CN112599246A - Vital sign data processing method, system, device and computer readable medium - Google Patents
Vital sign data processing method, system, device and computer readable medium Download PDFInfo
- Publication number
- CN112599246A CN112599246A CN202110235788.4A CN202110235788A CN112599246A CN 112599246 A CN112599246 A CN 112599246A CN 202110235788 A CN202110235788 A CN 202110235788A CN 112599246 A CN112599246 A CN 112599246A
- Authority
- CN
- China
- Prior art keywords
- vital sign
- sign data
- data processing
- model
- learning
- Prior art date
- Legal status (The legal status 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 status listed.)
- Granted
Links
Images
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
- 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
- G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
-
- 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
- G16H50/50—ICT 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
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Public Health (AREA)
- Medical Informatics (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Pathology (AREA)
- Epidemiology (AREA)
- Primary Health Care (AREA)
- Databases & Information Systems (AREA)
- Biophysics (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The present invention relates to a data processing technology. The invention discloses a vital sign data processing method, a vital sign data processing system, a vital sign data processing device and a computer readable medium. The vital sign data processing method of the invention comprises the following steps: a. preprocessing vital sign data; b. learning random latent variables in the vital sign data and constraining the variables to be Gaussian distribution; c. learning the feature representation of vital sign data in each subset through two fully-connected network layers of a neural network according to a specific task, and performing fusion learning on the learned feature of each subtask and the random latent variable feature; d. training the model, verifying the model by using a verification set every 30 iterations, and reserving the optimal model; e. and testing the stored optimal model by using the test set to finally obtain the evaluation result of each task. The method can be used for accurately evaluating the capability and risk of the vital sign data of the healthy body, and the evaluation accuracy is greatly improved.
Description
Technical Field
The invention relates to a technology for analyzing and processing medical data by utilizing a neural network and a deep learning technology in the field, in particular to a vital sign data processing method, a vital sign data processing system, a vital sign data processing device and a computer readable medium.
Background
The vital sign data mainly comprise health indexes such as blood pressure, blood glucose concentration, blood oxygen size and the like, and according to the medical research, the current physical state of a healthy body and the possible risks of some abilities or the future can be directly or indirectly known through mining and analyzing some vital sign data. Therefore, accurate modeling analysis and evaluation of vital sign data is crucial in today's medical health field.
With the rapid development of artificial intelligence (especially deep learning), a series of new cross-fields have appeared, wherein the cross-fields of medical treatment and deep learning have been greatly successful in many medical applications. For example, in medical image analysis, Convolutional Neural Networks (CNNs) are used to assist in the detection of cancer. On the basis of physiotherapy services such as a clinical diagnosis auxiliary system and the like, the deep learning method is applied to scenes of early screening, diagnosis, rehabilitation, risk assessment and the like.
For modeling and analyzing vital sign data of a healthy body, a traditional machine learning or data mining method is often adopted in the past literature. For example, a Support Vector Machine (SVM) and a Decision Tree (DT) are combined to perform risk and capability assessment on healthy living bodies.
However, these conventional methods described above perform well in the face of a single classification task, but often do not perform satisfactorily if multiple classification tasks are to be processed simultaneously.
This is because conventional machine learning methods ignore interdependencies and associations between vital sign data features. A representative example is that many health evidences hide relationships, changes in one item of data may cause changes in other evidences, and different connections between different tasks may exist. For example, how to model these dependencies from vital sign data becomes a challenge to be solved.
Disclosure of Invention
The invention mainly aims to provide a vital sign data processing method, a vital sign data processing system, a vital sign data processing device and a computer readable medium, which can be used for simultaneously learning various tasks and automatically establishing dependency relationships among tasks, tasks and features and between features, so that the ability and risk of a user can be accurately classified and evaluated.
In order to achieve the above object, according to an aspect of an embodiment of the present invention, there is provided a vital sign data processing method, including the steps of:
a. performing linear normalization on the vital sign data, converting the vital sign data into a range of [0,1], and dividing the vital sign data into a full set and N subsets according to tasks, wherein N is an integer; corresponding to the specific task number;
b. introducing a variational automatic encoder to learn random latent variables in the vital sign data, and constraining the variables to be Gaussian distribution by maximizing an evidence lower bound;
c. learning the feature representation of the vital sign data in each subset through two fully-connected network layers of a neural network according to a specific task, and performing fusion learning on the learned feature of each subtask and the random latent variable feature in the step b to complete association learning between the feature of each subtask and the feature;
d. training the model by adopting a Gradient Decline (GD) and small Batch (Mini-Batch) training mode, verifying the model by using a verification set every 30 iterations, and reserving the optimal model;
e. and after the whole model is finished, testing the stored optimal model by using the test set, and finally obtaining the evaluation result of each task.
