CN109363632A - The deciphering method of pulse profile data and the solution read apparatus of pulse profile data - Google Patents
The deciphering method of pulse profile data and the solution read apparatus of pulse profile data Download PDFInfo
- Publication number
- CN109363632A CN109363632A CN201811122932.8A CN201811122932A CN109363632A CN 109363632 A CN109363632 A CN 109363632A CN 201811122932 A CN201811122932 A CN 201811122932A CN 109363632 A CN109363632 A CN 109363632A
- Authority
- CN
- China
- Prior art keywords
- pulse
- profile data
- pulse profile
- network
- data
- 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.)
- Pending
Links
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/48—Other medical applications
- A61B5/4854—Diagnosis based on concepts of traditional oriental medicine
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7246—Details of waveform analysis using correlation, e.g. template matching or determination of similarity
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
Abstract
The present invention provides the solution read apparatus of a kind of deciphering method of pulse profile data and pulse profile data, wherein, the deciphering method of pulse profile data, the pulse profile data of the pulse condition is obtained by pulse-taking instrument, the method of the pulse condition diagnosis includes: acquisition training data, acquires the pulse profile data, and the pulse-taking instrument sample frequency is 100Hz, the label of the pulse profile data is corresponding with constitution, establishes indefinite long period, the pulse profile data as unit of per second and constitution label pair;Network and training depth network model, the pulse profile data that network inputs are each cycle 60 seconds are established, network output is the result of pulse condition;Model application carries out physical fitness diagnosis using the trained depth network model.According to the technical solution of the present invention, the judgement of multiple dimensioned pulse profile data is realized, a plurality of pulse profile data can correspond to a variety of pulse conditions as a result, it is possible to achieve single people, the identification of a variety of pulse conditions.
Description
Technical field
The present invention relates to field of medical technology, deciphering method and a kind of pulse condition in particular to a kind of pulse profile data
The solution read apparatus of data.
Background technique
Wang Qi tcm constitution rule of Nine " --- including gentle matter, deficiency of vital energy matter, deficiency of yang matter, deficiency of Yin matter, phlegm wet matter, damp and hot matter,
Hemostasis matter, obstruction of the circulation of vital energy matter, special official report matter.Forefathers have carried out Primary Study in terms of pulse condition and the association study of 9 kinds of constitutions.Shen Xiao etc.
Investigated arteries and veins position, 3 kinds of arteries and veins power, pulse frequency pulse condition informations are associated with constitution, find different constitutions in terms of sex composition difference compared with
Greatly, arteries and veins power can be used as the important evidence for distinguishing damp and hot matter and deficiency of yang matter.Wang Yingchun etc. is input, building with pulse-taking instrument measurement result
The BP neural networks of 9 kinds of constitutions.He Yan etc. has carried out discriminant analysis to the association of arteries and veins figure information and constitution, but effect is not enough managed
Think.
There are the problem of mainly have 3 aspect: first is that the sample size for establishing database is few, it is difficult to reflect general characteristic;Second is that
Acquisition method is often that single position acquires information of pulse examination, can not embody the diagnosis by feeling the pulse characteristic of three nine marquis of Chinese medicine;Third is that analysis method
Shortcomings.
Summary of the invention
The present invention is directed to solve at least one of the technical problems existing in the prior art or related technologies.
For this purpose, it is an object of the present invention to provide a kind of deciphering methods of pulse profile data.
It is another object of the present invention to provide a kind of solution read apparatus of pulse profile data.
In view of this, the technical solution of the first aspect of the present invention provides a kind of method of pulse condition diagnosis, by pulse-taking instrument
The pulse profile data of the pulse condition is obtained, the deciphering method of the pulse profile data includes: acquisition training data, acquires the pulse condition number
Be 100Hz according to, the pulse-taking instrument sample frequency, the label of the pulse profile data is corresponding with constitution, establish indefinite long period, with
Pulse profile data and constitution label pair per second for unit;Network and training depth network model are established, network inputs are each cycle
60 seconds pulse profile datas, network output are the result of pulse condition;Model application, using the trained depth network model
Carry out physical fitness diagnosis.
Further, in the acquisition training data, the length of the sample of each pulse profile data is 100Hz.
