WO2021159740A1 - 一种样本生成方法、装置、服务器及存储介质 - Google Patents

一种样本生成方法、装置、服务器及存储介质 Download PDF

Info

Publication number
WO2021159740A1
WO2021159740A1 PCT/CN2020/124610 CN2020124610W WO2021159740A1 WO 2021159740 A1 WO2021159740 A1 WO 2021159740A1 CN 2020124610 W CN2020124610 W CN 2020124610W WO 2021159740 A1 WO2021159740 A1 WO 2021159740A1
Authority
WO
WIPO (PCT)
Prior art keywords
compliance condition
patient
generation model
trained
original
Prior art date
Application number
PCT/CN2020/124610
Other languages
English (en)
French (fr)
Inventor
张渊
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 WO2021159740A1 publication Critical patent/WO2021159740A1/zh

Links

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
    • G16H20/70—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to mental therapies, e.g. psychological therapy or autogenous training
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems

Definitions

  • This application relates to the field of artificial intelligence technology, and in particular to a sample generation method, device, server, and storage medium.
  • the hierarchical management of patients with different compliances by medical staff can help improve the work efficiency of medical staff and reduce the workload.
  • the inventor realizes that the hierarchical management of patients depends on analyzing the patient's medical behavior pattern, and predicting the patient's compliance based on this, and identifying the signs of patients with different compliances. According to the characteristics of patients with different compliances, medical staff can guide the patient's medical behavior. Therefore, how to guide patients' medical behaviors based on compliance has become an urgent problem to be solved.
  • the embodiments of the present application provide a sample generation method, device, service, and storage medium, which can guide the patient's medical behavior based on compliance.
  • an embodiment of the present application provides a server, including a processor and a memory, the processor and the memory are connected to each other, wherein the memory is used to store a computer program, and the computer program includes program instructions,
  • the processor is configured to call the program instructions to perform the following steps: receiving a patient sample generation request sent by a terminal device, the patient sample generation request carrying preset compliance conditions; invoking a pre-trained depth generation model, According to the compliance condition, a target patient sample corresponding to the compliance condition is generated, and the target patient sample is used to guide the medical treatment behavior of patients meeting the compliance condition; and the target patient sample is sent to the terminal Equipment so that the terminal device displays the target patient sample.
  • an embodiment of the present application provides a sample generation method, including: receiving a patient sample generation request sent by a terminal device, the patient sample generation request carrying preset compliance conditions; calling a pre-trained depth generation model, According to the compliance condition, a target patient sample corresponding to the compliance condition is generated, and the target patient sample is used to guide the medical treatment behavior of patients meeting the compliance condition; and the target patient sample is sent to the terminal Equipment so that the terminal device displays the target patient sample.
  • an embodiment of the present application provides a sample generation device, including: a communication module for receiving a patient sample generation request sent by a terminal device, the patient sample generation request carrying preset compliance conditions; a processing module, Used to call a pre-trained deep generation model to generate a target patient sample corresponding to the compliance condition according to the compliance condition, and the target patient sample is used to guide the medical treatment behavior of patients who meet the compliance condition; The communication module is also used to send the target patient sample to the terminal device, so that the terminal device can display the target patient sample.
  • an embodiment of the present application provides a computer-readable storage medium, the computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement the following method: receiving a patient sample sent by a terminal device Generate a request, the patient sample generation request carries a preset compliance condition; the pre-trained depth generation model is called to generate a target patient sample corresponding to the compliance condition according to the compliance condition, and the target patient sample is used To guide the medical treatment behavior of patients who meet the compliance condition; send the target patient sample to the terminal device, so that the terminal device displays the target patient sample.
  • This application uses a pre-trained deep generation model to generate corresponding target patient samples according to compliance conditions for guiding the patient's medical behavior, thereby realizing the process of guiding the patient's medical behavior based on compliance.
  • Fig. 1 is a schematic flowchart of a sample generation method provided by an embodiment of the present application.
  • Fig. 2 is a schematic diagram of a model training process provided by an embodiment of the present application.
  • FIG. 3 is a schematic flowchart of another sample generation method provided by an embodiment of the present application.
  • Fig. 4 is a schematic diagram of a network architecture of a sample generation system provided by an embodiment of the present application.
  • Fig. 5 is a schematic structural diagram of a sample generating device provided by an embodiment of the present application.
  • Fig. 6 is a schematic structural diagram of a server provided by an embodiment of the present application.
  • the technical solution of this application can be applied to the fields of artificial intelligence, digital medical care, smart city, blockchain and/or big data technology to realize smart medical care.
  • the data involved in this application such as patient samples, can be stored in a database, or can be stored in a blockchain, or can be stored in other ways, and this application is not limited.
  • FIG. 1 is a schematic flowchart of a sample generation method provided by an embodiment of this application.
  • This method can be applied to the server.
  • the server can be a server or a server cluster. Specifically, the method may include the following steps.
  • S101 Receive a patient sample generation request sent by a terminal device, where the patient sample generation request carries preset compliance conditions.
  • the terminal devices here include, but are not limited to, smart terminals such as notebook computers and desktop computers.
  • the compliance condition may include at least one of the following: compliance level, age group, and gender.
  • the compliance level here can be, for example, good, good, medium, or poor. Compliance here is also called compliance and compliance. Compliance refers to the consistency of the patient's behavior with the doctor's order. Compliance not only affects the patient's normal recovery, but even interferes with the normal progress of medical care.
  • a doctor can use a terminal device to enter the auxiliary diagnosis page, and after inputting compliance conditions based on the auxiliary diagnosis page, click the sample generation button included in the auxiliary diagnosis page, and the terminal device can respond to the sample generation button. Click the operation to send a patient sample generation request to the server.
  • the patient sample request carries preset compliance conditions, and the server can receive the patient sample generation request sent by the terminal device.
  • the deep generative model may be a Generative Adversarial Networks (GAN) model or a Variational Auto-encoder (VAE).
  • GAN Generative Adversarial Networks
  • VAE Variational Auto-encoder
  • the target patient sample refers to the optimal sample corresponding to the compliance condition.
  • the target patient sample may include patient data.
  • the patient data may include at least one of the following: basic patient information, medical visit information, disease information, and medication information.
  • the basic information of the patient may include information such as gender and age
  • the medical treatment information may include information such as the type of hospital visited, the frequency of medical visits
  • the disease information may include information about basic diseases and information about medical diseases.
  • the pre-trained depth generation model can be obtained in the following manner: the server obtains the first patient data corresponding to each compliance condition in at least one compliance condition, and corresponds to each compliance condition according to the Training the original depth generation model to obtain the pre-trained depth generation model.
  • the first patient data may refer to real patient data used for model training.
  • the first patient data may be one or more.
  • the server trains the original depth generation model according to the first patient data corresponding to each compliance condition
  • the process of obtaining the pretrained depth generation model may be: the server corresponds to each compliance condition Perform normalization processing on the first patient data to obtain the first processed data.
  • the first processed data here refers to data obtained after normalization processing is performed on the first patient data corresponding to each compliance condition.
  • the server uses the first processed data as input data of the original depth generation model to train the original depth generation model to obtain a pre-trained depth generation model.
  • the server uses the first processed data as the input data of the original depth generation model to train the original depth generation model to obtain pre-training
  • the process of generating the model in depth can be specifically as follows.
  • the server uses the first processed data as the input data of the original variational autoencoder, and performs encoding processing according to the first processed data through the encoding module included in the original variational autoencoder to obtain the mean value and variance .
  • the first patient data may be patient data x
  • the server may normalize the patient data x corresponding to each compliance condition and input it into the original variational autoencoder. Specifically, the server may obtain the first processed data after normalizing the patient data x corresponding to each compliance condition, and use the first processed data as the input data of the original variational autoencoder, and pass the
  • the encoding encoder module included in the original variational autoencoder performs encoding processing according to the first processed data to obtain the mean value and the variance.
  • the encoding module may be a neural network, such as a 5-layer convolutional neural network.
  • the encoding process can be understood as the compression process.
  • the server samples the Gaussian distribution that the mean and variance obey to obtain hidden variables.
  • the server can sample the Gaussian distribution that the mean and variance obey to obtain the hidden variable Z.
  • the server inputs the latent variable into the decoding module included in the original variational autoencoder, and performs decoding processing according to the latent variable through the decoding module to obtain the second patient corresponding to each compliance condition data.
  • the server can input the latent variable Z into the decoding decoder module included in the original variational autoencoder, and use the decoding module to decode the latent variable Z to obtain the second corresponding to each compliance condition.
  • Patient data that is, patient data x'.
  • the server performs decoding processing according to the hidden variable Z through the decoding module to obtain the patient data x′ corresponding to each compliance condition: the server performs decoding processing according to the hidden variable Z through the decoding module to obtain the second processed data , And perform denormalization processing on the second processed data to obtain the patient data x'corresponding to each compliance condition.
  • the second processed data here refers to the data obtained after decoding processing according to the hidden variable Z.
  • the second patient data refers to the patient data generated by the model, and specifically may be the patient data generated by the decoding model included in the variational autoencoder.
  • the server constructs the loss function of the original variational autoencoder according to the first patient data and the second patient data, and uses the loss function to train the original variational autoencoder to obtain pre-training The variational self-encoder.
  • the server can construct the loss function of the original variational autoencoder according to the patient data x and patient data x', and train the original variational autoencoder by using the loss function to obtain the pre-trained variation. Sub-self encoder.
  • the server trains the original variational autoencoder according to the first patient data corresponding to each compliance condition
  • the process of obtaining the pre-trained variational autoencoder may be:
  • the first patient data corresponding to the sexual condition is used as the input data of the original variational autoencoder to train the original variational autoencoder to obtain the pre-trained variational autoencoder.
  • the server may input the first patient data into the original variational autoencoder to train the original variational autoencoder.
  • the server uses the first patient data corresponding to each compliance condition as the input data of the original variational autoencoder to train the original variational autoencoder to obtain the pre-trained variational autoencoder
  • the process of the device may be: the server uses the first patient data corresponding to each compliance condition as the input data of the original variational autoencoder, and the encoding module included in the original variational autoencoder is processed according to the first The data is encoded to obtain the mean value and variance; the server samples the Gaussian distribution that the mean value and variance obey to obtain the hidden variable, and inputs the hidden variable to the decoding module included in the original variational autoencoder, through the decoding The module performs decoding processing according to the hidden variable to obtain the second patient data corresponding to each compliance condition; the server constructs the loss function of the original variational autoencoder according to the first patient data and the second patient data, Use the loss function to train the original variational autoencoder to obtain a pre-trained variational autoencoder.
  • the server calls the pre-trained depth generation model to generate the target patient sample corresponding to the compliance condition according to the compliance condition:
  • the trained variational self-encoder includes a decoding module to generate a target patient sample corresponding to the compliance condition according to the compliance condition.
  • the server uses the decoding module included in the pre-trained variational autoencoder to generate the compliance condition according to the compliance condition.
  • the process of the corresponding target patient sample may be: the server obtains output data according to the compliance condition through the decoding module included in the pre-trained variational self-encoder, and performs denormalization processing on the output data to obtain the compliance condition The corresponding target patient sample.
  • the server may send the target patient sample to the terminal device, and the terminal device may display the target patient sample.
  • patients can be accurately managed based on compliance.
  • you can input the compliance condition (such as 60 years old, good compliance, male) and output the optimal patient sample under the compliance condition.
  • the medical staff can guide the medical treatment behavior of the patient who meets the compliance condition based on the sample.
  • the server can receive a patient sample generation request sent by the terminal device, and the patient sample generation request carries a preset compliance condition; the server invokes a pre-trained deep generation model according to the compliance The condition generates a target patient sample corresponding to the compliance condition, and the target patient sample is used to guide the medical treatment behavior of patients meeting the compliance condition; the server sends the target patient sample to the terminal device so that the terminal device can display the target patient Samples, in this embodiment of the application, a pre-trained depth generation model is used to generate corresponding target patient samples according to compliance conditions to guide the patient's medical behavior, thereby realizing the process of guiding the patient's medical behavior based on compliance .
