模型梯度参数;
对所述加密后的模型梯度参数进行解密,得到每个客户端对应的模型梯度参数;
对所述每个客户端对应的模型梯度参数执行联合运算,得到更新后的模型梯度参数;
将所述更新后的模型梯度参数分发给每个客户端。
本申请还提供一种疾病类型检测方法,所述方法包括:
从服务端获取初始梯度参数,根据所述初始梯度参数构建疾病类型检测模型,通过本地的图像训练数据对所述疾病类型检测模型进行训练,得到所述疾病类型检测模型训练后的模型梯度参数;
访问服务端的监听端口,与所述服务端成功建立连接后,将训练后的所述模型梯度参数加密运算后上传至服务端;
接收服务端传送的更新后的模型梯度参数,根据所述更新后的模型梯度参数得到标准疾病类型检测模型;
接收待检测图像数据,将所述待检测图像数据依次通过所述标准疾病类型检测模型的卷积层,归一化层,线性整流层,随机失活层,全连接层及逻辑回归层,得到疾病类型检测结果。
本申请还提供一种疾病类型检测方法,所述方法包括:
开启K个监听端口,其中,K为客户端的数量;
利用所述监听端口接收多个客户端发送的加密后的模型梯度参数;
对所述加密后的模型梯度参数进行解密,得到每个客户端对应的模型梯度参数;
对所述每个客户端对应的模型梯度参数执行联合运算,得到更新后的模型梯度参数;
将所述更新后的模型梯度参数分发给每个客户端。
附图说明
图1为本申请第一实施例提供的电子设备的内部结构示意图;
图2为本申请第二实施例提供的疾病类型检测方法的流程示意图;
图3为本申请第三实施例提供的疾病类型检测方法的流程示意图;
图4为本申请第四实施例提供的疾病类型检测装置的模块示意图;
本申请目的的实现、功能特点及优点将结合实施例,参照附图做进一步说明。
具体实施方式
应当理解,此处所描述的具体实施例仅仅用以解释本申请,并不用于限定本申请。
本申请实施例提供一种电子设备。所述电子设备可以是,例如服务端、终端等的至少一种。所述服务端包括但不限于:单台服务端、服务端集群、云端服务端或云端服务端集群等。
如图1所示,为本申请一实施例提供的电子设备的内部结构示意图。
所述电子设备1可以包括处理器10、存储器11和总线,还可以包括存储在所述存储器11中并可在所述处理器10上运行的计算机程序,如疾病类型检测程序12。
其中,所述存储器11至少包括一种类型的可读存储介质,所述可读存储介质包括闪存、移动硬盘、多媒体卡、卡型存储器(例如:SD或DX存储器等)、磁性存储器、磁盘、光盘等。所述存储器11在一些实施例中可以是电子设备1的内部存储单元,例如该电子设备1的移动硬盘。所述存储器11在另一些实施例中也可以是电子设备1的外部存储设备,例如电子设备1上配备的插接式移动硬盘、智能存储卡(Smart Media Card,SMC)、安全数字(Secure Digital,SD)卡、闪存卡(Flash Card)等。进一步地,所述存储器11还可以既包括电子设备1的内部存储单元也包括外部存储设备。所述存储器11不仅可以用于存储安装于电子设备1的应用软件及各类数据,例如疾病类型检测程序12的代码等,还可以用于暂时地存储已经输出或者将要输出的数据。
所述处理器10在一些实施例中可以由集成电路组成,例如可以由单个封装的集成电路所组成,也可以是由多个相同功能或不同功能封装的集成电路所组成,包括一个或者多个中央处理器(Central Processing unit,CPU)、微处理器、数字处理芯片、图形处理器及各种控制芯片的组合等。所述处理器10是所述电子设备的控制核心(Control Unit),利用各种接口和线路连接整个电子设备的各个部件,通过运行或执行存储在所述存储器11内的程序或者模块(例如执行疾病类型检测程序等),以及调用存储在所述存储器11内的数据,以执行电子设备1的各种功能和处理数据。
所述总线可以是外设部件互连标准(peripheral component interconnect,简称PCI)总线或扩展工业标准结构(extended industry standard architecture,简称EISA)总线等。该总线可以分为地址总线、数据总线、控制总线等。所述总线被设置为实现所述存储器11以及至少一个处理器10等之间的连接通信。
图1仅示出了具有部件的电子设备,本领域技术人员可以理解的是,图1示出的结构 并不构成对所述电子设备1的限定,可以包括比图示更少或者更多的部件,或者组合某些部件,或者不同的部件布置。
例如,尽管未示出,所述电子设备1还可以包括给各个部件供电的电源(比如电池),优选地,电源可以通过电源管理装置与所述至少一个处理器10逻辑相连,从而通过电源管理装置实现充电管理、放电管理、以及功耗管理等功能。电源还可以包括一个或一个以上的直流或交流电源、再充电装置、电源故障检测电路、电源转换器或者逆变器、电源状态指示器等任意组件。所述电子设备1还可以包括多种传感器、蓝牙模块、Wi-Fi模块等,在此不再赘述。