Further, each vital sign number includes 57 different vital signs.
Further, in the step a, the whole data corpus is divided into training sets according to the proportion of 6:2:2
Verification set
And test set
。
Further, in step c, a Dropout mechanism is added between the two fully-connected layers, and an activation function is added after each fully-connected layer, so that the target function is converted into a nonlinear mapping.
To achieve the above object, according to another aspect of an embodiment of the present invention, there is provided a vital sign data processing system including:
the preprocessing module is used for performing linear normalization on the vital sign data, converting the vital sign data into a range of [0,1], and dividing the vital sign data into a full set and N subsets according to tasks; n is an integer corresponding to the specific task number;
the variational automatic encoder module is used for learning random latent variables in the vital sign data and constraining the variables to be Gaussian distribution by maximizing the lower bound of evidence;
the fusion learning module is used for learning the feature representation of the vital sign data in each subset through two fully-connected network layers of the neural network according to the specific task, performing fusion learning on the learned feature of each subtask and the random latent variable feature in the step b, and completing the associated learning between the feature of each subtask and the feature;
the model training module is used for training the model by adopting a Gradient Decline (GD) and small Batch (Mini-Batch) training mode, verifying the model by using a verification set every 30 iterations and reserving the optimal model;
and the test module is used for testing the stored optimal model by using the test set after the whole model is finished, and finally obtaining the evaluation result of each task.
Further, each vital sign number includes 57 different vital signs.
Further, the preprocessing module divides the data corpus into training sets according to the proportion of 6:2:2
Verification set
And test set
。
Further, the fusion learning module adds a Dropout mechanism between the two fully-connected layers, adds an activation function after each fully-connected layer, and converts the target function into a nonlinear mapping.
In order to achieve the above object, according to another aspect of an embodiment of the present invention, there is provided a vital sign data processing apparatus including:
a processor, a memory;
the memory is used for storing executable instructions;
the processor is configured to perform the vital sign data processing method described above.
In order to achieve the above object, according to another aspect of specific embodiments of the present invention, there is provided a computer-readable storage medium, characterized by comprising a stored program, which when executed, performs the above vital sign data processing method.
According to the technical scheme of the invention and the technical scheme of further improvement in certain embodiments, the invention has the following beneficial effects:
based on multi-task learning, the method can be used for accurately evaluating the capability and risk of vital sign data of a healthy body, and compared with the traditional method, the method is greatly improved in evaluation accuracy.
The potential variable characteristics of the vital sign data are learned by using a variation automatic encoder, and the dependency between the vital sign data is captured under the condition that the potential variables are subjected to Gaussian distribution.
The potential variable features contain all vital sign information and have rich global feature information, and the defect of insufficient sub-task features can be overcome by adding the potential variable features into each specific sub-task for joint learning, so that the overall performance of the model is improved.
The analysis and evaluation of a plurality of different tasks can be completed through one sign data input. The association between tasks is completed through feature interaction between different tasks, and the association between the tasks and features is completed through adding random latent variable features, so that the performance of the whole evaluation model is further improved.
The invention is further described with reference to the following figures and detailed description. Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this application, illustrate embodiments of the invention and, together with the description, serve to explain the invention and not to limit the invention. In the drawings:
FIG. 1 is a flow chart of a vital sign data processing method according to the present invention;
FIG. 2 is a schematic diagram of a vital signs data processing system according to the present invention;
fig. 3 is a schematic structural diagram of the vital sign data processing device of the present invention.
Wherein:
210 is a preprocessing module;
220 is a variational automatic encoder module;
230 is a fusion learning module;
240 is a model training module;
250 is a test module;
310 is a processor;
320 is a memory;
330 is a bus.
Detailed Description
It should be noted that the specific embodiments, examples and features thereof may be combined with each other in the present application without conflict. The present invention will now be described in detail with reference to the attached figures in conjunction with the following.