  • This application can be used in the field of medical technology and involves blockchain technology.
  • target patient samples or compressed data of target patient samples can be written into the blockchain.
  • FIG. 3 is a schematic flowchart of another sample generation method provided by an embodiment of this application.
  • This method can be applied to the server.
  • the server can be a server or a server cluster. Specifically, the method may include the following steps.
  • S301 Receive a patient sample generation request sent by a terminal device, where the patient sample generation request carries preset compliance conditions.
  • step S301 to step S303 may refer to step S101 to step S103 in the embodiment of FIG.
  • S304 Determine a target psychological care strategy that matches the target patient sample.
  • the server can determine the target psychological care strategy matching the target patient sample, and send the target psychological care strategy to the terminal device, so that the terminal device can display the target psychological care strategy.
  • the target psychological care strategy here refers to the psychological care strategy that matches the target patient sample.
  • the server can obtain the target patient's patient data, and use the target patient's patient data to update the target psychological care strategy , Get the updated psychological care strategy, and send the updated psychological care strategy to the terminal device for display.
  • the updated psychological care strategy will be more suitable for the patient's personal situation for psychological care, which is more conducive to improving the patient's compliance.
  • the server may also match the information of the target medical staff who executes the psychological counseling strategy from the medical staff information collection, and send the information of the target medical staff to the terminal device for display.
  • the matching manner may include a matching manner determined according to the historical psychological counseling data of the medical staff or a matching manner determined according to the work content of the medical staff, etc., which are not described in detail in the embodiment of the present application.
  • the server may determine the target psychological care strategy matching the target patient sample, and send the target psychological care strategy to the terminal device, so that the terminal device can display the target psychological care strategy according to the target psychological care strategy.
  • the target psychological nursing strategy guides the target patient's medical treatment behavior, which is conducive to improving the patient's compliance.
  • FIG. 4 is a schematic diagram of a network architecture of a sample generation system provided by an embodiment of this application.
  • the sample generation system shown in FIG. 4 may include a server 10 and a terminal device 20.
  • the terminal device 20 can send a patient sample generation request to the server 10
  • the server 10 can generate the target patient sample according to the compliance conditions carried in the patient sample generation request and the pre-trained depth generation model by performing steps S101 and S102
  • pass Step S103 is executed to display the target patient sample through the terminal device 20, thereby realizing the process of guiding the patient's medical behavior according to compliance.
  • the sample generation apparatus may include: a communication module 501, configured to receive a patient sample generation request sent by a terminal device, the patient sample generation request carrying preset compliance conditions.
  • the processing module 502 is configured to call a pre-trained depth generation model to generate a target patient sample corresponding to the compliance condition according to the compliance condition, and the target patient sample is used to guide the patients who meet the compliance condition Medical treatment.
  • the communication module 501 is further configured to send the target patient sample to the terminal device, so that the terminal device can display the target patient sample.
  • the processing module 502 is further configured to obtain first patient data corresponding to each compliance condition in the at least one compliance condition; according to the first patient data corresponding to each compliance condition , Train the original depth generation model, and get the pre-trained depth generation model.
  • the processing module 502 trains the original depth generation model according to the first patient data corresponding to each compliance condition to obtain a pre-trained depth generation model, specifically for each of the The first patient data corresponding to the compliance condition is normalized to obtain the first processed data; the first processed data refers to the data obtained after the normalization processing is performed on the first patient data corresponding to each compliance condition ; Use the first processed data as input data of the original depth generation model to train the original depth generation model to obtain a pre-trained depth generation model.
  • the depth generation model is a variational autoencoder
  • the processing module 502 uses the first processed data as input data of the original depth generation model to generate the original depth
  • the model is trained to obtain a pre-trained depth generation model.
  • the first processed data is used as the input data of the original variational autoencoder, and the encoding module included in the original variational autoencoder is used according to the The first processed data is encoded to obtain the mean value and variance; the Gaussian distribution subject to the mean value and variance is sampled to obtain a hidden variable; the hidden variable is input to the decoding module included in the original variational autoencoder ,
  • the second patient data corresponding to each compliance condition is obtained by the decoding module according to the hidden variable; the original patient data is constructed according to the first patient data and the second patient data
  • the loss function of the variational autoencoder is used to train the original variational autoencoder to obtain a pre-trained variational autoencoder.
  • the processing module 502 performs decoding processing according to the latent variable through the decoding module to obtain the second patient data corresponding to each compliance condition, specifically through the decoding module according to The latent variable is decoded to obtain the second processed data; the second processed data refers to the data obtained after the decoding process is performed on the latent variable; the second processed data is denormalized to obtain the The second patient data corresponding to each compliance condition.
  • the processing module 502 trains the original variational autoencoder according to the first patient data corresponding to each compliance condition to obtain a pre-trained variational autoencoder, specifically In order to use the first patient data corresponding to each compliance condition as the input data of the original variational autoencoder, to train the original variational autoencoder to obtain a pre-trained variational autoencoder.
  • the processing module 502 is further configured to determine a target psychological care strategy matching the target patient sample; the target psychological care strategy is sent to the terminal device through the communication module 501, so that The terminal device displays the target psychological care strategy.
  • the sample generation device can receive a patient sample generation request sent by the terminal device, and the patient sample generation request carries preset compliance conditions; the sample generation device calls the pre-trained depth generation model to According to the compliance condition, a target patient sample corresponding to the compliance condition is generated, and the target patient sample is used to guide the medical treatment behavior of patients meeting the compliance condition; the sample generating device sends the target patient sample to the terminal device for the The terminal device displays the target patient sample.
  • the embodiment of the present application generates the corresponding target patient sample according to the compliance condition through a pre-trained deep generation model for guiding the patient's medical behavior, thereby realizing the patient's compliance based on the compliance The process of guiding medical treatment.
  • FIG. 6 is a schematic structural diagram of a server provided by an embodiment of this application.
  • the server described in this embodiment may include: one or more processors 1000 and a memory 2000.
  • the processor 1000 and the memory 2000 may be connected by a bus.
  • the processor 1000 may be a central processing unit (Central Processing Unit, CPU), and the processor may also be other general-purpose processors, digital signal processors (digital signal processors, DSP), and application-specific integrated circuits (Application Specific Integrated Circuits). Specific Integrated Circuit, ASIC), ready-made programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc.
  • the general-purpose processor may be a microprocessor or the processor may also be any conventional processor or the like.
  • the memory 2000 can be a high-speed RAM memory or a non-volatile memory (non-volatile memory), such as disk storage.
  • the memory 2000 is used to store a computer program, the computer program includes program instructions, and the processor 1000 is configured to call the program instructions to perform the following steps: receiving a patient sample generation request sent by a terminal device, and the patient sample generation Request to carry a preset compliance condition; call a pre-trained deep generation model to generate a target patient sample corresponding to the compliance condition according to the compliance condition, and the target patient sample is used to guide satisfying the compliance condition The patient’s medical behavior; sending the target patient sample to the terminal device so that the terminal device can display the target patient sample.
  • the embodiment of the present application may receive the patient sample generation request sent by the terminal device through the input device (not shown in the figure).
  • the target patient sample may be sent to the terminal device through an output device (not shown in the figure).
  • the input device and output device can be a standard wired/wireless interface.
  • the processor 1000 is configured to call the program instructions, and further perform the following steps: obtain the first patient data corresponding to each compliance condition in the at least one compliance condition; The first patient data corresponding to the condition trains the original depth generation model to obtain the pre-trained depth generation model.
  • the processor 1000 when the original depth generation model is trained according to the first patient data corresponding to each compliance condition to obtain the pre-trained depth generation model, the processor 1000 is configured to call the program instructions , Perform the following steps: normalize the first patient data corresponding to each compliance condition to obtain first processed data; the first processed data refers to the first patient corresponding to each compliance condition Data obtained after normalization of the data; using the first processed data as input data of the original depth generation model to train the original depth generation model to obtain a pre-trained depth generation model.
  • the depth generation model is a variational autoencoder
  • the first processed data is used as the input data of the original depth generation model to train the original depth generation model to obtain a prediction
  • the processor 1000 is configured to call the program instructions to perform the following steps: use the first processed data as the input data of the original variational autoencoder, and pass the original variational autoencoder.
  • the encoding module included in the autoencoder performs encoding processing according to the first processed data to obtain a mean value and a variance; samples the Gaussian distribution subject to the mean value and variance to obtain a hidden variable; and inputs the hidden variable to the original
  • the variational autoencoder includes a decoding module, through which the decoding module performs decoding processing according to the hidden variables to obtain the second patient data corresponding to each compliance condition; according to the first patient data and the For the second patient data, a loss function of the original variational autoencoder is constructed, and the original variational autoencoder is trained by using the loss function to obtain a pre-trained variational autoencoder.
  • the processor 1000 when the decoding module performs decoding processing according to the hidden variables to obtain the second patient data corresponding to each compliance condition, the processor 1000 is configured to call the program instructions, The following steps are performed: the decoding module performs decoding processing according to the hidden variable to obtain second processed data; the second processed data refers to the data obtained after decoding processing according to the hidden variable; and the second processed data Perform denormalization processing to obtain the second patient data corresponding to each compliance condition.
  • the processor 1000 when the original variational autoencoder is trained according to the first patient data corresponding to each compliance condition to obtain the pre-trained variational autoencoder, the processor 1000 is configured to call The program instructions execute the following steps: use the first patient data corresponding to each compliance condition as the original variational autoencoder input data to train the original variational autoencoder to obtain pre-trained Variational self-encoder.
  • the processor 1000 is configured to call the program instructions, and further perform the following steps: determine a target psychological care strategy that matches the target patient sample; send the target psychological care strategy to the terminal Device so that the terminal device displays the target psychological care strategy.
  • the embodiment of the present application may send the target psychological care strategy to the terminal device through an output device.
  • the processor 1000 described in the embodiment of the present application can perform the implementation described in the embodiment of FIG. 1 and the embodiment of FIG. .
  • the functional modules in each embodiment of the present application may be integrated into one processing module, or each module may exist alone physically, or two or more modules may be integrated into one module.
  • the above-mentioned integrated modules can be implemented in the form of sampling hardware, and can also be implemented in the form of sampling software functional modules.
  • the embodiment of the present application also provides a computer-readable storage medium, and the computer-readable storage medium stores a computer program.
  • the function of each module/unit of the device in the above-mentioned embodiment is realized during execution, which will not be repeated here.
  • the storage medium involved in this application such as a computer-readable storage medium, may be non-volatile or volatile.
  • the program can be stored in a computer readable storage medium. During execution, it may include the procedures of the above-mentioned method embodiments.
  • the computer-readable storage medium may be volatile or non-volatile.
  • the computer storage medium may be a magnetic disk, an optical disk, a read-only memory (Read-Only Memory, ROM), or a random access memory (Random Access Memory). Memory, RAM) etc.
  • the computer-readable storage medium may mainly include a storage program area and a storage data area, where 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 information based on the blockchain node Use the created data, etc.
  • Blockchain essentially a decentralized database, is a series of data blocks associated with cryptographic methods. Each data block contains a batch of network transaction information for verification. The validity of the information (anti-counterfeiting) and the generation of the next block.
  • the blockchain can include the underlying platform of the blockchain, the platform product service layer, and the application service layer.