The technical solution of the second aspect of the present invention provides a kind of solution read apparatus of pulse profile data, comprising: obtains training
Data cell, for acquiring the pulse profile data, the pulse-taking instrument sample frequency is 100Hz, the label of the pulse profile data with
Constitution is corresponding, establishes indefinite long period, the pulse profile data as unit of per second and constitution label pair;Training depth network model
Unit, for pulse profile data input network to be trained, the pulse profile data that network inputs are each cycle 60 seconds, net
Network output is the result of pulse condition;Model applying unit, for carrying out physical fitness diagnosis using the trained depth network model.
One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:
The possibility of database is established with input great amount of samples, thus further improve reacts the accurate of general characteristic
Property, the judgement of multiple dimensioned pulse profile data is realized, a plurality of pulse profile data can correspond to a variety of pulse conditions as a result, it is possible to achieve single
People, the identification of a variety of pulse conditions.
Additional aspect and advantage of the invention will be set forth in part in the description, and will partially become from the following description
Obviously, or practice through the invention is recognized.
Detailed description of the invention
Above-mentioned and/or additional aspect of the invention and advantage will become from the description of the embodiment in conjunction with the following figures
Obviously and it is readily appreciated that, in which:
Fig. 1 shows the network diagram according to an embodiment of the invention with residual error structure.
Specific embodiment
It is with reference to the accompanying drawing and specific real in order to be more clearly understood that aforementioned aspect of the present invention, feature and advantage
Applying mode, the present invention is further described in detail.It should be noted that in the absence of conflict, the implementation of the application
Feature in example and embodiment can be combined with each other.
In the following description, numerous specific details are set forth in order to facilitate a full understanding of the present invention, still, the present invention may be used also
To be implemented using other than the one described here other modes, therefore, protection scope of the present invention is not by described below
Specific embodiment limitation.
Embodiment 1:
The deciphering method of the pulse profile data of embodiment according to the present invention, comprising:
Step A1 extracts pulse profile data and carries out the pretreatment of training data:
The vein data for collecting 20000 famous persons and expert are to the judgement result of individual physique.Each of them is a company
The data in continuous sampling period, the pulse condition for including are denoted as x1... xkCorresponding individual physique classification is m1... mn, wherein classification mi
For one of gentle matter, deficiency of vital energy matter, deficiency of yang matter, deficiency of Yin matter, phlegm wet matter, damp and hot matter, hemostasis matter, obstruction of the circulation of vital energy matter, special official report matter.
In training data pretreatment, we are per second to cut the pulse condition sequence of random length according to a unit, and
A data pair are formed with corresponding constitution classification, the frequency of sampling is 100Hz, then the dimension of data pair is (100,1).
Step A2, the network structure that pulse condition classification distinguishes.
We use convolutional neural networks (CNN) to network structure, and multiple dimensioned arteries and veins can be extracted by being mainly in view of convolution kernel
Picture signals information.There are the variations of a variety of scales for pulse signal feature, we can be obtained using CNN according to the convolutional coding structure of deep layer
Obtain receptive field bigger in signal time domain space, the characteristic information of effective signal acquisition.Convolutional neural networks are by convolutional layer
It is constituted with pond layer, convolutional layer is noteworthy characterized by the parameter that depth network structure is reduced using shared weight, reduces training
Complexity.Simultaneously in order to guarantee the loss for reducing signal to the greatest extent during Internet communication, uses He Kaiming in 2015 and propose
Residual error structure, establish 16 layers of network structure.
Data input:
For the data structure of pulse condition with image difference, the pulse profile data of single location is one-dimensional tensor.Chinese medicine pulse diagnosis
According to the comprehensive diagnos of the pulse condition information at 6 positions as a result, the differentiation network inputs of pulse condition type are the 1 dimension tensor in 6 channels.?
It is the data pair of (96,1) that we obtained in the processing of training set, which is according to the fixed length that one second is input, is denoted as respectively (a, b),
Therefore it is (6,100,1) that the single sample of input, which is dimension,.
Network structure:
We are using the network with residual error structure, as shown in Figure 1, residual error structure is deep neural network jump connection
A kind of structure of convolutional layer, depth network are equivalent to an identical transformation.
Residual error network is finely adjusted, each layer increases a regularization layer, the layer be accomplished that by batch sample into
Row regularization variation, regularization layer help to improve the Generalization Capability of depth model, reduce over-fitting.
