CN116649924A - Pulse analysis method and device - Google Patents
Pulse analysis method and device Download PDFInfo
- Publication number
- CN116649924A CN116649924A CN202310661630.2A CN202310661630A CN116649924A CN 116649924 A CN116649924 A CN 116649924A CN 202310661630 A CN202310661630 A CN 202310661630A CN 116649924 A CN116649924 A CN 116649924A
- Authority
- CN
- China
- Prior art keywords
- pulse
- features
- gaussian mixture
- frequency domain
- time domain
- 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
- 238000004458 analytical method Methods 0.000 title claims abstract description 28
- 239000000203 mixture Substances 0.000 claims abstract description 72
- 238000003062 neural network model Methods 0.000 claims abstract description 49
- 238000012549 training Methods 0.000 claims description 41
- 238000012360 testing method Methods 0.000 claims description 28
- 238000000034 method Methods 0.000 claims description 15
- 238000012847 principal component analysis method Methods 0.000 claims description 13
- 239000011159 matrix material Substances 0.000 claims description 10
- 238000002372 labelling Methods 0.000 claims description 8
- 238000005315 distribution function Methods 0.000 claims description 5
- 238000000605 extraction Methods 0.000 claims description 3
- 238000009826 distribution Methods 0.000 description 11
- 238000003745 diagnosis Methods 0.000 description 5
- 210000004204 blood vessel Anatomy 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 210000001367 artery Anatomy 0.000 description 3
- 201000010099 disease Diseases 0.000 description 3
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 3
- 210000004369 blood Anatomy 0.000 description 2
- 239000008280 blood Substances 0.000 description 2
- 230000017531 blood circulation Effects 0.000 description 2
- 230000008602 contraction Effects 0.000 description 2
- 230000007812 deficiency Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 239000003814 drug Substances 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 238000012935 Averaging Methods 0.000 description 1
- 108010076504 Protein Sorting Signals Proteins 0.000 description 1
- 238000012952 Resampling Methods 0.000 description 1
- 210000000709 aorta Anatomy 0.000 description 1
- 230000000747 cardiac effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000003205 diastolic effect Effects 0.000 description 1
- 230000002526 effect on cardiovascular system Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 239000010985 leather Substances 0.000 description 1
- 210000005240 left ventricle Anatomy 0.000 description 1
- 238000012821 model calculation Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 238000000513 principal component analysis Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000004393 prognosis Methods 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 230000000644 propagated effect Effects 0.000 description 1
- 230000010349 pulsation Effects 0.000 description 1
- 230000035485 pulse pressure Effects 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 210000002321 radial artery Anatomy 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 210000000707 wrist Anatomy 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/02—Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
-
- 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
-
- 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
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
-
- 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
- G06N3/047—Probabilistic or stochastic 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
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2123/00—Data types
- G06F2123/02—Data types in the time domain, e.g. time-series data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
- G06F2218/06—Denoising by applying a scale-space analysis, e.g. using wavelet analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
- G06F2218/10—Feature extraction by analysing the shape of a waveform, e.g. extracting parameters relating to peaks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E30/00—Energy generation of nuclear origin
- Y02E30/30—Nuclear fission reactors
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Heart & Thoracic Surgery (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Surgery (AREA)
- Public Health (AREA)
- Veterinary Medicine (AREA)
- Animal Behavior & Ethology (AREA)
- Pathology (AREA)
- Physiology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Psychiatry (AREA)
- Signal Processing (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Fuzzy Systems (AREA)
- Computational Linguistics (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Alternative & Traditional Medicine (AREA)
- Cardiology (AREA)
- Probability & Statistics with Applications (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
A pulse analysis method and device comprises the following steps: s1, acquiring three groups of pulse data sets corresponding to three parts of an cun, guan and chi; s2, acquiring frequency domain features, time domain features and Gaussian mixture model features of each pulse data; s3, inputting the frequency domain features, the time domain features and the Gaussian mixture model features of each pulse data into a pre-trained neural network model, and outputting a predicted pulse classification result. By adopting the technical scheme, compared with the prior art, the pulse analysis method and the device can realize objective and accurate analysis of the pulse of the patient. On the other hand, since each group of pulse data sets includes a plurality of pulse data acquired by a plurality of pulse acquisition units, such multi-channel pulse data can improve the accuracy of prediction relative to single-channel pulse data. Secondly, the introduction of Gaussian mixture model features can further improve the accuracy of the model.