In order to make the technical solutions of the present invention better understood, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, but not all embodiments. All other embodiments and examples obtained by a person skilled in the art without any inventive step should fall within the protection scope of the present invention.
The terms "comprising," "including," "having," and any variations thereof in this specification and claims and in any related parts thereof, are intended to cover non-exclusive inclusions.
Fig. 1 is a flow chart of a vital sign data processing method according to the present invention. As shown in fig. 1, the vital sign data processing method of the present invention includes the steps of:
s110, performing linear normalization on the vital sign data, converting the vital sign data into a range of [0,1], and dividing the vital sign data into a full set and N subsets according to tasks, wherein N is an integer; corresponding to a specific number of tasks.
Each vital sign count includes 57 different vital signs.
For all vital sign data, different characteristics have dimensional and magnitude influences, so that linear normalization (Min-Max Scaling) processing is required to convert the original vital sign data into a range of [0,1 ].
The training speed of the model can be accelerated by linear normalization processing, and the performance of the model can be improved.
In addition, for each complete piece of training data
Which contains a plurality of different vital sign data. Thus, for different evaluation tasks, the training data also needs to be divided into N subsets according to the different tasks, i.e.
= 
And performing feature learning on different subtask learning.
S120, a variational automatic encoder is introduced to learn random Latent Variables (Stochastic tension Variables SLV) in the vital sign data, and the Variables are constrained to be Gaussian distribution by maximizing the lower bound of evidence.
The Variational Auto Encoder (Variational Auto Encoder VAE) can model the uncertainty between features from the perspective of distribution learning, and can model the interdependency between different subtasks and the association between each vital sign feature, i.e. the task-to-task, feature-to-feature associations.
A variational autoencoder consists of two main networks, an inference network (encoder) and a generation network (decoder).
Inferring mean and variance of network learning data samples to generate random latent variables
. It is often necessary to constrain this random latent variable
Obeying a gaussian distribution. This is because most data in the real world obey a gaussian distribution, and theoretically any other complex distribution can be fitted from the gaussian distribution.
Random latent variables can be assigned by minimizing KL divergence
The distribution of (a) is to be increasingly close to gaussian:
wherein the parameters
And
respectively representing inference networks
And generating a network
Is determined by the parameters of (a) and (b),
a random latent variable is represented that is,
each vital sign training data is represented.
Obtaining random latent variables obeying a Gaussian distribution
Thereafter, downstream tasks are performed in two different directions:
the first direction is from a random latent variable
The mean and variance of (a) are Resampled (RT), and the resulting samples are put into a decoder for source vital sign data reconstruction:
the second direction is to combine random latent variables
And as an intermediate variable, performing fusion learning by combining the characteristics of each subtask, and further establishing the association among the vital sign characteristics.
S130, learning the feature representation of the vital sign data in each subset through two fully-connected network layers of the neural network according to the specific task, performing fusion learning on the learned feature of each subtask and the random latent variable feature in the step b, and finishing the associated learning between the feature of each subtask and the feature.
By performing independent feature extraction on each subtask, good features which are irrelevant to each different task are learned as far as possible.
And learning the characteristics of different subtasks by utilizing the fully-connected network layer.
To reduce the degree of model overfitting and enhance the generalization capability of the model, a Dropout mechanism is added between the two fully connected network layers. And an activation function (ReLU) is added behind each fully-connected network layer to convert the target function into nonlinear mapping, so that the performance of the whole model is improved:
wherein the content of the first and second substances,
represents the input of the ith sub-task,
and
representing learnable parameters and bias terms in a fully connected network layer,
features learned for the ith subtask.
After obtaining the features of each subtask, the random latent variables learned in S120 need to be learned
And combining the vital sign data and the feature fusion data together to carry out feature fusion learning, wherein the aim is to establish the relevance between different features among the vital sign data.
After obtaining the fusion features we adopt
A final evaluation classification is performed:
and S140, training the model by adopting a Gradient Decline (GD) and small Batch (Mini-Batch) training mode, verifying the model by using a verification set every 30 iterations, and reserving the optimal model.
Training vital sign data evaluation model according to gradient descent method
A cross-entropy loss function commonly used for classification problems can be selected:
wherein the content of the first and second substances,
representation evaluation model
All of the learning parameters in (1) are,
representing the final evaluation classification result.