进一步地,所述电子设备1还可以包括网络接口,可选地,所述网络接口可以包括有线接口和/或无线接口(如WI-FI接口、蓝牙接口等),通常用于在该电子设备1与其他电子设备之间建立通信连接。
可选地,该电子设备1还可以包括用户接口,用户接口可以是显示器(Display)、输入单元(比如键盘(Keyboard)),可选地,用户接口还可以是标准的有线接口、无线接口。可选地,在一些实施例中,显示器可以是LED显示器、液晶显示器、触控式液晶显示器以及OLED(Organic Light-Emitting Diode,有机发光二极管)触摸器等。其中,显示器也可以适当的称为显示屏或显示单元,用于显示在电子设备1中处理的信息以及用于显示可视化的用户界面。
应该了解,所述实施例仅为说明之用,在专利申请范围上并不受此结构的限制。
所述电子设备1中的所述存储器11存储的疾病类型检测程序12是多个指令的组合,在所述处理器10中运行时,可以实现一种疾病类型检测方法。详细地,所述疾病类型检测方法可参照下述关于图2所示的流程图中的描述。
参照图2所示,为本申请第二实施例提供的一种疾病类型检测方法的流程示意图。本实施例第二实施例中所述的疾病类型检测方法应用于客户端中,包括:
S11、从服务端获取初始梯度参数,根据所述初始梯度参数构建疾病类型检测模型,通过本地的图像训练数据对所述疾病类型检测模型进行训练,得到所述疾病类型检测模型训练后的模型梯度参数。
本申请实施例中,所述本地的图像训练数据包括医院的数据库中所保存的病人的心电图像。
本申请实施例利用所述本地的图像训练数据对从服务端端获取到的疾病类型检测模型进行训练,得到训练后的模型梯度参数。
S12、访问服务端的监听端口,成功建立连接后,将训练后的所述模型梯度参数进行加密后上传至服务端。
本申请实施例中,所述对所述模型梯度参数进行加密,包括:
随机选取大质数p,q,使得pq与(p-1)(q-1)的最大公倍数为1;
计算n=p×q,且满足λ(n)=lcm(p-1,q-1),其中,lcm为最小公倍数,λ为卡迈克尔函数;
随机选择一个小于n 2的正整数g,并计算μ=(L(g λmodn 2)) -1modn;
根据所述n、g、λ,和μ,得到公钥为(n,g),私钥为(λ,μ);
利用所述私钥(λ,μ)对所述模型梯度参数进行加密处理,得到加密后的模型梯度参数。
其中,质数是指在大于1的自然数中,除了1和它本身以外不再有其他因数的自然数,所述大质数则是指满足质数定义的自然数中最大的一个或者多个。
进一步地,本申请实施例将所述公钥传输给服务端,利用所述私钥(λ,μ)对所述模型梯度参数进行加密处理,得到加密后的模型梯度参数。
本申请实施例在多个客户端,如多个医院,执行所述疾病类型检测模型的训练,得到每个客户端对应的模型梯度参数。例如,所述多个医院可以包括A医院、B医院和C医院 等,A医院对应的模型梯度参数为w 1,B医院对应的模型梯度参数为w 2,C医院对应的模型梯度参数为w 3等,以此类推。本申请实施例利用所述私钥对所述模型梯度参数进行加密,方便后续上传更新,达到了拓展数据样本并保护病人隐私的目的。
进一步地,本申请实施例利用http协议通过服务端开的监听端口将加密后的模型梯度参数上传至所述服务端。
S13、接收服务端传送的更新后的模型梯度参数,根据所述更新后的模型梯度参数得到标准疾病类型检测模型。
本申请实施例利用服务端传送的更新后的模型梯度参数对所述疾病类型检测模型进行更新,得到标准疾病类型检测模型。
S14、接收待检测图像数据,将所述待检测图像数据依次通过所述标准疾病类型检测模型的卷积层,归一化层,线性整流层,随机失活层,全连接层及逻辑回归层,得到疾病类型检测结果。
本申请实施例中,接收待检测图像数据,将所述待检测图像数据依次通过所述标准疾病类型检测模型的卷积层,归一化层,线性整流层,随机失活层,全连接层及逻辑回归层,得到疾病类型检测结果。
其中,所述卷积层用于对所述待检测图像数据进行特征提取,所述归一化层用于防止梯度爆炸和梯度消失,所述线性整流层用于提高梯度下降和反向传播过程的效率,所述随机失活层用于实现所述标准疾病类型检测模型的正则化,降低其结构风险,所述全连接层用于将特征提取出的局部特征进行组装,以及所述逻辑回归层用于进行预测。
进一步地,所述待检测图像数据以及对应的注释数据包括心电图像与其对应的包含了心脏病专家的注释数据的json文件。