Description
Technical Field
The application relates to the technical field of pulse analysis, in particular to a pulse analysis method and device.
Background
In the blood circulation process in the cardiac blood vessel, blood flows into the aorta from the left ventricle under the contraction action of the heart, so that the wall of the proximal end of the artery expands and the internal pressure is increased; when the heart is relaxed, the ejection of blood is stopped temporarily, and the artery resumes its contraction under the action of the elasticity of the blood vessel. In the systolic and diastolic processes, blood is rapidly propagated to the distal end along the blood vessel with the heart as a starting point under the action of the pressure difference, so that the distal artery also has similar pulsation to form a pulse wave, and the signal carries a large amount of information related to the cardiovascular state. In pulse diagnosis of the traditional Chinese medicine diagnosis method in China, a doctor senses the periodic expansion change of the blood vessel by pressing the wrist radial artery, and divides the sensed pulse form into different pulse conditions. The pulse condition can be used for judging the nature of diseases such as cold, heat, deficiency and excess, and also can be used for judging disease prognosis, and is an important objective index for identifying diseases in traditional Chinese medicine.
Pulse diagnosis is a noninvasive diagnosis method, has the characteristics of convenient diagnosis, low cost, effective result and the like, and has a relatively perfect evaluation and intervention system. However, different doctors have different standards on the pulse intensity due to uneven medical level, so that the pulse of a patient cannot be objectively and accurately analyzed simply by subjective feeling and experience of the doctors. Therefore, there is a need to develop a pulse analysis method and device capable of objectively and accurately analyzing the pulse of a patient.
Disclosure of Invention
The application solves the problems that the pulse intensity of different doctors has different standards due to uneven medical level of different doctors in the prior art, so that objective and accurate analysis of the pulse of a patient cannot be performed.
In order to achieve the above purpose, the following technical scheme is adopted in the embodiment of the application.
In one aspect, a pulse analysis method is provided, including the steps of:
s1, three groups of pulse data sets corresponding to three parts of an cun, an guan and an chi are acquired, and each group of pulse data set comprises a plurality of pulse data acquired by a plurality of pulse acquisition units;
s2, acquiring frequency domain features, time domain features and Gaussian mixture model features of each pulse data; the Gaussian mixture model is characterized by weight, mean and covariance matrix of each Gaussian distribution function in the Gaussian mixture model;
s3, inputting the frequency domain features, the time domain features and the Gaussian mixture model features of each pulse data into a pre-trained neural network model, and outputting a predicted pulse classification result.
The pulse analysis method can realize objective and accurate analysis of the pulse of the patient. On the other hand, since each group of pulse data sets includes a plurality of pulse data acquired by a plurality of pulse acquisition units, such multi-channel pulse data can improve the accuracy of prediction relative to single-channel pulse data. Secondly, the introduction of Gaussian mixture model features can further improve the accuracy of the model.
In order to accelerate the model calculation speed, in some embodiments, a step of removing redundant frequency domain features, time domain features and Gaussian mixture model features by adopting a principal component analysis method is further included between the steps S2 and S3.
In some embodiments, the trained neural network model is obtained according to the following steps:
the method comprises the steps of obtaining a pulse sample data set, performing pulse condition labeling on the pulse sample data set to obtain pulse condition classification, dividing the pulse sample data set into a training set and a testing set, obtaining frequency domain characteristics, time domain characteristics and Gaussian mixture model characteristics of each pulse sample data in the training set and the testing set, and removing redundant frequency domain characteristics, time domain characteristics and Gaussian mixture models by adopting a principal component analysis method;
training the neural network model by using the frequency domain features, the time domain features and the Gaussian mixture model features of all pulse sample data in the training set, and evaluating the accuracy of the neural network model by using the frequency domain features, the time domain features and the Gaussian mixture model features of all pulse sample data in the testing set to obtain a trained neural network model.