In order to obtain the optimal evaluation model, the model is verified by a verification set after each iterative training is finished
The verification is carried out in such a way that,and stores the primary model with the best evaluation result.
After all vital sign data are trained, the stored optimal model is used as a test set
Tests were performed to verify the final model.
Therefore, the vital sign analysis and evaluation method based on multi-task learning is realized, and the processing of vital sign data is completed.
Fig. 2 is a schematic structural diagram of the vital sign data processing system of the present invention. As shown in fig. 2, the vital sign data processing system of the present invention includes:
the preprocessing module 210 is configured to perform linear normalization on the vital sign data, convert the vital sign data into a range of [0,1], and divide the vital sign data into a full set and N subsets according to tasks; n is an integer corresponding to the specific number of tasks.
Each vital sign number here includes 57 different vital signs.
The preprocessing module 210 divides the data corpus into training sets by a ratio of 6:2:2
Verification set
And test set
。
A variational autoencoder module 220 for learning random latent variables in the vital sign data and constraining the variables to a gaussian distribution by maximizing the lower bound of evidence.
And a fusion learning module 230, configured to learn, according to the specific task, feature representations of the vital sign data in the subsets through two fully-connected layers of the neural network, perform fusion learning on the learned features of each subtask and the random latent variable features in step b, and complete association learning between the features of each subtask.
The fusion learning module 230 adds Dropout mechanism between two fully connected layers and adds an activation function after each fully connected layer to convert the target function into a non-linear mapping.
And the model training module 240 is configured to train the model by using Gradient Descent (GD) and small Batch (Mini-Batch) training modes, verify the model by using a verification set every 30 iterations, and retain the optimal model.
And the test module 250 is used for testing the stored optimal model by using the test set after the whole model is completed, and finally obtaining the evaluation result of each task.
Fig. 3 is a schematic structural diagram of the vital sign data processing device of the present invention. As shown in fig. 3, the vital sign data processing apparatus of the present invention includes a processor 310 and a memory 320.
The memory 320 is used to store executable instructions.
The processor 310 is configured to perform the vital signs data processing method described above.
The processor 310 includes a Central Processing Unit (CPU), a Digital Signal Processor (DSP) or an Application Specific Integrated Circuit (ASIC), or one or more Integrated circuits that may be configured to implement the vital sign data processing methods described above.
The processor 310 reads and executes the computer program instructions stored in the memory 320 to implement the vital sign data processing method described above.
In fig. 3, bus 330 is used to transport program instructions and data exchanges. By way of example, and not limitation, bus 330 may include an Accelerated Graphics Port (AGP) bus or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hypertransport (HT) interconnect, a memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, or other suitable bus, or a combination of two or more of these. Bus 330 may include one or more buses, where appropriate. Although this disclosure describes and illustrates a particular bus, this disclosure contemplates any suitable bus or interconnect.
The computer-readable storage medium of the present invention includes a stored program, and the program executes the above vital sign data processing method.
Claims (10)
1. The vital sign data processing method is characterized by comprising the following steps:
a. performing linear normalization on the vital sign data, converting the vital sign data into a range of [0,1], and dividing the vital sign data into a full set and N subsets according to tasks, wherein N is an integer; corresponding to the specific task number;
b. introducing a variational automatic encoder to learn random latent variables in the vital sign data, and constraining the variables to be Gaussian distribution by maximizing an evidence lower bound;
c. learning the feature representation of the vital sign data in each subset through two fully-connected network layers of a neural network according to a specific task, and performing fusion learning on the learned feature of each subtask and the random latent variable feature in the step b to complete association learning between the feature of each subtask and the feature;
d. training the model by adopting a Gradient Descent Gradient Descent, GD and small Batch Mini-Batch training mode, verifying the model by using a verification set every 30 iterations, and reserving the optimal model;
e. and after the whole model is finished, testing the stored optimal model by using the test set, and finally obtaining the evaluation result of each task.
2. The vital sign data processing method of claim 1, wherein each vital sign number includes 57 different vital signs.
3. The method for processing vital sign data according to claim 1, wherein in step a, the whole data corpus is divided into training sets according to a ratio of 6:2:2
Verification set
And test set
。
4. The vital sign data processing method of claim 1, wherein in step c, a Dropout mechanism is added between two fully connected layers, and an activation function is added after each fully connected layer to convert the objective function into a non-linear mapping.