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Public Health (AREA)
  • Medical Informatics (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Linguistics (AREA)
  • Epidemiology (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Biophysics (AREA)
  • Primary Health Care (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Pathology (AREA)
  • Databases & Information Systems (AREA)
  • Child & Adolescent Psychology (AREA)
  • Developmental Disabilities (AREA)
  • Hospice & Palliative Care (AREA)
  • Psychiatry (AREA)
  • Psychology (AREA)
  • Social Psychology (AREA)
  • Medical Treatment And Welfare Office Work (AREA)

Abstract

一种样本生成方法、装置、服务器及存储介质，应用于医疗科技领域，该服务器包括处理器和存储器，处理器和存储器相互连接，其中，存储器用于存储计算机程序，计算机程序包括程序指令，处理器被配置用于调用程序指令，执行以下步骤：接收终端设备发送的患者样本生成请求，患者样本生成请求携带预置的依从性条件（S101）；调用预训练的深度生成模型，以根据依从性条件生成依从性条件对应的目标患者样本，目标患者样本用于指导满足依从性条件的患者的就医行为（S102）；将目标患者样本发送至终端设备，以便终端设备展示目标患者样本（S103）。可以根据依从性对患者的就医行为进行指导。涉及区块链技术，可将目标患者样本写入区块链。