In some embodiments, the pulse data comprises pulse width; in step S3, the frequency domain feature, the time domain feature, the gaussian mixture model feature and the pulse width of each pulse data are input into a pre-trained neural network model, and a predicted pulse classification result is output.
In some embodiments, a pulse sample data set is obtained, pulse condition labeling is carried out on the pulse sample data set to obtain pulse condition classification, the pulse sample data set is divided into a training set and a testing set, frequency domain characteristics, time domain characteristics and Gaussian mixture model characteristics of each pulse sample data in the training set and the testing set are obtained, and a principal component analysis method is adopted to remove redundant frequency domain characteristics, time domain characteristics and Gaussian mixture models;
training the neural network model by using the frequency domain features, the time domain features, the Gaussian mixture model features and the pulse width of all pulse sample data in the training set, and evaluating the accuracy of the neural network model by using the frequency domain features, the time domain features, the Gaussian mixture model features and the pulse width of all pulse sample data in the testing set to obtain a trained neural network model.
In yet another aspect, there is provided a pulse analysis device including:
the first acquisition module is used for acquiring three groups of pulse data sets corresponding to three parts of the cun, guan and chi, wherein each group of pulse data set comprises a plurality of pulse data acquired by a plurality of pulse acquisition units;
the feature extraction module is used for acquiring frequency domain features, time domain features and Gaussian mixture model features of each pulse data; the Gaussian mixture model is characterized by weight, mean and covariance matrix of each Gaussian distribution function in the Gaussian mixture model; and
the prediction module is used for inputting the frequency domain characteristics, the time domain characteristics and the Gaussian mixture model characteristics of each pulse data into a pre-trained neural network model and outputting a predicted pulse classification result.
In some embodiments, the pulse analysis device further includes a feature elimination module for eliminating redundant frequency domain features, time domain features, and gaussian mixture model features using principal component analysis.
In some embodiments, the pulse analysis device further comprises a model training module for obtaining a trained neural network model according to the steps of:
the method comprises the steps of obtaining a pulse sample data set, performing pulse condition labeling on the pulse sample data set to obtain pulse condition classification, dividing the pulse sample data set into a training set and a testing set, obtaining frequency domain characteristics, time domain characteristics and Gaussian mixture model characteristics of each pulse sample data in the training set and the testing set, and removing redundant frequency domain characteristics, time domain characteristics and Gaussian mixture models by adopting a principal component analysis method;
training the neural network model by using the frequency domain features, the time domain features and the Gaussian mixture model features of all pulse sample data in the training set, and evaluating the accuracy of the neural network model by using the frequency domain features, the time domain features and the Gaussian mixture model features of all pulse sample data in the testing set to obtain a trained neural network model.
In some embodiments, the pulse data comprises pulse width; the prediction module is used for inputting the frequency domain characteristics, the time domain characteristics, the Gaussian mixture model characteristics and the pulse width of each pulse data into a pre-trained neural network model, and outputting a predicted pulse classification result.
In some embodiments, the method further comprises a model training module for obtaining a trained neural network model according to the steps of: the method comprises the steps of obtaining a pulse sample data set, performing pulse condition labeling on the pulse sample data set to obtain pulse condition classification, dividing the pulse sample data set into a training set and a testing set, obtaining frequency domain characteristics, time domain characteristics and Gaussian mixture model characteristics of each pulse sample data in the training set and the testing set, and removing redundant frequency domain characteristics, time domain characteristics and Gaussian mixture models by adopting a principal component analysis method;
training the neural network model by using the frequency domain features, the time domain features, the Gaussian mixture model features and the pulse width of all pulse sample data in the training set, and evaluating the accuracy of the neural network model by using the frequency domain features, the time domain features, the Gaussian mixture model features and the pulse width of all pulse sample data in the testing set to obtain a trained neural network model.
One or more technical solutions provided in the embodiments of the present application at least have the following technical effects or advantages: the problem of objective and accurate analysis of the pulse of the patient can be realized. On the other hand, since each group of pulse data sets includes a plurality of pulse data acquired by a plurality of pulse acquisition units, such multi-channel pulse data can improve the accuracy of prediction relative to single-channel pulse data. Secondly, the introduction of Gaussian mixture model features can further improve the accuracy of the model.