Wherein activation primitive form is as follows:
Whole convolution are finally attached by full articulamentum, then by soft max function output category result, are denoted as
y。
softmax(y)iexp(yi)/∑jexp(yj)
Training process:
In the training of the network, loss function uses following form:
Wherein, n is the batch size of training, and p (*) is output and the consistent probability of label.Training is using under stochastic gradient
The method of drop successively trains each layer parameter using backpropagation, shown in the network structure design table 1 of each layer.
Table 1
Wherein Conv indicates that convolutional layer, BN indicate that regularization layer, ReLU indicate hidden layer, what Residual block was indicated
It is residual error structure sheaf.
Step A3, constitution judgement: retaining weight according to training pattern, and realizes the constitution judgement of people.Acquire the pulse condition of people
Continuous pulse profile data is split as single input data in seconds by data.Before being passed through using the network that training obtains
The pulse condition result of people is obtained to communication process.We realize the judgement of multiple dimensioned pulse profile data to the fractionation of pulse profile data, more
Pulse profile data can correspond to a variety of pulse conditions as a result, it is possible to achieve single people, the identification of a variety of pulse conditions.In order to avoid splitting part
Feature be destroyed, we using 0.5 times coincidence sample, i.e., sequentially in time realize 0.5s pulse profile data coincidence adopt
Sample obtains a series of sampled data.The available multiple pulse condition labels of network obtained using training, remove duplicate knot
Fruit, final pulse condition set are the constitution judging result of observation object.Example: the pulse condition sample of patient amounts to 120s, we obtain
It is the data of 11520 frames to overall length, 230 sample datas can be obtained according to repeated sampling, corresponded to by our network
Label, it is assumed that all Qi deficiency physiques, then last judging result be Qi deficiency physique.It is exported if label difference whole
Constitution result.
Embodiment 2:
The solution read apparatus of the pulse profile data of embodiment according to the present invention, comprising: training data unit is obtained, for acquiring
The pulse profile data, the pulse-taking instrument sample frequency are 100Hz, and the label of the pulse profile data is corresponding with constitution, are established not
Fixed length period, the pulse profile data as unit of per second and constitution label pair;Training depth network model unit, is used for the arteries and veins
Image data input network is trained, the pulse profile data that network inputs are each cycle 60 seconds, and network output is the knot of pulse condition
Fruit;Model applying unit, for carrying out physical fitness diagnosis using the trained depth network model.
Technical solution of the present invention has been described with reference to the drawings above, according to the technical solution of the present invention, by training data,
It establishes model and applies, there is input great amount of samples to establish the possibility of database, thus further improve reacts general characteristic
Accuracy, realize the judgement of multiple dimensioned pulse profile data, a plurality of pulse profile data can correspond to a variety of pulse conditions as a result, it is possible to achieve
Single people, the identification of a variety of pulse conditions.
In the description of this specification, the description of term " one embodiment ", " some embodiments ", " specific embodiment " etc.
Mean that particular features, structures, materials, or characteristics described in conjunction with this embodiment or example are contained at least one reality of the invention
It applies in example or example.In the present specification, schematic expression of the above terms are not necessarily referring to identical embodiment or reality
Example.Moreover, description particular features, structures, materials, or characteristics can in any one or more of the embodiments or examples with
Suitable mode combines.
The foregoing is only a preferred embodiment of the present invention, is not intended to restrict the invention, for the skill of this field
For art personnel, the invention may be variously modified and varied.All within the spirits and principles of the present invention, made any to repair
Change, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.
Claims (3)
1. a kind of deciphering method of pulse profile data, the pulse profile data of the pulse condition is obtained by pulse-taking instrument, which is characterized in that the arteries and veins
The deciphering method of image data includes:
Training data is obtained, the pulse profile data is acquired, the pulse-taking instrument sample frequency is 100Hz, the mark of the pulse profile data
Label are corresponding with constitution, establish indefinite long period, the pulse profile data as unit of per second and constitution label pair;
Network and training depth network model are established, pulse profile data input network is trained, network inputs are weekly
The phase 60 seconds pulse profile datas, network output are the result of pulse condition;
Model application carries out physical fitness diagnosis using the trained depth network model.