Description

一种样本生成方法、装置、服务器及存储介质
本申请要求于2020年9月28日提交中国专利局、申请号为202011045854.3，发明名称为“一种样本生成方法、装置、服务器及存储介质”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
本申请涉及人工智能技术领域，尤其涉及一种样本生成方法、装置、服务器及存储介质。
 背景技术
在诊疗过程中，患者的依从性对患者预后有很大影响，医护人员对依从性不同的患者进行分级管理有助于提高医护人员的工作效率，减轻工作负担。发明人意识到，对患者进行分级管理有赖于对患者的就医行为模式进行分析，并据此对患者的依从性进行预测，识别出不同依从性的患者的表征。根据不同依从性的患者的表征，医护人员可以对患者的就医行为进行指导。因此，如何依据依从性对患者的就医行为进行指导成为亟待解决的问题。
 技术问题
本申请实施例提供了一种样本生成方法、装置、服务及存储介质，可以根据依从性对患者的就医行为进行指导。
 技术解决方案
第一方面，本申请实施例提供了一种服务器，包括处理器和存储器，所述处理器和所述存储器相互连接，其中，所述存储器用于存储计算机程序，所述计算机程序包括程序指令，所述处理器被配置用于调用所述程序指令，执行以下步骤：接收终端设备发送的患者样本生成请求，所述患者样本生成请求携带预置的依从性条件；调用预训练的深度生成模型，以根据所述依从性条件生成所述依从性条件对应的目标患者样本，所述目标患者样本用于指导满足所述依从性条件的患者的就医行为；将所述目标患者样本发送至所述终端设备，以便所述终端设备展示所述目标患者样本。
第二方面，本申请实施例提供了一种样本生成方法，包括：接收终端设备发送的患者样本生成请求，所述患者样本生成请求携带预置的依从性条件；调用预训练的深度生成模型，以根据所述依从性条件生成所述依从性条件对应的目标患者样本，所述目标患者样本用于指导满足所述依从性条件的患者的就医行为；将所述目标患者样本发送至所述终端设备，以便所述终端设备展示所述目标患者样本。
第三方面，本申请实施例提供了一种样本生成装置，包括：通信模块，用于接收终端设备发送的患者样本生成请求，所述患者样本生成请求携带预置的依从性条件；处理模块，用于调用预训练的深度生成模型，以根据所述依从性条件生成所述依从性条件对应的目标患者样本，所述目标患者样本用于指导满足所述依从性条件的患者的就医行为；所述通信模块，还用于将所述目标患者样本发送至所述终端设备，以便所述终端设备展示所述目标患者样本。
第四方面，本申请实施例提供了一种计算机可读存储介质，所述计算机可读存储介质存储有计算机程序，所述计算机程序被处理器执行以实现以下方法：接收终端设备发送的患者样本生成请求，所述患者样本生成请求携带预置的依从性条件；调用预训练的深度生成模型，以根据所述依从性条件生成所述依从性条件对应的目标患者样本，所述目标患者样本用于指导满足所述依从性条件的患者的就医行为；将所述目标患者样本发送至所述终端设备，以便所述终端设备展示所述目标患者样本。
 有益效果
本申请通过预训练的深度生成模型来根据依从性条件生成对应的目标患者样本以用于对患者的就医行为进行指导，从而实现了根据依从性对患者的就医行为进行指导的过程。
 附图说明
图1是本申请实施例提供的一种样本生成方法的流程示意图。
图2是本申请实施例提供的一种模型训练过程的示意图。
图3是本申请实施例提供的另一种样本生成方法的流程示意图。
图4是本申请实施例提供的一种样本生成系统的网络架构示意图。
图5是本申请实施例提供的一种样本生成装置的结构示意图。
图6是本申请实施例提供的一种服务器的结构示意图。
 本发明的实施方式
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行描述。
本申请的技术方案可应用于人工智能、数字医疗、智慧城市、区块链和/或大数据技术领域，以实现智慧医疗。可选的，本申请涉及的数据如患者样本等可存储于数据库中，或者可以存储于区块链中，或者可采用其他方式存储，本申请不做限定。
请参阅图1，为本申请实施例提供的一种样本生成方法的流程示意图。该方法可以应用于服务器。该服务器可以为一个服务器或服务器集群。具体地，该方法可以包括以下步骤。
S101、接收终端设备发送的患者样本生成请求，所述患者样本生成请求携带预置的依从性条件。
其中，此处的终端设备包括但不限于笔记本电脑、台式电脑等智能终端。依从性条件可以包括以下至少一项：依从性等级、年龄段、性别。此处的依从性等级如可以为好、较好、中或差。此处的依从性也称顺从性、顺应性。依从性是指患者行为与医嘱的一致性，依从性不但影响患者的正常康复，甚至会干扰医护工作的正常进行。
在一个的应用场景中，医生可以使用终端设备进入辅助诊断页面，并在基于该辅助诊断页面输入依从性条件后，点击辅助诊断页面包括的样本生成按钮，终端设备可以响应于对样本生成按钮的点击操作，发送患者样本生成请求至服务器，该患者样本请求携带预置的依从性条件，服务器可以接收终端设备发送的患者样本生成请求。
S102、调用预训练的深度生成模型，以根据所述依从性条件生成所述依从性条件对应的目标患者样本，所述目标患者样本用于指导满足所述依从性条件的患者的就医行为。
其中，该深度生成模型可以为生成对抗网络（Generative Adversarial Networks ，GAN）模型或变分自编码器（Variational Auto-encoder，VAE）。目标患者样本是指该依从性条件对应的最优样本。目标患者样本可以包括患者数据。该患者数据可以包括以下至少一项：患者基本信息、就诊信息、疾病信息、用药信息。该患者基本信息如可以包括性别、年龄等信息，就诊信息如可以包括就诊医院的类别、就诊的频次等信息，疾病信息如可以包括基础疾病的信息以及就诊疾病的信息。
在一个实施例中，所述的预训练的深度生成模型可以通过如下方式得到：服务器获取至少一个依从性条件中每个依从性条件对应的第一患者数据，并根据该每个依从性条件对应的第一患者数据，训练该原始的深度生成模型，得到预训练的深度生成模型。其中，第一患者数据可以是指用于模型训练的真实患者数据。该第一患者数据可以为一个或多个。
在一个实施例中，服务器根据该每个依从性条件对应的第一患者数据，训练该原始的深度生成模型，得到预训练的深度生成模型的过程可以为：服务器对该每个依从性条件对应的第一患者数据进行归一化处理，得到第一处理数据。此处的第一处理数据是指根据该每个依从性条件对应的第一患者数据进行归一化处理后得到的数据。服务器将该第一处理数据作为原始的深度生成模型的输入数据，以对该原始的深度生成模型进行训练，得到预训练的深度生成模型。通过对各第一患者数据进行归一化处理，可以使得不同数量级的第一患者数据处于同一数量级，便于模型训练。
在一个实施例中，在该深度生成模型为变分自编码器时，服务器将该第一处理数据作为原始的深度生成模型的输入数据，以对该原始的深度生成模型进行训练，得到预训练的深度生成模型的过程具体可以如下。
①服务器将所述第一处理数据作为原始的变分自编码器的输入数据，通过所述原始的变分自编码器包括的编码模块根据所述第一处理数据进行编码处理，得到均值和方差。
结合图2来看，第一患者数据可以为患者数据x，服务器可以将每个依从性条件对应的患者数据x归一化处理后输入原始的变分自编码器。具体地，服务器可以在对每个依从性条件对应的患者数据x进行归一化处理后，得到第一处理数据，并将第一处理数据作为原始的变分自编码器的输入数据，通过该原始的变分自编码器包括的编码encoder模块根据该第一处理数据进行编码处理，得到均值和方差。其中，该编码模块可以为神经网络，比如5层卷积神经网络。该编码处理的过程可以理解为压缩处理的过程。
②服务器对所述均值和方差服从的高斯分布进行采样，得到隐变量。
结合图2来看，服务器可以对所述均值和方差服从的高斯分布进行采样，得到隐变量Z。
③服务器将所述隐变量输入到所述原始的变分自编码器包括的解码模块，通过所述解码模块根据所述隐变量进行解码处理，得到所述每个依从性条件对应的第二患者数据。
结合图2来看，服务器可以将隐变量Z输入到原始的变分自编码器包括的解码decoder模块，并通过该解码模块根据隐变量Z进行解码处理，得到每个依从性条件对应的第二患者数据，也就是患者数据x’。