例如,本申请实施例获取病人的心电图数据,将所述心电图数据输入至所述标准疾病类型检测模型中,通过所述标准疾病类型检测模型的卷积层,归一化层,线性整流层,随机失活层,全连接层及逻辑回归层,得到诊断结果。其中,所述诊断结果是心脏疾病的具体分类,包括但不限于心房纤颤和扑动,房室传导阻滞,二联律,异位心房节律,交接性心律,窦性心律,室上性心动过速。
图3所示的流程示意图描述了本申请第三实施例提供的疾病类型检测方法。本实施例第三实施例所述的疾病类型检测方法应用于服务端,包括:
S21、开启K个监听端口,其中,K为客户端的数量。
本申请实施例中,所述服务端根据客户端的数量开启K个监听端口,以便与每一个客户端之间执行数据传输。
S22、利用所述监听端口接收多个客户端发送的加密后的模型梯度参数。
S23、对所述加密后的模型梯度参数进行解密,得到每个客户端对应的模型梯度参数。
本申请实施例利用所述公钥(n,g)对所述加密后的模型梯度参数进行解密处理,得到每个客户端对应的模型梯度参数。
具体地,所述对所述加密后的模型梯度参数进行解密,包括:
根据下述解密公式对所述加密后的模型梯度参数进行解密:
m=L(c λmodn 2)*μmodn
Figure PCTCN2020131987-appb-000001
其中,m是解密后的模型梯度参数,c是指加密后的模型梯度参数,mod是指取模运算符,n=p×q,其中,p,q为满足pq与(p-1)(q-1)的最大公倍数为1的大质数,λ为卡迈克尔函数,μ为预设参数。
例如,所述各个客户端包括但不限于A医院、B医院、C医院,各个客户端对所述加密后的模型梯度参数进行解密处理,得到每个客户端对应的模型梯度参数。
S24、对所述每个客户端对应的模型梯度参数执行联合运算,得到更新后的模型梯度参数。
本申请实施例中,所述对所述每个客户端对应的模型梯度参数执行联合运算,得到更新后的模型梯度参数,包括:
采用下述方法执行联合运算,得到更新后的模型梯度参数:
Figure PCTCN2020131987-appb-000002
Figure PCTCN2020131987-appb-000003
其中,f(w)为更新后的模型梯度参数,f i(w)为模型梯度参数,F k(w)表示中间参数,K为所述客户端数量,P k代表存储在第k个客户端中的训练数据,n k为训练数据的数量。
S25、将所述更新后的模型梯度参数分发给每个客户端。
本申请实施例中,对所述客户端与所述服务端进行安全连接,得到成功建立连接的所述客户端和所述服务端,将所述更新后的模型梯度参数分发给每个客户端。
如图4所示,是本申请第四实施例提供的疾病类型检测装置的模块示意图。
本申请所述疾病类型检测装置可以被划分为第一疾病类型检测装置100以及第二疾病类型检测装置200。其中,所述第一疾病类型检测装置100可以安装于客户端中以及所述第二疾病类型检测装置200可以安装在服务端中。
根据实现的功能,所述第一疾病类型检测装置100可以包括加密模块101、模型更新模块102及分类模块103;以及所述第二疾病类型检测装置200可以包括解密模块201及参数更新模块202。
本申请所述模块也可以称之为单元,是指一种能够被电子设备处理器所执行,并且能够完成固定功能的一系列计算机程序段,其存储在电子设备的存储器中。
在本实施例中,所述第一疾病类型检测装置100中和所述第一疾病类型检测装置200各模块的功能如下:
所述加密模块101,用于从服务端获取初始梯度参数,根据所述初始梯度参数构建疾病类型检测模型,通过本地的图像训练数据对所述疾病类型检测模型进行训练,得到所述疾病类型检测模型的训练后的模型梯度参数,访问服务端的监听端口,成功建立连接后,将训练后的所述模型梯度参数进行加密后上传至服务端;
所述解密模块201,用于开启K个监听端口,其中,K为客户端的数量,利用所述监听端口接收多个客户端发送的加密后的模型梯度参数,对所述加密后的模型梯度参数进行解密,得到每个客户端对应的模型梯度参数;
所述参数更新模块202,用于对所述每个客户端对应的模型梯度参数执行联合运算,得到更新后的模型梯度参数,将所述更新后的模型梯度参数分发给每个客户端。
所述模型更新模块102,用于接收服务端传送的更新后的模型梯度参数,根据所述更新后的模型梯度参数得到标准疾病类型检测模型;
所述分类模块103,用于接收待检测图像数据,将所述待检测图像数据依次通过所述标准疾病类型检测模型的卷积层,归一化层,线性整流层,随机失活层,全连接层及逻辑回归层,得到疾病类型检测结果。
详细地,所述第一疾病类型检测装置100各模块在客户端中执行下述操作:
步骤一、所述加密模块101从服务端获取初始梯度参数,根据所述初始梯度参数构建疾病类型检测模型,通过本地的图像训练数据对所述疾病类型检测模型进行训练,得到所 述疾病类型检测模型训练后的模型梯度参数;本申请实施例中,所述本地的图像训练数据包括医院的数据库中所保存的病人的心电图像。