2. the deciphering method of pulse profile data according to claim 1, which is characterized in that in the acquisition training data, often
The length of the sample of a pulse profile data is 100Hz.
3. a kind of solution read apparatus of pulse profile data characterized by comprising
Training data unit is obtained, for acquiring the pulse profile data, the pulse-taking instrument sample frequency is 100Hz, the pulse condition
The label of data is corresponding with constitution, establishes indefinite long period, the pulse profile data as unit of per second and constitution label pair;
Training depth network model unit, for pulse profile data input network to be trained, network inputs are each cycle
60 seconds pulse profile datas, network output are the result of pulse condition;
Model applying unit, for carrying out physical fitness diagnosis using the trained depth network model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811122932.8A CN109363632A (en) | 2018-09-26 | 2018-09-26 | The deciphering method of pulse profile data and the solution read apparatus of pulse profile data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811122932.8A CN109363632A (en) | 2018-09-26 | 2018-09-26 | The deciphering method of pulse profile data and the solution read apparatus of pulse profile data |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109363632A true CN109363632A (en) | 2019-02-22 |
Family
ID=65401875
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811122932.8A Pending CN109363632A (en) | 2018-09-26 | 2018-09-26 | The deciphering method of pulse profile data and the solution read apparatus of pulse profile data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109363632A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110752032A (en) * | 2019-12-02 | 2020-02-04 | 山东浪潮人工智能研究院有限公司 | Traditional Chinese medicine diagnosis method based on convolutional neural network and laser vibration measurement |
CN113076878A (en) * | 2021-04-02 | 2021-07-06 | 郑州大学 | Physique identification method based on attention mechanism convolution network structure |
CN113229798A (en) * | 2021-05-18 | 2021-08-10 | 平安科技(深圳)有限公司 | Model migration training method and device, computer equipment and readable storage medium |
CN115374125A (en) * | 2022-09-01 | 2022-11-22 | 无锡市华焯光电科技有限公司 | Pulse condition diagnosis and classification method, database construction method, device and storage medium |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20110098286A (en) * | 2010-02-26 | 2011-09-01 | 신라대학교 산학협력단 | Self health diagnosis system of oriental medicine using fuzzy inference method |
KR20130024535A (en) * | 2011-08-31 | 2013-03-08 | 한국 한의학 연구원 | Apparatus and method for determining health state using pulse wave |
CN103886173A (en) * | 2012-12-19 | 2014-06-25 | 财团法人工业技术研究院 | Pulse wave and physical health risk assessment system and method |
CN106709448A (en) * | 2016-12-21 | 2017-05-24 | 华南理工大学 | Face convolutional neural network-based traditional Chinese medicine constitution recognition method |
CN108022213A (en) * | 2017-11-29 | 2018-05-11 | 天津大学 | Video super-resolution algorithm for reconstructing based on generation confrontation network |
CN108280827A (en) * | 2018-01-24 | 2018-07-13 | 北京红云视界技术有限公司 | Coronary artery pathological changes automatic testing method, system and equipment based on deep learning |
CN108319977A (en) * | 2018-01-30 | 2018-07-24 | 浙江大学 | Cervical biopsy area recognizing method based on the multi-modal network of channel information and device |
CN108309257A (en) * | 2018-03-29 | 2018-07-24 | 深圳市易成自动驾驶技术有限公司 | Diagnosis by feeling the pulse method, apparatus based on deep learning and computer storage media |
-
2018
- 2018-09-26 CN CN201811122932.8A patent/CN109363632A/en active Pending
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20110098286A (en) * | 2010-02-26 | 2011-09-01 | 신라대학교 산학협력단 | Self health diagnosis system of oriental medicine using fuzzy inference method |
KR20130024535A (en) * | 2011-08-31 | 2013-03-08 | 한국 한의학 연구원 | Apparatus and method for determining health state using pulse wave |
CN103886173A (en) * | 2012-12-19 | 2014-06-25 | 财团法人工业技术研究院 | Pulse wave and physical health risk assessment system and method |
CN106709448A (en) * | 2016-12-21 | 2017-05-24 | 华南理工大学 | Face convolutional neural network-based traditional Chinese medicine constitution recognition method |
CN108022213A (en) * | 2017-11-29 | 2018-05-11 | 天津大学 | Video super-resolution algorithm for reconstructing based on generation confrontation network |