此处，服务器通过解码模块根据隐变量Z进行解码处理，得到每个依从性条件对应的患者数据x’的过程可以为：服务器通过解码模块根据该隐变量Z进行解码处理，得到第二处理数据，并对该第二处理数据进行反归一化处理，得到该每个依从性条件对应的患者数据x’。此处的第二处理数据是指根据该隐变量Z进行解码处理后得到的数据。第二患者数据是指模型生成的患者数据，具体可为变分自编码器包括的解码模型生成的患者数据。
④服务器根据所述第一患者数据和所述第二患者数据，构建所述原始的变分自编码器的损失函数，利用所述损失函数训练所述原始的变分自编码器，得到预训练的变分自编码器。
结合图2来看，服务器可以根据患者数据x与患者数据x’构建该原始的变分自编码器的损失函数，利用所述损失函数训练该原始的变分自编码器，得到预训练的变分自编码器。
在一个实施例中，服务器根据该每个依从性条件对应的第一患者数据，训练原始的变分自编码器，得到预训练的变分自编码器的过程可以为：服务器将该每个依从性条件对应的第一患者数据作为原始的变分自编码器的输入数据，以训练原始的变分自编码器，得到预训练的变分自编码器。此处，服务器可以将第一患者数据输入原始的变分自编码器，以训练该原始的变分自编码器。
在一个实施例中，服务器将该每个依从性条件对应的第一患者数据作为原始的变分自编码器的输入数据，以训练原始的变分自编码器，得到预训练的变分自编码器的过程可以为：服务器将该每个依从性条件对应的第一患者数据作为原始的变分自编码器的输入数据，通过该原始的变分自编码器包括的编码模块根据该第一处理数据进行编码处理，得到均值和方差；服务器对该均值和方差服从的高斯分布进行采样，得到隐变量，并将该隐变量输入到该原始的变分自编码器包括的解码模块，通过该解码模块根据该隐变量进行解码处理，得到该每个依从性条件对应的第二患者数据；服务器根据该第一患者数据和该第二患者数据，构建该原始的变分自编码器的损失函数，利用该损失函数训练该原始的变分自编码器，得到预训练的变分自编码器。
在一个实施例中，当深度生成模型为变分自编码器时，服务器调用预训练的深度生成模型，以根据该依从性条件生成该依从性条件对应的目标患者样本的过程为：服务器通过预训练的变分自编码器包括的解码模块以根据该依从性条件生成该依从性条件对应的目标患者样本。
在一个实施例中，当训练原始的变分自编码器的过程涉及归一化处理过程，服务器通过预训练的变分自编码器包括的解码模块，以根据该依从性条件生成该依从性条件对应的目标患者样本的过程可以为：服务器通过预训练的变分自编码器包括的解码模块以根据该依从性条件得到输出数据，对该输出数据进行反归一化处理，得到该依从性条件对应的目标患者样本。
S103、将所述目标患者样本发送至所述终端设备，以便所述终端设备展示所述目标患者样本。
本申请实施例中，服务器可以将目标患者样本发送至终端设备，终端设备可以展示该目标患者样本。
在一些智慧医疗的场景中，可以根据依从性对患者，尤其是慢性病患者进行精准管理。具体可以输入依从性条件（如60岁、依从性好、男性）后输出该依从性条件下的最优患者样本。在得到该最优患者样本后，医护人员可以根据此样本指导符合该依从性条件的患者的就医行为。
可见，图1所示的实施例中，服务器可以接收终端设备发送的患者样本生成请求，该患者样本生成请求携带预置的依从性条件；服务器调用预训练的深度生成模型，以根据该依从性条件生成该依从性条件对应的目标患者样本，该目标患者样本用于指导满足该依从性条件的患者的就医行为；服务器将该目标患者样本发送至该终端设备，以便该终端设备展示该目标患者样本，本申请实施例通过预训练的深度生成模型来根据依从性条件生成对应的目标患者样本以用于对患者的就医行为进行指导，从而实现了根据依从性对患者的就医行为进行指导的过程。
本申请可用于医疗科技领域，涉及区块链技术，如可将目标患者样本或目标患者样本的压缩数据写入区块链中。
请参阅图3，为本申请实施例提供的另一种样本生成方法的流程示意图。该方法可以应用于服务器。该服务器可以为一个服务器或服务器集群。具体地，该方法可以包括以下步骤。
S301、接收终端设备发送的患者样本生成请求，所述患者样本生成请求携带预置的依从性条件。
S302、调用预训练的深度生成模型，以根据所述依从性条件生成所述依从性条件对应的目标患者样本，所述目标患者样本用于指导满足所述依从性条件的患者的就医行为。
S303、将所述目标患者样本发送至所述终端设备，以便所述终端设备展示所述目标患者样本。
其中，步骤S301-步骤S303可以参见图1实施例中的步骤S101-步骤S103，本申请实施例在此不做赘述。
S304、确定与所述目标患者样本匹配的目标心理护理策略。
S305、将所述目标心理护理策略发送至所述终端设备，以便所述终端设备展示所述目标心理护理策略。
随着医学模式向生物-心理-社会模式转化，心理护理已经成为不可缺少的措施，特别是在如今医护人员对患者有告知义务，病人有知情同意权的时代，掌握患者心理护理知识，实施个性化护理，有着深远的现实意义。因此，本申请实施例中，服务器可以确定与该目标患者样本匹配的目标心理护理策略，并将该目标心理护理策略发送至该终端设备，以便该终端设备展示该目标心理护理策略。此处的目标心理护理策略是指与该目标患者样本匹配的心理护理策略。
在一个实施例中，考虑到患者样本包括的患者数据不一定与目标患者的患者数据完全匹配，因此服务器可以获取目标患者的患者数据，并利用目标患者的患者数据对该目标心理护理策略进行更新，得到更新后的心理护理策略，并将更新后的心理护理策略发送至终端设备以进行展示。更新后的心理护理策略将更加贴合患者的个人情况以进行心理护理，从而更有利于提升患者的依从性。
在一个实施例中，服务器还可以从医护人员信息集合，匹配出执行该心理辅导策略的目标医护人员的信息，并将该目标医护人员的信息发送至该终端设备以进行显示。匹配方式可以包括根据医护人员的历史心理辅导数据确定的匹配方式或根据医护人员的工作内容确定的匹配方式等方式，本申请实施例在此不做赘述。
图3所示的实施例中，服务器可以确定与该目标患者样本匹配的目标心理护理策略，并将该目标心理护理策略发送至该终端设备，以便该终端设备展示该目标心理护理策略，依据该目标心理护理策略指导目标患者的就诊行为，有利于提升患者的依从性。
请参阅图4，为本申请实施例提供的一种样本生成系统的网络架构示意图。图4所示的样本生成系统可以包括服务器10和终端设备20。其中：终端设备20可以向服务器10发送患者样本生成请求，服务器10可以通过执行步骤S101和步骤S102以根据患者样本生成请求携带的依从性条件和预训练的深度生成模型生成目标患者样本，并通过执行步骤S103以通过终端设备20展示该目标患者样本，进而实现根据依从性对患者的就医行为进行指导的过程。
请参阅图5，为本申请实施例提供的一种样本生成装置的结构示意图。该装置可以应用于前述提及的服务器。具体地，该样本生成装置可以包括：通信模块501，用于接收终端设备发送的患者样本生成请求，所述患者样本生成请求携带预置的依从性条件。处理模块502，用于调用预训练的深度生成模型，以根据所述依从性条件生成所述依从性条件对应的目标患者样本，所述目标患者样本用于指导满足所述依从性条件的患者的就医行为。通信模块501，还用于将所述目标患者样本发送至所述终端设备，以便所述终端设备展示所述目标患者样本。
在一种可选的实施方式中，处理模块502，还用于获取至少一个依从性条件中每个依从性条件对应的第一患者数据；根据所述每个依从性条件对应的第一患者数据，训练原始的深度生成模型，得到预训练的深度生成模型。
在一种可选的实施方式中，处理模块502根据所述每个依从性条件对应的第一患者数据，训练原始的深度生成模型，得到预训练的深度生成模型，具体为对所述每个依从性条件对应的第一患者数据进行归一化处理，得到第一处理数据；第一处理数据是指根据所述每个依从性条件对应的第一患者数据进行归一化处理后得到的数据；将所述第一处理数据作为原始的深度生成模型的输入数据，以对所述原始的深度生成模型进行训练，得到预训练的深度生成模型。
在一种可选的实施方式中，所述深度生成模型为变分自编码器，处理模块502将所述第一处理数据作为原始的深度生成模型的输入数据，以对所述原始的深度生成模型进行训练，得到预训练的深度生成模型，具体为将所述第一处理数据作为原始的变分自编码器的输入数据，通过所述原始的变分自编码器包括的编码模块根据所述第一处理数据进行编码处理，得到均值和方差；对所述均值和方差服从的高斯分布进行采样，得到隐变量；将所述隐变量输入到所述原始的变分自编码器包括的解码模块，通过所述解码模块根据所述隐变量进行解码处理，得到所述每个依从性条件对应的第二患者数据；根据所述第一患者数据和所述第二患者数据，构建所述原始的变分自编码器的损失函数，利用所述损失函数训练所述原始的变分自编码器，得到预训练的变分自编码器。