Drawings
FIG. 1 is a flow chart of a pulse analysis method according to an embodiment of the application;
FIG. 2 is a diagram showing the comparison of the original pulse signal and the denoised reconstructed pulse signal according to an embodiment of the present application;
fig. 3 is a schematic diagram of pulse wave according to an embodiment of the application.
Detailed Description
In order to better understand the above technical solutions, the following detailed description will refer to the accompanying drawings and specific embodiments.
Referring to fig. 1, a pulse analysis method includes the steps of:
s1, three groups of pulse data sets corresponding to three parts of an cun, an guan and an chi are acquired, and each group of pulse data set comprises a plurality of pulse data acquired by a plurality of pulse acquisition units;
note that the pulse data set is a set of preprocessed pulse data. In this embodiment, wavelet transformation is adopted to preprocess pulse data, and the specific steps are as follows: and performing multi-scale wavelet decomposition on the original signal, performing threshold quantization processing according to the obtained high-frequency detail component and low-frequency approximate component, and performing wavelet reconstruction to obtain a denoised signal. Fig. 2 is a graph comparing an original signal and a reconstructed signal after denoising.
S2, obtaining frequency domain features, time domain features and Gaussian mixture model features of each pulse data, and removing redundant frequency domain features, time domain features and Gaussian mixture model features by adopting a principal component analysis method. The Gaussian mixture model is characterized by weight, mean and covariance matrix of each Gaussian distribution function in the Gaussian mixture model.
After preprocessing the pulse signal data, carrying out time domain feature extraction on the single-period pulse signal after resampling and averaging on each sample. As shown in fig. 3, the time-domain features include 6 amplitude scale features (h 2/h1 (down-branch height/main wave height), h3/h1 (tidal wave height/main wave height), h4/h1 (down-branch canyon height/main wave height), h5/h1 (microblog height/main wave height) and abscissa width w at main wave height 2/3), 6 time scale features (T1 (main wave time)/T, T2 (down-branch time)/T, T3 (tidal wave time)/T, T4 (down-branch time)/T, T5 (microblog wave time)/T), and also mean, variance, root mean square and offset of pulse pressure, waveform factors, peak factors, pulse factors, margin factors, and 20 time-domain features.
The frequency domain features are required to be extracted from the filtered multicycle signals, the power spectrum estimation is carried out on the reconstructed pulse signal sequence, and the extracted frequency domain features comprise 6 frequency domain features, namely a frequency maximum value, a frequency domain variance, a frequency domain range, a rectification average value (an average value of signal absolute value integration), a frequency domain skewness and a frequency domain root mean square.
The monocycle pulse wave may be modeled as a gaussian mixture assuming that the monocycle pulse wave dataset is x=x 1 ，x 2 ，x 3 ，...x N The pulse wave of single period has 2 Gaussian distributions, and each Gaussian distribution has a parameter of theta k ＝(π k ，μ k ，∑ k ) Wherein pi is k Is the weight of the gaussian distribution (satisfy)，μ k Is the mean value of the Gaussian distribution, sigma k Is the covariance matrix of the gaussian distribution.
Then for each sample x i Its responsivity to the kth gaussian distribution is:
where N (x|μ, Σ) is a probability density function of a multidimensional gaussian distribution, representing the probability density value of x in a gaussian distribution with mean μ and covariance matrix Σ.
From the responsivity described above, a new gaussian weight, gaussian mean and Gao Sixie variance matrix for each gaussian distribution can be calculated:
1. gaussian weight pi k The estimated values of (2) are:
i.e. for each sample x i An average value of responsivity to the kth gaussian distribution.
2. Gaussian mean mu k The estimated values of (2) are:
i.e. for each sample x i Multiplying the weighted average of its responsivity to the kth gaussian distributionAnd (5) an average value.
3. Gao Sixie variance matrix Σ k The estimated values of (2) are:
i.e. for each sample x i Subtracting the corresponding mean mu k Then multiplied by the weighted covariance matrix after its weighting of the responsivity to the kth gaussian distribution.