5. Vital sign data processing system, comprising:
the preprocessing module is used for performing linear normalization on the vital sign data, converting the vital sign data into a range of [0,1], and dividing the vital sign data into a full set and N subsets according to tasks; n is an integer corresponding to the specific task number;
the variational automatic encoder module is used for learning random latent variables in the vital sign data and constraining the variables to be Gaussian distribution by maximizing the lower bound of evidence;
the fusion learning module is used for learning the feature representation of the vital sign data in each subset through two full-connection layers of the neural network according to a specific task, performing fusion learning on the learned feature of each subtask and the random latent variable feature in the step b, and completing the associated learning between the feature of each subtask and the feature;
the model training module is used for training the model by adopting a training mode of Gradient Descent, GD and a small Batch Mini-Batch, verifying the model by using a verification set every 30 iterations, and reserving the optimal model;
and the test module is used for testing the stored optimal model by using the test set after the whole model is finished, and finally obtaining the evaluation result of each task.
6. The vital signs data processing system of claim 5, wherein each vital signs number comprises 57 different vital signs.
7. The vital sign data processing system of claim 5, wherein the pre-processing module divides the corpus of data into training sets by a ratio of 6:2:2
Verification set
And test set
。
8. The vital signs data processing system of claim 5, wherein the fusion learning module incorporates a Dropout mechanism between two fully connected layers and an activation function after each fully connected layer to convert the objective function to a non-linear mapping.
9. Vital sign data processing apparatus, comprising:
a processor, a memory;
the memory is used for storing executable instructions;
the processor is configured to perform the vital sign data processing method of any one of claims 1 to 4.
10. Computer-readable storage medium, characterized in that it comprises a stored program which when running executes the vital sign data processing method of any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110235788.4A CN112599246B (en) | 2021-03-03 | 2021-03-03 | Vital sign data processing method, system, device and computer readable medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110235788.4A CN112599246B (en) | 2021-03-03 | 2021-03-03 | Vital sign data processing method, system, device and computer readable medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112599246A true CN112599246A (en) | 2021-04-02 |
| CN112599246B CN112599246B (en) | 2021-06-29 |
Family
ID=75210352
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110235788.4A Active CN112599246B (en) | 2021-03-03 | 2021-03-03 | Vital sign data processing method, system, device and computer readable medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112599246B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113393934A (en) * | 2021-06-07 | 2021-09-14 | 义金(杭州)健康科技有限公司 | Health trend estimation method and prediction system based on vital sign big data |
| CN117850601A (en) * | 2024-03-08 | 2024-04-09 | 南昌大学第二附属医院 | System and method for automatically detecting vital signs of handheld PDA |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108389125A (en) * | 2018-02-27 | 2018-08-10 | 挖财网络技术有限公司 | The overdue Risk Forecast Method and device of credit applications |
| CN108897947A (en) * | 2018-06-27 | 2018-11-27 | 西安交通大学 | A kind of equipment degeneration multi-source data fusion method based on improvement variation autocoding |
| CN109598939A (en) * | 2018-12-24 | 2019-04-09 | 中国科学院地理科学与资源研究所 | A kind of prediction of short-term traffic volume method based on multitask multiple view learning model |
| CN110363283A (en) * | 2019-06-06 | 2019-10-22 | 哈尔滨工业大学（深圳） | User property prediction technique and relevant apparatus based on deep learning |
| CN111899882A (en) * | 2020-08-07 | 2020-11-06 | 北京科技大学 | Method and system for predicting cancer |
| US20200364505A1 (en) * | 2018-09-27 | 2020-11-19 | Deepmind Technologies Limited | Committed information rate variational autoencoders |
| CN112241783A (en) * | 2019-07-17 | 2021-01-19 | 罗伯特·博世有限公司 | Machine-learnable system with conditional normalized flow |
-
2021
- 2021-03-03 CN CN202110235788.4A patent/CN112599246B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108389125A (en) * | 2018-02-27 | 2018-08-10 | 挖财网络技术有限公司 | The overdue Risk Forecast Method and device of credit applications |