在一种可选的实施方式中，处理模块502通过所述解码模块根据所述隐变量进行解码处理，得到所述每个依从性条件对应的第二患者数据，具体为通过所述解码模块根据所述隐变量进行解码处理，得到第二处理数据；第二处理数据是指根据所述隐变量进行解码处理后得到的数据；对所述第二处理数据进行反归一化处理，得到所述每个依从性条件对应的第二患者数据。
在一种可选的实施方式中，处理模块502根据所述每个依从性条件对应的第一患者数据，训练所述原始的变分自编码器，得到预训练的变分自编码器，具体为将所述每个依从性条件对应的第一患者数据作为所述原始的变分自编码的输入数据，以训练所述原始的变分自编码器，得到预训练的变分自编码器。
在一种可选的实施方式中，处理模块502，还用于确定与所述目标患者样本匹配的目标心理护理策略；通过通信模块501将所述目标心理护理策略发送至所述终端设备，以便所述终端设备展示所述目标心理护理策略。
可见，图5所示的实施例中，样本生成装置可以接收终端设备发送的患者样本生成请求，该患者样本生成请求携带预置的依从性条件；样本生成装置调用预训练的深度生成模型，以根据该依从性条件生成该依从性条件对应的目标患者样本，该目标患者样本用于指导满足该依从性条件的患者的就医行为；样本生成装置将该目标患者样本发送至该终端设备，以便该终端设备展示该目标患者样本，本申请实施例通过预训练的深度生成模型来根据依从性条件生成对应的目标患者样本以用于对患者的就医行为进行指导，从而实现了根据依从性对患者的就医行为进行指导的过程。
请参阅图6，为本申请实施例提供的一种服务器的结构示意图。本实施例中所描述的服务器可以包括：一个或多个处理器1000和存储器2000。处理器1000、和存储器2000可以通过总线连接。
处理器1000可以是中央处理模块(Central Processing Unit，CPU)，该处理器还可以是其他通用处理器、数字信号处理器(Digital Signal Processor，DSP)、专用集成电路(Application Specific Integrated Circuit，ASIC)、现成可编程门阵列(Field-Programmable Gate Array，FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件等。通用处理器可以是微处理器或者该处理器也可以是任何常规的处理器等。
存储器2000可以是高速RAM存储器，也可为非不稳定的存储器（non-volatile memory），例如磁盘存储器。其中，存储器2000用于存储计算机程序，所述计算机程序包括程序指令，处理器1000被配置用于调用所述程序指令，执行以下步骤：接收终端设备发送的患者样本生成请求，所述患者样本生成请求携带预置的依从性条件；调用预训练的深度生成模型，以根据所述依从性条件生成所述依从性条件对应的目标患者样本，所述目标患者样本用于指导满足所述依从性条件的患者的就医行为；将所述目标患者样本发送至所述终端设备，以便所述终端设备展示所述目标患者样本。
在一个实施例中，本申请实施例可以通过输入装置（图未示）接收终端设备发送的患者样本生成请求。本申请实施例可以通过输出装置（图未示）将所述目标患者样本发送至终端设备。所述的输入装置和输出装置可以为标准的有线/无线接口。
在一个实施例中，处理器1000被配置用于调用所述程序指令，还执行以下步骤：获取至少一个依从性条件中每个依从性条件对应的第一患者数据；根据所述每个依从性条件对应的第一患者数据，训练原始的深度生成模型，得到预训练的深度生成模型。
在一个实施例中，在根据所述每个依从性条件对应的第一患者数据，训练原始的深度生成模型，得到预训练的深度生成模型时，处理器1000被配置用于调用所述程序指令，执行以下步骤：对所述每个依从性条件对应的第一患者数据进行归一化处理，得到第一处理数据；第一处理数据是指根据所述每个依从性条件对应的第一患者数据进行归一化处理后得到的数据；将所述第一处理数据作为原始的深度生成模型的输入数据，以对所述原始的深度生成模型进行训练，得到预训练的深度生成模型。
在一个实施例中，所述深度生成模型为变分自编码器，在将所述第一处理数据作为原始的深度生成模型的输入数据，以对所述原始的深度生成模型进行训练，得到预训练的深度生成模型时，处理器1000被配置用于调用所述程序指令，执行以下步骤：将所述第一处理数据作为原始的变分自编码器的输入数据，通过所述原始的变分自编码器包括的编码模块根据所述第一处理数据进行编码处理，得到均值和方差；对所述均值和方差服从的高斯分布进行采样，得到隐变量；将所述隐变量输入到所述原始的变分自编码器包括的解码模块，通过所述解码模块根据所述隐变量进行解码处理，得到所述每个依从性条件对应的第二患者数据；根据所述第一患者数据和所述第二患者数据，构建所述原始的变分自编码器的损失函数，利用所述损失函数训练所述原始的变分自编码器，得到预训练的变分自编码器。
在一个实施例中，在通过所述解码模块根据所述隐变量进行解码处理，得到所述每个依从性条件对应的第二患者数据时，处理器1000被配置用于调用所述程序指令，执行以下步骤：通过所述解码模块根据所述隐变量进行解码处理，得到第二处理数据；第二处理数据是指根据所述隐变量进行解码处理后得到的数据；对所述第二处理数据进行反归一化处理，得到所述每个依从性条件对应的第二患者数据。
在一个实施例中，在根据所述每个依从性条件对应的第一患者数据，训练原始的变分自编码器，得到预训练的变分自编码器时，处理器1000被配置用于调用所述程序指令，执行以下步骤：将所述每个依从性条件对应的第一患者数据作为原始的变分自编码的输入数据，以训练所述原始的变分自编码器，得到预训练的变分自编码器。
在一个实施例中，处理器1000被配置用于调用所述程序指令，还执行以下步骤：确定与所述目标患者样本匹配的目标心理护理策略；将所述目标心理护理策略发送至所述终端设备，以便所述终端设备展示所述目标心理护理策略。
在一个实施例中，本申请实施例可以通过输出装置将所述目标心理护理策略发送至所述终端设备。
具体实现中，本申请实施例中所描述的处理器1000可执行图1实施例、图3实施例所描述的实现方式，也可执行本申请实施例所描述的实现方式，在此不再赘述。
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模块中，也可以是各个模块单独物理存在，也可以是两个或两个以上模块集成在一个模块中。上述集成的模块既可以采样硬件的形式实现，也可以采样软件功能模块的形式实现。
本申请实施例还提供了一种计算机可读存储介质，该计算机可读存储介质上存储有计算机程序，计算机程序被处理器执行时实现上述实施例中方法的步骤，或者，计算机程序被处理器执行时实现上述实施例中装置的各模块/单元的功能，这里不再赘述。可选的，本申请涉及的存储介质如计算机可读存储介质可以是非易失性的，也可以是易失性的。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程，是可以通过计算机程序来指令相关的硬件来完成，所述的程序可存储于一计算机可读取存储介质中，该程序在执行时，可包括如上述各方法的实施例的流程。其中，所述的计算机可读存储介质可为易失性的或非易失性的。例如，该计算机存储介质可以为磁碟、光盘、只读存储记忆体（Read-Only Memory，ROM）或随机存储记忆体（Random Access Memory，RAM）等。所述的计算机可读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。
其中，本申请所指区块链是分布式数据存储、点对点传输、共识机制、加密算法等计算机技术的新型应用模式。区块链(Blockchain)，本质上是一个去中心化的数据库，是一串使用密码学方法相关联产生的数据块，每一个数据块中包含了一批次网络交易的信息，用于验证其信息的有效性(防伪)和生成下一个区块。区块链可以包括区块链底层平台、平台产品服务层以及应用服务层等。
以上所揭露的仅为本申请一种较佳实施例而已，当然不能以此来限定本申请之权利范围，本领域普通技术人员可以理解实现上述实施例的全部或部分流程，并依本申请权利要求所作的等同变化，仍属于本申请所涵盖的范围。