本申请实施例利用所述本地的图像训练数据对从服务端端获取到的疾病类型检测模型进行训练,得到训练后的模型梯度参数。
步骤二、所述加密模块101访问服务端的监听端口,成功建立连接后,
对所述模型梯度参数进行加密后上传至服务端。
本申请实施例中,所述加密模块101对所述模型梯度参数进行加密,包括:
随机选取大质数p,q,使得pq与(p-1)(q-1)的最大公倍数为1;
计算n=p×q,且满足λ(n)=lcm(p-1,q-1),其中,lcm为最小公倍数,λ为卡迈克尔函数;
随机选择一个小于n 2的正整数g,并计算μ=(L(g λmodn 2)) -1modn;
根据所述n、g、λ,和μ,得到公钥为(n,g),私钥为(λ,μ);
利用所述私钥(λ,μ)对所述模型梯度参数进行加密处理,得到加密后的模型梯度参数。
其中,质数是指在大于1的自然数中,除了1和它本身以外不再有其他因数的自然数,所述大质数则是指满足质数定义的自然数中最大的一个或者多个。
进一步地,本申请实施例将所述公钥传输给服务端,利用所述私钥(λ,μ)对所述模型梯度参数进行加密处理,得到加密后的模型梯度参数。
本申请实施例在多个客户端,如多个医院,执行所述疾病类型检测模型的训练,得到每个客户端对应的模型梯度参数。例如,所述多个医院可以包括A医院、B医院和C医院等,A医院对应的模型梯度参数为w 1,B医院对应的模型梯度参数为w 2,C医院对应的模型梯度参数为w 3等,以此类推。本申请实施例利用所述私钥对所述模型梯度参数进行加密,方便后续上传更新,达到了拓展数据样本并保护病人隐私的目的。
进一步地,本申请实施例利用http协议通过服务端开的监听端口将加密后的模型梯度参数上传至所述服务端。
步骤三、所述模型更新模块102接收服务端传送的更新后的梯度参数,根据所述更新后的模型梯度参数得到标准疾病类型检测模型。
本申请实施例所述模型更新模块102利用服务端传送的更新后的模型梯度参数对所述疾病类型检测模型进行更新,得到标准疾病类型检测模型。
步骤四、所述分类模块103接收待检测图像数据,将所述待检测图像数据依次通过所述标准疾病类型检测模型的卷积层,归一化层,线性整流层,随机失活层,全连接层及逻辑回归层,得到疾病类型检测结果。
其中,所述卷积层用于对所述待检测图像数据进行特征提取,所述归一化层用于防止梯度爆炸和梯度消失,所述线性整流层用于提高梯度下降和反向传播过程的效率,所述随机失活层用于实现所述标准疾病类型检测模型的正则化,降低其结构风险,所述全连接层用于将特征提取出的局部特征进行组装,以及所述逻辑回归层用于进行预测。
本申请实施例中,所述分类模块103接收待检测图像数据,将所述待检测图像数据依次通过所述标准疾病类型检测模型的卷积层,归一化层,线性整流层,随机失活层,全连接层及逻辑回归层,得到疾病类型检测结果。
进一步地,所述待检测图像数据以及对应的注释数据包括心电图像与其对应的包含了心脏病专家的注释数据的json文件。
例如,本申请实施例获取病人的心电图数据,将所述心电图数据输入至所述标准疾病类型检测模型中,通过所述标准疾病类型检测模型的卷积层,归一化层,线性整流层,随机失活层,全连接层及逻辑回归层,得到诊断结果。其中,所述诊断结果是心脏疾病的具体分类,包括但不限于心房纤颤和扑动,房室传导阻滞,二联律,异位心房节律,交接性心律,窦性心律,室上性心动过速。
详细地,所述第二疾病类型检测装置200各模块在服务端中执行下述操作:
步骤一、所述解密模块201开启K个监听端口,其中,K为客户端的数量。
本申请实施例中,所述服务端根据客户端的数量开启K个监听端口,以便与每一个客户端之间执行数据传输。
步骤二、所述解密模块201利用所述监听端口接收多个客户端发送的加密后的模型梯度参数。
步骤三、所述解密模块201对所述加密后的模型梯度参数进行解密,得到每个客户端对应的模型梯度参数。
本申请实施例所述解密模块201利用所述公钥(n,g)对所述加密后的模型梯度参数进行解密处理,得到每个客户端对应的模型梯度参数。
具体地,所述对所述加密后的模型梯度参数进行解密,包括:
根据下述解密公式对所述加密后的模型梯度参数进行解密:
m=L(c λmodn 2)*μmodn
Figure PCTCN2020131987-appb-000004
其中,m是解密后的模型梯度参数,c是指加密后的模型梯度参数,mod是指取模运算符,n=p×q,其中,p,q为满足pq与(p-1)(q-1)的最大公倍数为1的大质数,λ为卡迈克尔函数,μ为预设参数。
例如,所述各个客户端包括但不限于A医院、B医院、C医院,各个客户端对所述加密后的模型梯度参数进行解密处理,得到每个客户端对应的模型梯度参数。