The monocycle pulse waveform can be modeled by using two Gaussian functions, and the monocycle pulse waveform comprises 6 Gaussian mixture model parameter characteristics of Gaussian weights, gaussian mean values and Gao Sixie variances.
S3, inputting the frequency domain features, the time domain features and the Gaussian mixture model features of each pulse data into a pre-trained neural network model, and outputting a predicted pulse classification result.
The trained neural network model can be obtained according to the following steps:
the method comprises the steps of acquiring a pulse sample data set, wherein the pulse sample data set is also a preprocessed data set, and the preprocessing method is consistent with that of the pulse data set. The pulse sample data set should include data of all pulse conditions, and the number of samples corresponding to each pulse condition is not less than one thousand people.
Pulse condition classification (pulse condition categories include 28 pulse conditions of floating, sinking, slow, count, slippery, astringent, deficiency, real, long, short, flood, micro, tight, slow, chord, hollow, leather, firm, soft, weak, scattered, fine, volt, dynamic, promoting, knot, generation and big) is obtained by marking pulse sample data sets, the pulse sample data sets are divided into training sets and test sets, frequency domain characteristics, time domain characteristics and Gaussian mixture model characteristics of each pulse sample data in the training sets and the test sets are obtained, and redundant frequency domain characteristics, time domain characteristics and Gaussian mixture models are removed by adopting a principal component analysis method.
Training the neural network model by using the frequency domain features, the time domain features and the Gaussian mixture model features of all pulse sample data in the training set, and evaluating the accuracy of the neural network model by using the frequency domain features, the time domain features and the Gaussian mixture model features of all pulse sample data in the testing set to obtain a trained neural network model. When the accuracy of the neural network model is evaluated, the Adam optimizer is utilized to update parameters of the neural network model according to the first moment and the second moment of the loss function gradient of the neural network model until the accuracy reaches a preset threshold (such as 95%).
Example two
The present embodiment is basically identical to the first embodiment except that the pulse data includes pulse width in addition to the pulse wave data, and the pulse width is also input as one of the features into the neural network model (including training, testing, and prediction).
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present application, and are not limiting; although the application has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present application.
Claims (10)
1. A pulse analysis method comprising the steps of:
s1, three groups of pulse data sets corresponding to three parts of an cun, an guan and an chi are acquired, and each group of pulse data set comprises a plurality of pulse data acquired by a plurality of pulse acquisition units;
s2, acquiring frequency domain features, time domain features and Gaussian mixture model features of each pulse data; the Gaussian mixture model is characterized by weight, mean and covariance matrix of each Gaussian distribution function in the Gaussian mixture model;
s3, inputting the frequency domain features, the time domain features and the Gaussian mixture model features of each pulse data into a pre-trained neural network model, and outputting a predicted pulse classification result.
2. The pulse analysis method according to claim 1, wherein between the steps S2 and S3, a step of removing redundant frequency domain features, time domain features and gaussian mixture model features by using a principal component analysis method is further included.
3. The pulse analysis method according to claim 2, characterized in that: obtaining a trained neural network model according to the following steps:
the method comprises the steps of obtaining a pulse sample data set, performing pulse condition labeling on the pulse sample data set to obtain pulse condition classification, dividing the pulse sample data set into a training set and a testing set, obtaining frequency domain characteristics, time domain characteristics and Gaussian mixture model characteristics of each pulse sample data in the training set and the testing set, and removing redundant frequency domain characteristics, time domain characteristics and Gaussian mixture models by adopting a principal component analysis method;
training the neural network model by using the frequency domain features, the time domain features and the Gaussian mixture model features of all pulse sample data in the training set, and evaluating the accuracy of the neural network model by using the frequency domain features, the time domain features and the Gaussian mixture model features of all pulse sample data in the testing set to obtain a trained neural network model.
4. The pulse analysis method according to claim 2, characterized in that: the pulse data includes pulse width; in step S3, the frequency domain feature, the time domain feature, the gaussian mixture model feature and the pulse width of each pulse data are input into a pre-trained neural network model, and a predicted pulse classification result is output.