Claims (20)

  1. 一种服务器，包括处理器和存储器，所述处理器和所述存储器相互连接，所述存储器用于存储计算机程序，所述计算机程序包括程序指令，所述处理器被配置用于调用所述程序指令，执行以下步骤：
    接收终端设备发送的患者样本生成请求，所述患者样本生成请求携带预置的依从性条件；
    调用预训练的深度生成模型，以根据所述依从性条件生成所述依从性条件对应的目标患者样本，所述目标患者样本用于指导满足所述依从性条件的患者的就医行为；
    将所述目标患者样本发送至所述终端设备，以便所述终端设备展示所述目标患者样本。
  2. 根据权利要求1所述的服务器，其中，所述处理器被配置用于调用所述程序指令，还执行以下步骤：
    获取至少一个依从性条件中每个依从性条件对应的第一患者数据；
    根据所述每个依从性条件对应的第一患者数据，训练原始的深度生成模型，得到预训练的深度生成模型。
  3. 根据权利要求2所述的服务器，其中，在根据所述每个依从性条件对应的第一患者数据，训练原始的深度生成模型，得到预训练的深度生成模型时，所述处理器被配置用于调用所述程序指令，执行以下步骤：
    对所述每个依从性条件对应的第一患者数据进行归一化处理，得到第一处理数据；第一处理数据是指根据所述每个依从性条件对应的第一患者数据进行归一化处理后得到的数据；
    将所述第一处理数据作为原始的深度生成模型的输入数据，以对所述原始的深度生成模型进行训练，得到预训练的深度生成模型。
  4. 根据权利要求3所述的服务器，其中，所述深度生成模型为变分自编码器，在将所述第一处理数据作为原始的深度生成模型的输入数据，以对所述原始的深度生成模型进行训练，得到预训练的深度生成模型时，所述处理器被配置用于调用所述程序指令，执行以下步骤：
    将所述第一处理数据作为原始的变分自编码器的输入数据，通过所述原始的变分自编码器包括的编码模块根据所述第一处理数据进行编码处理，得到均值和方差；
    对所述均值和方差服从的高斯分布进行采样，得到隐变量；
    将所述隐变量输入到所述原始的变分自编码器包括的解码模块，通过所述解码模块根据所述隐变量进行解码处理，得到所述每个依从性条件对应的第二患者数据；
    根据所述第一患者数据和所述第二患者数据，构建所述原始的变分自编码器的损失函数，利用所述损失函数训练所述原始的变分自编码器，得到预训练的变分自编码器。
  5. 根据权利要求4所述的服务器，其中，在通过所述解码模块根据所述隐变量进行解码处理，得到所述每个依从性条件对应的第二患者数据时，所述处理器被配置用于调用所述程序指令，执行以下步骤：
    通过所述解码模块根据所述隐变量进行解码处理，得到第二处理数据；第二处理数据是指根据所述隐变量进行解码处理后得到的数据；
    对所述第二处理数据进行反归一化处理，得到所述每个依从性条件对应的第二患者数据。
  6. 根据权利要求2所述的服务器，其中，在根据所述每个依从性条件对应的第一患者数据，训练所述原始的变分自编码器，得到预训练的变分自编码器时，所述处理器被配置用于调用所述程序指令，执行以下步骤：
    将所述每个依从性条件对应的第一患者数据作为原始的变分自编码器的输入数据，以训练所述原始的变分自编码器，得到预训练的变分自编码器。
  7. 根据权利要求1-6任一项所述的服务器，其中，所述处理器被配置用于调用所述程序指令，还执行以下步骤：
    确定与所述目标患者样本匹配的目标心理护理策略；
    将所述目标心理护理策略发送至所述终端设备，以便所述终端设备展示所述目标心理护理策略。
  8. 一种样本生成方法，包括：
    接收终端设备发送的患者样本生成请求，所述患者样本生成请求携带预置的依从性条件；
    调用预训练的深度生成模型，以根据所述依从性条件生成所述依从性条件对应的目标患者样本，所述目标患者样本用于指导满足所述依从性条件的患者的就医行为；
    将所述目标患者样本发送至所述终端设备，以便所述终端设备展示所述目标患者样本。
  9. 根据权利要求8所述的方法，其中，所述方法还包括：
    获取至少一个依从性条件中每个依从性条件对应的第一患者数据；
    根据所述每个依从性条件对应的第一患者数据，训练原始的深度生成模型，得到预训练的深度生成模型。
  10. 根据权利要求9所述的方法，其中，所述根据所述每个依从性条件对应的第一患者数据，训练原始的深度生成模型，得到预训练的深度生成模型，包括：
    对所述每个依从性条件对应的第一患者数据进行归一化处理，得到第一处理数据；第一处理数据是指根据所述每个依从性条件对应的第一患者数据进行归一化处理后得到的数据；
    将所述第一处理数据作为原始的深度生成模型的输入数据，以对所述原始的深度生成模型进行训练，得到预训练的深度生成模型。
  11. 根据权利要求10所述的方法，其中，所述深度生成模型为变分自编码器，所述将所述第一处理数据作为原始的深度生成模型的输入数据，以对所述原始的深度生成模型进行训练，得到预训练的深度生成模型，包括：
    将所述第一处理数据作为原始的变分自编码器的输入数据，通过所述原始的变分自编码器包括的编码模块根据所述第一处理数据进行编码处理，得到均值和方差；
    对所述均值和方差服从的高斯分布进行采样，得到隐变量；
    将所述隐变量输入到所述原始的变分自编码器包括的解码模块，通过所述解码模块根据所述隐变量进行解码处理，得到所述每个依从性条件对应的第二患者数据；
    根据所述第一患者数据和所述第二患者数据，构建所述原始的变分自编码器的损失函数，利用所述损失函数训练所述原始的变分自编码器，得到预训练的变分自编码器。
  12. 根据权利要求11所述的方法，其中，所述通过所述解码模块根据所述隐变量进行解码处理，得到所述每个依从性条件对应的第二患者数据，包括：
    通过所述解码模块根据所述隐变量进行解码处理，得到第二处理数据；第二处理数据是指根据所述隐变量进行解码处理后得到的数据；
    对所述第二处理数据进行反归一化处理，得到所述每个依从性条件对应的第二患者数据。
  13. 根据权利要求8-12任一项所述的方法，其中，所述方法还包括：
    确定与所述目标患者样本匹配的目标心理护理策略；
    将所述目标心理护理策略发送至所述终端设备，以便所述终端设备展示所述目标心理护理策略。
  14. 一种样本生成装置，包括：
    通信模块，用于接收终端设备发送的患者样本生成请求，所述患者样本生成请求携带预置的依从性条件；
    处理模块，用于调用预训练的深度生成模型，以根据所述依从性条件生成所述依从性条件对应的目标患者样本，所述目标患者样本用于指导满足所述依从性条件的患者的就医行为；
    所述通信模块，还用于将所述目标患者样本发送至所述终端设备，以便所述终端设备展示所述目标患者样本。
  15. 一种计算机可读存储介质，所述计算机可读存储介质存储有计算机程序，所述计算机程序被处理器执行以实现以下方法：
    接收终端设备发送的患者样本生成请求，所述患者样本生成请求携带预置的依从性条件；
    调用预训练的深度生成模型，以根据所述依从性条件生成所述依从性条件对应的目标患者样本，所述目标患者样本用于指导满足所述依从性条件的患者的就医行为；
    将所述目标患者样本发送至所述终端设备，以便所述终端设备展示所述目标患者样本。
  16. 根据权利要求15所述的计算机可读存储介质，其中，所述计算机程序被处理器执行时还用于实现：
    获取至少一个依从性条件中每个依从性条件对应的第一患者数据；
    根据所述每个依从性条件对应的第一患者数据，训练原始的深度生成模型，得到预训练的深度生成模型。
  17. 根据权利要求16所述的计算机可读存储介质，其中，在根据所述每个依从性条件对应的第一患者数据，训练原始的深度生成模型，得到预训练的深度生成模型时，具体实现：
    对所述每个依从性条件对应的第一患者数据进行归一化处理，得到第一处理数据；第一处理数据是指根据所述每个依从性条件对应的第一患者数据进行归一化处理后得到的数据；
    将所述第一处理数据作为原始的深度生成模型的输入数据，以对所述原始的深度生成模型进行训练，得到预训练的深度生成模型。
  18. 根据权利要求17所述的计算机可读存储介质，其中，所述深度生成模型为变分自编码器，在将所述第一处理数据作为原始的深度生成模型的输入数据，以对所述原始的深度生成模型进行训练，得到预训练的深度生成模型时，具体实现：
    将所述第一处理数据作为原始的变分自编码器的输入数据，通过所述原始的变分自编码器包括的编码模块根据所述第一处理数据进行编码处理，得到均值和方差；
    对所述均值和方差服从的高斯分布进行采样，得到隐变量；
    将所述隐变量输入到所述原始的变分自编码器包括的解码模块，通过所述解码模块根据所述隐变量进行解码处理，得到所述每个依从性条件对应的第二患者数据；
    根据所述第一患者数据和所述第二患者数据，构建所述原始的变分自编码器的损失函数，利用所述损失函数训练所述原始的变分自编码器，得到预训练的变分自编码器。
  19. 根据权利要求18所述的计算机可读存储介质，其中，在通过所述解码模块根据所述隐变量进行解码处理，得到所述每个依从性条件对应的第二患者数据时，具体实现：
    通过所述解码模块根据所述隐变量进行解码处理，得到第二处理数据；第二处理数据是指根据所述隐变量进行解码处理后得到的数据；
    对所述第二处理数据进行反归一化处理，得到所述每个依从性条件对应的第二患者数据。
  20. 根据权利要求15-19任一项所述的计算机可读存储介质，其中，所述计算机程序被处理器执行时还用于实现：
    确定与所述目标患者样本匹配的目标心理护理策略；
    将所述目标心理护理策略发送至所述终端设备，以便所述终端设备展示所述目标心理护理策略。
PCT/CN2020/124610 2020-09-28 2020-10-29 一种样本生成方法、装置、服务器及存储介质 WO2021159740A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202011045854.3A CN111899844B (zh) 2020-09-28 2020-09-28 一种样本生成方法、装置、服务器及存储介质
CN202011045854.3 2020-09-28