CN108280827A (en) * | 2018-01-24 | 2018-07-13 | 北京红云视界技术有限公司 | Coronary artery pathological changes automatic testing method, system and equipment based on deep learning |
CN108319977A (en) * | 2018-01-30 | 2018-07-24 | 浙江大学 | Cervical biopsy area recognizing method based on the multi-modal network of channel information and device |
CN108309257A (en) * | 2018-03-29 | 2018-07-24 | 深圳市易成自动驾驶技术有限公司 | Diagnosis by feeling the pulse method, apparatus based on deep learning and computer storage media |
Non-Patent Citations (1)
Title |
---|
LI H , XU B , WANG N , ET AL: "Deep Convolutional Neural Networks for Classifying Body Constitution", 《INTERNATIONAL CONFERENCE ON ARTIFICIAL NEURAL NETWORKS》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110752032A (en) * | 2019-12-02 | 2020-02-04 | 山东浪潮人工智能研究院有限公司 | Traditional Chinese medicine diagnosis method based on convolutional neural network and laser vibration measurement |
CN113076878A (en) * | 2021-04-02 | 2021-07-06 | 郑州大学 | Physique identification method based on attention mechanism convolution network structure |
CN113229798A (en) * | 2021-05-18 | 2021-08-10 | 平安科技(深圳)有限公司 | Model migration training method and device, computer equipment and readable storage medium |
CN113229798B (en) * | 2021-05-18 | 2023-08-22 | 平安科技(深圳)有限公司 | Model migration training method, device, computer equipment and readable storage medium |
CN115374125A (en) * | 2022-09-01 | 2022-11-22 | 无锡市华焯光电科技有限公司 | Pulse condition diagnosis and classification method, database construction method, device and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109363632A (en) | The deciphering method of pulse profile data and the solution read apparatus of pulse profile data | |
CN110507335B (en) | Multi-mode information based criminal psychological health state assessment method and system | |
CN106295139B (en) | A kind of tongue body autodiagnosis health cloud service system based on depth convolutional neural networks | |
CN109117744A (en) | A kind of twin neural network training method for face verification | |
CN106682616A (en) | Newborn-painful-expression recognition method based on dual-channel-characteristic deep learning | |
CN110516161A (en) | A kind of recommended method and device | |
CN108986912A (en) | Chinese medicine stomach trouble tongue based on deep learning is as information intelligent processing method | |
CN111785366B (en) | Patient treatment scheme determination method and device and computer equipment | |
CN108363979A (en) | Neonatal pain expression recognition method based on binary channels Three dimensional convolution neural network | |
Yu et al. | A hybrid user experience evaluation method for mobile games | |
CN108597621A (en) | Health status monitoring device, system and method based on theory of traditional Chinese medical science | |
CN110189305A (en) | A kind of multitask tongue picture automatic analysis method | |
CN109147880A (en) | Management system and method based on Internet of Things block chain quantum bit artificial intelligence | |
CN110037710A (en) | The System and method for of non-intrusion type estimation HBA1C and blood glucose value | |
Abe et al. | The influence of facial muscle training on the facial soft tissue profile: A brief review | |
CN109106345A (en) | Pulse signal characteristic detection method and device | |
Coelho et al. | Multimodal biometric authentication method by federated learning | |
Li et al. | A Comparison of Personalized and Generalized Approaches to Emotion Recognition Using Consumer Wearable Devices: Machine Learning Study | |
Chen et al. | DFML: Dynamic federated meta-learning for rare disease prediction | |
CN208422436U (en) | Health status monitoring device and system based on theory of traditional Chinese medical science | |
Borup et al. | Automatic sleep scoring using patient-specific ensemble models and knowledge distillation for ear-EEG data | |
Daniel et al. | Relationship between EMG and fNIRS during Dynamic Movements | |
Ninh et al. | An improved subject-independent stress detection model applied to consumer-grade wearable devices | |
Jaiswal et al. | Understanding cognitive fatigue from fmri scans with self-supervised learning | |
Aitken et al. | A contribution to the discussion of probabilities and human hair comparisons |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190222 |
|
RJ01 | Rejection of invention patent application after publication |