| CN108897947A (en) * | 2018-06-27 | 2018-11-27 | 西安交通大学 | A kind of equipment degeneration multi-source data fusion method based on improvement variation autocoding |
| US20200364505A1 (en) * | 2018-09-27 | 2020-11-19 | Deepmind Technologies Limited | Committed information rate variational autoencoders |
| CN109598939A (en) * | 2018-12-24 | 2019-04-09 | 中国科学院地理科学与资源研究所 | A kind of prediction of short-term traffic volume method based on multitask multiple view learning model |
| CN110363283A (en) * | 2019-06-06 | 2019-10-22 | 哈尔滨工业大学（深圳） | User property prediction technique and relevant apparatus based on deep learning |
| CN112241783A (en) * | 2019-07-17 | 2021-01-19 | 罗伯特·博世有限公司 | Machine-learnable system with conditional normalized flow |
| CN111899882A (en) * | 2020-08-07 | 2020-11-06 | 北京科技大学 | Method and system for predicting cancer |
Non-Patent Citations (2)
| Title |
|---|
| ÇAĞLAR HIZLI等: "Joint Source Separation and Classification Using Variational Autoencoders", 《2020 28TH SIGNAL PROCESSING AND COMMUNICATIONS APPLICATIONS CONFERENCE (SIU)》 * |
| 刘霄宇等: "基于主动学习的机器学习算法研究进展", 《现代计算机》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113393934A (en) * | 2021-06-07 | 2021-09-14 | 义金(杭州)健康科技有限公司 | Health trend estimation method and prediction system based on vital sign big data |
| CN113393934B (en) * | 2021-06-07 | 2022-07-12 | 义金(杭州)健康科技有限公司 | Health trend estimation method and prediction system based on vital sign big data |
| CN117850601A (en) * | 2024-03-08 | 2024-04-09 | 南昌大学第二附属医院 | System and method for automatically detecting vital signs of handheld PDA |
| CN117850601B (en) * | 2024-03-08 | 2024-05-14 | 南昌大学第二附属医院 | System and method for automatically detecting vital signs of handheld PDA |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112599246B (en) | 2021-06-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112599246B (en) | Vital sign data processing method, system, device and computer readable medium | |
| CN110796199B (en) | Image processing method and device and electronic medical equipment | |
| CN111444873B (en) | Method and device for detecting authenticity of person in video, electronic equipment and storage medium | |
| CN110232123B (en) | Text emotion analysis method and device, computing device and readable medium | |
| CN111738269B (en) | Model training method, image processing device, model training apparatus, and storage medium | |
| CN109948680B (en) | Classification method and system for medical record data | |
| CN107945210B (en) | Target tracking method based on deep learning and environment self-adaption | |
| CN111539480A (en) | Multi-class medical image identification method and equipment | |
| CN110363830B (en) | Element image generation method, device and system | |
| CN112668486A (en) | Method, device and carrier for identifying facial expressions of pre-activated residual depth separable convolutional network | |
| Huber et al. | Benchmarking perturbation-based saliency maps for explaining Atari agents | |
| CN113592842A (en) | Sample serum quality identification method and identification device based on deep learning | |
| CN111753859B (en) | Sample generation method, device and equipment | |
| CN117036941A (en) | Building change detection method and system based on twin Unet model | |
| CN113780394B (en) | Training method, device and equipment for strong classifier model | |
| Kazaj et al. | U-net-based models for skin lesion segmentation: More attention and augmentation | |
| CN112329735B (en) | Training method of face recognition model and online education system | |
| EP3975071A1 (en) | Identifying and quantifying confounding bias based on expert knowledge | |
| CN115240843A (en) | Fairness prediction system based on structure causal model | |
| CN114049303A (en) | Progressive bone age assessment method based on multi-granularity feature fusion | |
| CN113706546A (en) | Medical image segmentation method and device based on lightweight twin network | |
| Suyal et al. | An Agile Review of Machine Learning Technique | |
| CN116030526B (en) | Emotion recognition method, system and storage medium based on multitask deep learning | |
| CN116992979B (en) | Deep learning-based cognitive training task pushing method and system | |
| CN112712130B (en) | Visual understanding model training method and device, computer equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right |
Denomination of invention: Vital sign data processing method, system, device and computer-readable medium Effective date of registration: 20220429 Granted publication date: 20210629 Pledgee: Bank of Chengdu science and technology branch of Limited by Share Ltd. Pledgor: SICHUAN HWADEE INFORMATION TECHNOLOGY CO.,LTD. Registration number: Y2022980004936 |
|
| PE01 | Entry into force of the registration of the contract for pledge of patent right |