Publications (1)

Publication Number Publication Date
WO2021159740A1 true WO2021159740A1 (zh) 2021-08-19

Family

ID=73224003

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/124610 WO2021159740A1 (zh) 2020-09-28 2020-10-29 一种样本生成方法、装置、服务器及存储介质

Country Status (2)

Country Link
CN (1) CN111899844B (zh)
WO (1) WO2021159740A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112927812B (zh) * 2021-03-30 2024-03-19 平安科技（深圳）有限公司 治疗策略的动态干预方法、装置、电子设备及存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107103201A (zh) * 2017-05-10 2017-08-29 北京大数医达科技有限公司 医疗导航路径的生成方法、装置及医疗路径导航方法
WO2019068073A1 (en) * 2017-09-29 2019-04-04 The Brigham And Women's Hospital, Inc. AUTOMATED EVALUATION OF HUMAN EMBRYOS
CN109698017A (zh) * 2018-12-12 2019-04-30 中电健康云科技有限公司 医疗病历数据生成方法及装置
CN109793645A (zh) * 2019-01-21 2019-05-24 徐州医科大学附属医院 一种辅助帕金森病人步态康复训练装置
CN110572696A (zh) * 2019-08-12 2019-12-13 浙江大学 一种变分自编码器与生成对抗网络结合的视频生成方法

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3367386A1 (en) * 2006-05-02 2018-08-29 Proteus Digital Health, Inc. Patient customized therapeutic regimens
CN110998744B (zh) * 2017-08-01 2024-04-05 西门子医疗有限公司 针对弥漫性和串联性病变中冠状动脉疾病的非侵入性评估和治疗指导
US20200253547A1 (en) * 2017-10-31 2020-08-13 Apple Inc. Monitoring System for Assessing Control of a Disease State
US10859657B2 (en) * 2018-05-31 2020-12-08 The Board Of Trustees Of The Leland Stanford Junior University MRI reconstruction using deep learning, generative adversarial network and acquisition signal model

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107103201A (zh) * 2017-05-10 2017-08-29 北京大数医达科技有限公司 医疗导航路径的生成方法、装置及医疗路径导航方法
WO2019068073A1 (en) * 2017-09-29 2019-04-04 The Brigham And Women's Hospital, Inc. AUTOMATED EVALUATION OF HUMAN EMBRYOS
CN109698017A (zh) * 2018-12-12 2019-04-30 中电健康云科技有限公司 医疗病历数据生成方法及装置
CN109793645A (zh) * 2019-01-21 2019-05-24 徐州医科大学附属医院 一种辅助帕金森病人步态康复训练装置
CN110572696A (zh) * 2019-08-12 2019-12-13 浙江大学 一种变分自编码器与生成对抗网络结合的视频生成方法

Also Published As

Publication number Publication date
CN111899844B (zh) 2021-11-23
CN111899844A (zh) 2020-11-06

Similar Documents

Publication Publication Date Title
US10014080B2 (en) Evidence based medical record
US20170293722A1 (en) Insurance Evaluation Engine
US10984024B2 (en) Automatic processing of ambiguously labeled data
WO2021180244A1 (zh) 一种疾病风险预测系统、方法、装置、设备及介质
WO2021120688A1 (zh) 医疗误诊检测方法、装置、电子设备及存储介质
WO2020228636A1 (zh) 训练方法和装置、对话处理方法和系统及介质
US20210022688A1 (en) Methods and systems for generating a diagnosis via a digital health application
WO2021159814A1 (zh) 文本数据的错误检测方法、装置、终端设备及存储介质
CN113707299A (zh) 基于问诊会话的辅助诊断方法、装置及计算机设备
WO2021159740A1 (zh) 一种样本生成方法、装置、服务器及存储介质
CN111128325A (zh) 医疗数据存储方法及装置、电子设备和存储介质
Yu et al. The development and impact of China's digital transformation in the medical industry
CN117573835A (zh) 一种大模型微调指令自动化生成方法、装置、设备及介质
CN111081325B (zh) 医疗数据处理方法与装置
CN112116994A (zh) 基于区块链的社会化医疗与信息服务平台
CN117198446A (zh) 一种基于医护对讲设备的查房报告自动生成方法
CN115718809B (zh) 知识图谱补全模型的训练方法及装置
Chua et al. Enhancing serious illness communication using artificial intelligence
CN112397195A (zh) 用于生成体格检查模型的方法、装置、电子设备和介质
WO2021151330A1 (zh) 用户分群方法、装置、设备和计算机可读存储介质
WO2021114626A1 (zh) 一种病历数据的质量检测方法和相关装置
CN113113124A (zh) 一种神经外科多患者的整合护理方法及系统
CN112394924A (zh) 用于生成提问模型的方法、装置、电子设备和介质
Lewis et al. Applying" big data" and business intelligence insights to improving clinical care for cancer
CN117174319B (zh) 一种基于知识图谱的脓毒症时序预测方法及系统

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: 20918743

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 24/05/2023)

122 Ep: pct application non-entry in european phase

Ref document number: 20918743

Country of ref document: EP

Kind code of ref document: A1