步骤四、所述参数更新模块202对所述每个客户端对应的模型梯度参数执行联合运算,得到更新后的模型梯度参数。
本申请实施例中,所述参数更新模块202对所述每个客户端对应的模型梯度参数执行联合运算,得到更新后的模型梯度参数,包括:
采用下述方法执行联合运算,得到更新后的模型梯度参数:
Figure PCTCN2020131987-appb-000005
Figure PCTCN2020131987-appb-000006
其中,f(w)为更新后的模型梯度参数,f i(w)为模型梯度参数,F k(w)表示中间参数,K为所述客户端数量,P k代表存储在第k个客户端中的训练数据,n k为训练数据的数量。
步骤五、所述参数更新模块202将所述更新后的模型梯度参数分发给每个客户端。
本申请实施例中,所述参数更新模块202对所述客户端与所述服务端进行安全连接,得到成功建立连接的所述客户端和所述服务端,将所述更新后的模型梯度参数分发给每个客户端。
进一步地,所述疾病类型检测装置的模块/单元如果以软件功能单元的形式实现并作为独立的产品销售或使用时,可以存储在一个计算机可读取存储介质中,所述计算机可读存储介质可以是易失性,也可以是非易失性。所述计算机可读介质可以包括:能够携带所述计算机程序代码的任何实体或装置、记录介质、U盘、移动硬盘、磁碟、光盘、计算机存储器、只读存储器(ROM,Read-Only Memory)。
进一步地,所述计算机可用存储介质可主要包括存储程序区和存储数据区,其中,存储程序区可存储操作系统、至少一个功能所需的应用程序等;存储数据区可存储根据区块链节点的使用所创建的数据等。
所述计算机可读存储介质的存储程序区存储有计算机程序;其中,所述计算机程序被 处理器执行时实现如下步骤:
从服务端获取初始梯度参数,根据所述初始梯度参数构建疾病类型检测模型,通过本地的图像训练数据对所述疾病类型检测模型进行训练,得到所述疾病类型检测模型的训练后的模型梯度参数;
访问服务端的监听端口,与所述服务端成功建立连接后,将训练后的所述模型梯度参数加密运算后上传至服务端;
接收服务端传送的更新后的模型梯度参数,根据所述更新后的模型梯度参数得到标准疾病类型检测模型;
接收待检测图像数据,将所述待检测图像数据依次通过所述标准疾病类型检测模型的卷积层,归一化层,线性整流层,随机失活层,全连接层及逻辑回归层,得到疾病类型检测结果。
进一步的,所述计算机可读存储介质的存储程序区存储有计算机程序;其中,所述计算机程序被处理器执行时实现如下步骤:
开启K个监听端口,其中,K为客户端的数量;
利用所述监听端口接收多个客户端发送的加密后的模型梯度参数;
对所述加密后的模型梯度参数进行解密,得到每个客户端对应的模型梯度参数;
对所述每个客户端对应的模型梯度参数执行联合运算,得到更新后的模型梯度参数;
将所述更新后的模型梯度参数分发给每个客户端。
在本申请所提供的几个实施例中,应该理解到,所揭露的设备,装置和方法,可以通过其它的方式实现。例如,以上所描述的装置实施例仅仅是示意性的,例如,所述模块的划分,仅仅为一种逻辑功能划分,实际实现时可以有另外的划分方式。
所述作为分离部件说明的模块可以是或者也可以不是物理上分开的,作为模块显示的部件可以是或者也可以不是物理单元,即可以位于一个地方,或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部模块来实现本实施例方案的目的。
另外,在本申请各个实施例中的各功能模块可以集成在一个处理单元中,也可以是各个单元单独物理存在,也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式实现,也可以采用硬件加软件功能模块的形式实现。
对于本领域技术人员而言,显然本申请不限于上述示范性实施例的细节,而且在不背离本申请的精神或基本特征的情况下,能够以其他的具体形式实现本申请。
因此,无论从哪一点来看,均应将实施例看作是示范性的,而且是非限制性的,本申请的范围由所附权利要求而不是上述说明限定,因此旨在将落在权利要求的等同要件的含义和范围内的所有变化涵括在本申请内。不应将权利要求中的任何附关联图表记视为限制所涉及的权利要求。
此外,显然“包括”一词不排除其他单元或步骤,单数不排除复数。系统权利要求中陈述的多个单元或装置也可以由一个单元或装置通过软件或者硬件来实现。第二等词语用来表示名称,而并不表示任何特定的顺序。
最后应说明的是,以上实施例仅用以说明本申请的技术方案而非限制,尽管参照较佳实施例对本申请进行了详细说明,本领域的普通技术人员应当理解,可以对本申请的技术方案进行修改或等同替换,而不脱离本申请技术方案的精神和范围。