5. The pulse analysis method of claim 4, wherein the trained neural network model is obtained according to the steps of:
the method comprises the steps of obtaining a pulse sample data set, performing pulse condition labeling on the pulse sample data set to obtain pulse condition classification, dividing the pulse sample data set into a training set and a testing set, obtaining frequency domain characteristics, time domain characteristics and Gaussian mixture model characteristics of each pulse sample data in the training set and the testing set, and removing redundant frequency domain characteristics, time domain characteristics and Gaussian mixture models by adopting a principal component analysis method;
training the neural network model by using the frequency domain features, the time domain features, the Gaussian mixture model features and the pulse width of all pulse sample data in the training set, and evaluating the accuracy of the neural network model by using the frequency domain features, the time domain features, the Gaussian mixture model features and the pulse width of all pulse sample data in the testing set to obtain a trained neural network model.
6. A pulse analysis device, comprising:
the first acquisition module is used for acquiring three groups of pulse data sets corresponding to three parts of the cun, guan and chi, wherein each group of pulse data set comprises a plurality of pulse data acquired by a plurality of pulse acquisition units;
the feature extraction module is used for acquiring frequency domain features, time domain features and Gaussian mixture model features of each pulse data; wherein the method comprises the steps of
The Gaussian mixture model is characterized by weight, mean and covariance matrix of each Gaussian distribution function in the Gaussian mixture model; and
the prediction module is used for inputting the frequency domain characteristics, the time domain characteristics and the Gaussian mixture model characteristics of each pulse data into a pre-trained neural network model and outputting a predicted pulse classification result.
7. The pulse analysis device of claim 6, wherein: the feature eliminating module is used for eliminating redundant frequency domain features, time domain features and Gaussian mixture model features by adopting a principal component analysis method.
8. The pulse analysis device of claim 7, further comprising a model training module for obtaining a trained neural network model according to the steps of:
the method comprises the steps of obtaining a pulse sample data set, performing pulse condition labeling on the pulse sample data set to obtain pulse condition classification, dividing the pulse sample data set into a training set and a testing set, obtaining frequency domain characteristics, time domain characteristics and Gaussian mixture model characteristics of each pulse sample data in the training set and the testing set, and removing redundant frequency domain characteristics, time domain characteristics and Gaussian mixture models by adopting a principal component analysis method;
training the neural network model by using the frequency domain features, the time domain features and the Gaussian mixture model features of all pulse sample data in the training set, and evaluating the accuracy of the neural network model by using the frequency domain features, the time domain features and the Gaussian mixture model features of all pulse sample data in the testing set to obtain a trained neural network model.
9. The pulse analysis device of claim 6, wherein: the pulse data includes pulse width; the prediction module is used for inputting the frequency domain characteristics, the time domain characteristics, the Gaussian mixture model characteristics and the pulse width of each pulse data into a pre-trained neural network model, and outputting a predicted pulse classification result.