Claims (20)

  1. 一种电子设备,其中,所述电子设备包括:
    至少一个处理器;以及
    与所述至少一个处理器通信连接的存储器;其中,所述存储器存储有可被所述至少一个处理器执行的计算机程序,所述计算机程序被所述至少一个处理器执行,以使所述至少一个处理器执行如下步骤:
    从服务端获取初始梯度参数,根据所述初始梯度参数构建疾病类型检测模型,通过本地的图像训练数据对所述疾病类型检测模型进行训练,得到所述疾病类型检测模型的训练后的模型梯度参数;
    访问服务端的监听端口,与所述服务端成功建立连接后,将训练后的所述模型梯度参数加密运算后上传至服务端;
    接收服务端传送的更新后的模型梯度参数,根据所述更新后的模型梯度参数得到标准疾病类型检测模型;
    接收待检测图像数据,将所述待检测图像数据依次通过所述标准疾病类型检测模型的卷积层,归一化层,线性整流层,随机失活层,全连接层及逻辑回归层,得到疾病类型检测结果。
  2. 如权利要求1所述的电子设备,其中,所述对所述模型梯度参数进行加密后上传至服务端,包括:
    随机选取大质数p,q,使得pq与(p-1)(q-1)的最大公倍数为1;
    计算n=p×q,且满足λ(n)=lcm(p-1,q-1),其中,lcm为最小公倍数,λ为卡迈克尔函数;
    随机选择一个小于n 2的正整数g,并计算μ=(L(g λmodn 2)) -1modn;
    根据所述n、g、λ,和μ,得到公钥为(n,g),私钥为(λ,μ);
    利用所述私钥(λ,μ)对所述模型梯度参数进行加密处理,得到加密后的模型梯度参数。
  3. 如权利要求1所述的电子设备,其中,所述待检测图像数据以及对应的注释数据包括心电图像与其对应的包含了心脏病专家的注释数据的json文件。
  4. 一种电子设备,其中,所述电子设备包括:
    至少一个处理器;以及
    与所述至少一个处理器通信连接的存储器;其中,所述存储器存储有可被所述至少一个处理器执行的计算机程序,所述计算机程序被所述至少一个处理器执行,以使所述至少一个处理器执行如下步骤:
    开启K个监听端口,其中,K为客户端的数量;
    利用所述监听端口接收多个客户端发送的加密后的模型梯度参数;
    对所述加密后的模型梯度参数进行解密,得到每个客户端对应的模型梯度参数;
    对所述每个客户端对应的模型梯度参数执行联合运算,得到更新后的模型梯度参数;
    将所述更新后的模型梯度参数分发给每个客户端。
  5. 如权利要求4所述的电子设备,其中,所述对所述每个客户端对应的模型梯度参数执行联合运算,得到更新后的模型梯度参数,包括:
    采用下述方法执行联合运算,得到更新后的模型梯度参数:
    Figure PCTCN2020131987-appb-100001
    Figure PCTCN2020131987-appb-100002
    其中,f(w)为更新后的模型梯度参数,f i(w)为模型梯度参数,F k(w)表示中间参数,K为所述客户端数量,P k代表存储在第k个客户端中的训练数据,n k为训练数据的数量。
  6. 如权利要求4所述的电子设备,其中,所述对所述加密后的模型梯度参数进行解密,包括:
    根据下述解密公式对所述加密后的模型梯度参数进行解密:
    m=L(c λmodn 2)*μmodn
    Figure PCTCN2020131987-appb-100003
    其中,m是解密后的模型梯度参数,c是指加密后的模型梯度参数,mod是指取模运算符,n=p×q,其中,p,q为满足pq与(p-1)(q-1)的最大公倍数为1的大质数,λ为卡迈克尔函数,μ为预设参数。
  7. 一种疾病类型检测装置,其中,所述装置应用于客户端电子设备,所述装置包括:
    加密模块,用于从服务端获取初始梯度参数,根据所述初始梯度参数构建疾病类型检测模型,通过本地的图像训练数据对所述疾病类型检测模型进行训练,得到所述疾病类型检测模型的训练后的模型梯度参数,访问服务端的监听端口,成功建立连接后,将训练后的所述模型梯度参数进行加密后上传至服务端;
    模型更新模块,用于接收服务端传送的更新后的模型梯度参数,根据所述更新后的模型梯度参数得到标准疾病类型检测模型;
    分类模块,用于接收待检测图像数据,将所述待检测图像数据依次通过所述标准疾病类型检测模型的卷积层,归一化层,线性整流层,随机失活层,全连接层及逻辑回归层,得到疾病类型检测结果。
  8. 一种疾病类型检测装置,其中,所述装置应用于服务端电子设备,所述装置包括:
    解密模块,用于开启K个监听端口,其中,K为客户端的数量,利用所述监听端口接收多个客户端发送的加密后的模型梯度参数,对所述加密后的模型梯度参数进行解密,得到每个客户端对应的模型梯度参数;
    参数更新模块,用于对所述每个客户端对应的模型梯度参数执行联合运算,得到更新后的模型梯度参数,将所述更新后的模型梯度参数分发给每个客户端。
  9. 一种计算机可读存储介质,包括存储数据区和存储程序区,存储数据区存储创建的数据,存储程序区存储有计算机程序;其中,所述计算机程序被处理器执行时实现如下步骤:
    从服务端获取初始梯度参数,根据所述初始梯度参数构建疾病类型检测模型,通过本地的图像训练数据对所述疾病类型检测模型进行训练,得到所述疾病类型检测模型的训练后的模型梯度参数;
    访问服务端的监听端口,与所述服务端成功建立连接后,将训练后的所述模型梯度参数加密运算后上传至服务端;
    接收服务端传送的更新后的模型梯度参数,根据所述更新后的模型梯度参数得到标准疾病类型检测模型;
    接收待检测图像数据,将所述待检测图像数据依次通过所述标准疾病类型检测模型的卷积层,归一化层,线性整流层,随机失活层,全连接层及逻辑回归层,得到疾病类型检测结果。
  10. 如权利要求9所述的计算机可读存储介质,其中,所述对所述模型梯度参数进行加密后上传至服务端,包括:
    随机选取大质数p,q,使得pq与(p-1)(q-1)的最大公倍数为1;
    计算n=p×q,且满足λ(n)=lcm(p-1,q-1),其中,lcm为最小公倍数,λ为卡迈克尔函数;
    随机选择一个小于n 2的正整数g,并计算μ=(L(g λmodn 2)) -1modn;
    根据所述n、g、λ,和μ,得到公钥为(n,g),私钥为(λ,μ);
    利用所述私钥(λ,μ)对所述模型梯度参数进行加密处理,得到加密后的模型梯度参数。
  11. 如权利要求9所述的计算机可读存储介质,其中,所述待检测图像数据以及对应的注释数据包括心电图像与其对应的包含了心脏病专家的注释数据的json文件。
  12. 一种计算机可读存储介质,包括存储数据区和存储程序区,存储数据区存储创建的数据,存储程序区存储有计算机程序;其中,所述计算机程序被处理器执行时实现如下步骤:
    开启K个监听端口,其中,K为客户端的数量;
    利用所述监听端口接收多个客户端发送的加密后的模型梯度参数;
    对所述加密后的模型梯度参数进行解密,得到每个客户端对应的模型梯度参数;
    对所述每个客户端对应的模型梯度参数执行联合运算,得到更新后的模型梯度参数;
    将所述更新后的模型梯度参数分发给每个客户端。
  13. 如权利要求12所述的计算机可读存储介质,其中,所述对所述每个客户端对应的模型梯度参数执行联合运算,得到更新后的模型梯度参数,包括:
    采用下述方法执行联合运算,得到更新后的模型梯度参数:
    Figure PCTCN2020131987-appb-100004
    Figure PCTCN2020131987-appb-100005
    其中,f(w)为更新后的模型梯度参数,f i(w)为模型梯度参数,F k(w)表示中间参数,K为所述客户端数量,p k代表存储在第k个客户端中的训练数据,n k为训练数据的数量。
  14. 如权利要求12述的计算机可读存储介质,其中,所述对所述加密后的模型梯度参数进行解密,包括:
    根据下述解密公式对所述加密后的模型梯度参数进行解密:
    m=L(c λmodn 2)*μmodn
    Figure PCTCN2020131987-appb-100006
    其中,m是解密后的模型梯度参数,c是指加密后的模型梯度参数,mod是指取模运算符,n=p×q,其中,p,q为满足pq与(p-1)(q-1)的最大公倍数为1的大质数,λ为卡迈克尔函数,μ为预设参数。
  15. 一种疾病类型检测方法,其中,所述方法包括:
    从服务端获取初始梯度参数,根据所述初始梯度参数构建疾病类型检测模型,通过本地的图像训练数据对所述疾病类型检测模型进行训练,得到所述疾病类型检测模型的训练后的模型梯度参数;
    访问服务端的监听端口,与所述服务端成功建立连接后,将训练后的所述模型梯度参数加密运算后上传至服务端;
    接收服务端传送的更新后的模型梯度参数,根据所述更新后的模型梯度参数得到标准疾病类型检测模型;
    接收待检测图像数据,将所述待检测图像数据依次通过所述标准疾病类型检测模型的卷积层,归一化层,线性整流层,随机失活层,全连接层及逻辑回归层,得到疾病类型检测结果。
  16. 如权利要求15所述的疾病类型检测方法,其中,所述对所述模型梯度参数进行加密后上传至服务端,包括:
    随机选取大质数p,q,使得pq与(p-1)(q-1)的最大公倍数为1;
    计算n=p×q,且满足λ(n)=lcm(p-1,q-1),其中,lcm为最小公倍数,λ为卡迈克尔函数;
    随机选择一个小于n 2的正整数g,并计算μ=(L(g λmodn 2)) -1modn;
    根据所述n、g、λ,和μ,得到公钥为(n,g),私钥为(λ,μ);
    利用所述私钥(λ,μ)对所述模型梯度参数进行加密处理,得到加密后的模型梯度参数。
  17. 如权利要求15所述的疾病类型检测方法,其中,所述待检测图像数据以及对应的注释数据包括心电图像与其对应的包含了心脏病专家的注释数据的json文件。
  18. 一种疾病类型检测方法,其中,所述方法包括:
    开启K个监听端口,其中,K为客户端的数量;
    利用所述监听端口接收多个客户端发送的加密后的模型梯度参数;
    对所述加密后的模型梯度参数进行解密,得到每个客户端对应的模型梯度参数;
    对所述每个客户端对应的模型梯度参数执行联合运算,得到更新后的模型梯度参数;
    将所述更新后的模型梯度参数分发给每个客户端。
  19. 如权利要求18所述的疾病类型检测方法,其中,所述对所述每个客户端对应的模型梯度参数执行联合运算,得到更新后的模型梯度参数,包括:
    采用下述方法执行联合运算,得到更新后的模型梯度参数:
    Figure PCTCN2020131987-appb-100007
    Figure PCTCN2020131987-appb-100008
    其中,f(w)为更新后的模型梯度参数,f i(w)为模型梯度参数,F k(w)表示中间参数,K为所述客户端数量,P k代表存储在第k个客户端中的训练数据,n k为训练数据的数量。
  20. 如权利要求18所述的疾病类型检测方法,其中,所述对所述加密后的模型梯度参数进行解密,包括:
    根据下述解密公式对所述加密后的模型梯度参数进行解密:
    m=L(c λmodn 2)*μmodn
    Figure PCTCN2020131987-appb-100009
    其中,m是解密后的模型梯度参数,c是指加密后的模型梯度参数,mod是指取模运算符,n=p×q,其中,p,q为满足pq与(p-1)(q-1)的最大公倍数为1的大质数,λ为卡迈克尔函数,μ为预设参数。
PCT/CN2020/131987 2020-10-20 2020-11-27 电子设备、疾病类型检测方法、装置及介质 WO2021189907A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202011127805.4A CN112259238A (zh) 2020-10-20 2020-10-20 电子设备、疾病类型检测方法、装置及介质
CN202011127805.4 2020-10-20

Publications (1)

Publication Number Publication Date
WO2021189907A1 true WO2021189907A1 (zh) 2021-09-30

Family

ID=74244023

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/131987 WO2021189907A1 (zh) 2020-10-20 2020-11-27 电子设备、疾病类型检测方法、装置及介质

Country Status (2)

Country Link
CN (1) CN112259238A (zh)
WO (1) WO2021189907A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116913519A (zh) * 2023-07-24 2023-10-20 东莞莱姆森科技建材有限公司 基于智能镜的健康监护方法、装置、设备及存储介质

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113610205B (zh) * 2021-07-15 2022-12-27 深圳宇晰科技有限公司 一种基于机器视觉的二维码生成方法、装置及存储介质

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109545385A (zh) * 2018-11-30 2019-03-29 周立广 一种基于物联网的医疗大数据分析处理系统及其方法
US20190324856A1 (en) * 2018-04-18 2019-10-24 EMC IP Holding Company LLC Optimization of checkpoint operations for deep learning computing
CN111180061A (zh) * 2019-12-09 2020-05-19 广东工业大学 融合区块链与联邦学习的共享医疗数据智能辅助诊断系统

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110444263B (zh) * 2019-08-21 2024-04-26 深圳前海微众银行股份有限公司 基于联邦学习的疾病数据处理方法、装置、设备及介质
CN110795477A (zh) * 2019-09-20 2020-02-14 平安科技(深圳)有限公司 数据的训练方法及装置、系统
CN110797124B (zh) * 2019-10-30 2024-04-12 腾讯科技(深圳)有限公司 一种模型多端协同训练方法、医疗风险预测方法和装置
CN111190487A (zh) * 2019-12-30 2020-05-22 中国科学院计算技术研究所 一种建立数据分析模型的方法
CN111325726A (zh) * 2020-02-19 2020-06-23 腾讯医疗健康(深圳)有限公司 模型训练方法、图像处理方法、装置、设备及存储介质
CN111652863A (zh) * 2020-05-27 2020-09-11 刘君茹 一种医学影像检测方法、装置、设备和存储介质

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20190324856A1 (en) * 2018-04-18 2019-10-24 EMC IP Holding Company LLC Optimization of checkpoint operations for deep learning computing
CN109545385A (zh) * 2018-11-30 2019-03-29 周立广 一种基于物联网的医疗大数据分析处理系统及其方法
CN111180061A (zh) * 2019-12-09 2020-05-19 广东工业大学 融合区块链与联邦学习的共享医疗数据智能辅助诊断系统

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116913519A (zh) * 2023-07-24 2023-10-20 东莞莱姆森科技建材有限公司 基于智能镜的健康监护方法、装置、设备及存储介质

Also Published As

Publication number Publication date
CN112259238A (zh) 2021-01-22

Similar Documents

Publication Publication Date Title
US20210201268A1 (en) Multi-factor authentication for remote access of patient data
WO2021189907A1 (zh) 电子设备、疾病类型检测方法、装置及介质
US10572626B2 (en) Advanced telemedicine system with virtual doctor
US9135608B2 (en) Systems and methods for constructing a local electronic medical record data store using a remote personal health record server
US10765873B2 (en) Systems and methods for EMS device communications interface
US8977572B2 (en) Systems and methods for patient-controlled, encrypted, consolidated medical records
US11586742B2 (en) Data processing method, data processing device, and computer readable storage medium
US20170303119A1 (en) Information processing system, method of obtaining monitor information, and sensor device
WO2021208701A1 (zh) 代码变更的注释生成方法、装置、电子设备及存储介质
CN109361704A (zh) 云存储数据加密传输方法、系统、设备及存储介质
WO2022126903A1 (zh) 图像异常区域检测方法、装置、电子设备及存储介质
US20210392118A1 (en) Clinical infrastructure with features for the prevention of egress of private information
US20240184924A1 (en) Systems and methods for implementing a secure database for storing a patient operational longitudinal record
Srivastava et al. Automated remote patient monitoring: Data sharing and privacy using blockchain
CN102779285A (zh) 便携式电子健康病历卡
WO2021189903A1 (zh) 基于音频的用户状态识别方法、装置、电子设备及存储介质
WO2019095552A1 (zh) 区域医疗电子病历安全协同整合系统及方法
JP6707545B2 (ja) 健康管理情報の捕捉および管理
CN112398645A (zh) 一种基于区块链技术的医疗数据共享方法
TWI328249B (zh)
US20220020491A1 (en) Systems and methods for ems device communications interface
WO2022116422A1 (zh) 产品推荐方法、装置、电子设备及计算机可读存储介质
US20110145016A1 (en) Secure data card
NandhaKumar et al. A Survey on E-Health Care for Diabetes Using Cloud Framework
US12009110B1 (en) Method for using a clinician station to provide medical services remotely

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20928001

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20928001

Country of ref document: EP

Kind code of ref document: A1