10. The pulse analysis device of claim 9, wherein: the model training module is used for acquiring a trained neural network model according to the following steps:
the method comprises the steps of obtaining a pulse sample data set, performing pulse condition labeling on the pulse sample data set to obtain pulse condition classification, dividing the pulse sample data set into a training set and a testing set, obtaining frequency domain characteristics, time domain characteristics and Gaussian mixture model characteristics of each pulse sample data in the training set and the testing set, and removing redundant frequency domain characteristics, time domain characteristics and Gaussian mixture models by adopting a principal component analysis method;
training the neural network model by using the frequency domain features, the time domain features, the Gaussian mixture model features and the pulse width of all pulse sample data in the training set, and evaluating the accuracy of the neural network model by using the frequency domain features, the time domain features, the Gaussian mixture model features and the pulse width of all pulse sample data in the testing set to obtain a trained neural network model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310661630.2A CN116649924A (en) | 2023-06-06 | 2023-06-06 | Pulse analysis method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310661630.2A CN116649924A (en) | 2023-06-06 | 2023-06-06 | Pulse analysis method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116649924A true CN116649924A (en) | 2023-08-29 |
Family
ID=87718680
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310661630.2A Pending CN116649924A (en) | 2023-06-06 | 2023-06-06 | Pulse analysis method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116649924A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117224092A (en) * | 2023-11-16 | 2023-12-15 | 常熟理工学院 | Photoelectric volume pulse wave interference band real-time detection method and system based on decision tree |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107049269A (en) * | 2017-03-06 | 2017-08-18 | 武汉嫦娥医学抗衰机器人股份有限公司 | A kind of Pulse Signals Treatment Analysis system |
| CN107822608A (en) * | 2017-10-26 | 2018-03-23 | 中国民航大学 | Pulse wave feature extracting method based on gauss hybrid models |
| CN108670209A (en) * | 2018-03-29 | 2018-10-19 | 中国科学院微电子研究所 | A kind of method and system of automatic identification Chinese medicine pulse |
| CN112487945A (en) * | 2020-11-26 | 2021-03-12 | 上海贝业斯健康科技有限公司 | Pulse condition identification method based on double-path convolution neural network fusion |
| CN116089797A (en) * | 2023-02-08 | 2023-05-09 | 苏州大学 | Pulse condition identification method and system based on convolutional neural network |
-
2023
- 2023-06-06 CN CN202310661630.2A patent/CN116649924A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107049269A (en) * | 2017-03-06 | 2017-08-18 | 武汉嫦娥医学抗衰机器人股份有限公司 | A kind of Pulse Signals Treatment Analysis system |
| CN107822608A (en) * | 2017-10-26 | 2018-03-23 | 中国民航大学 | Pulse wave feature extracting method based on gauss hybrid models |
| CN108670209A (en) * | 2018-03-29 | 2018-10-19 | 中国科学院微电子研究所 | A kind of method and system of automatic identification Chinese medicine pulse |
| CN112487945A (en) * | 2020-11-26 | 2021-03-12 | 上海贝业斯健康科技有限公司 | Pulse condition identification method based on double-path convolution neural network fusion |
| CN116089797A (en) * | 2023-02-08 | 2023-05-09 | 苏州大学 | Pulse condition identification method and system based on convolutional neural network |
Non-Patent Citations (1)
| Title |
|---|
| 一种基于随机森林的脉象信号特征降维与分类研究: "一种基于随机森林的脉象信号特征降维与分类研究", 一种基于随机森林的脉象信号特征降维与分类研究, pages 2418 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117224092A (en) * | 2023-11-16 | 2023-12-15 | 常熟理工学院 | Photoelectric volume pulse wave interference band real-time detection method and system based on decision tree |
| CN117224092B (en) * | 2023-11-16 | 2024-02-09 | 常熟理工学院 | Photoelectric volume pulse wave interference band real-time detection method and system based on decision tree |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109998525B (en) | Arrhythmia automatic classification method based on discriminant deep belief network | |
| Chen et al. | Wrist pulse signal diagnosis using modified Gaussian models and Fuzzy C-Means classification | |
| Gupta et al. | Neural network classification of homomorphic segmented heart sounds | |
| CN109961017A (en) | A kind of cardiechema signals classification method based on convolution loop neural network | |
| CN108670209A (en) | A kind of method and system of automatic identification Chinese medicine pulse | |
| CN109948396B (en) | Heart beat classification method, heart beat classification device and electronic equipment | |
| CN109864714A (en) | A kind of ECG Signal Analysis method based on deep learning | |
| CN112307959B (en) | Wavelet denoising method for electrocardiosignal analysis | |
| Thakker et al. | Wrist pulse signal classification for health diagnosis | |
| CN116649924A (en) | Pulse analysis method and device | |
| CN114469124A (en) | Method for identifying abnormal electrocardiosignals in motion process | |
| CN113116300A (en) | Physiological signal classification method based on model fusion | |
| CN113729648B (en) | Wearable pulse-taking bracelet system based on multiple pulse sensors | |
| CN115089139A (en) | Personalized physiological parameter measuring method combining biological characteristic identification | |
| CN111370120A (en) | Method for detecting diastolic dysfunction based on heart sound signals | |
| Garcia et al. | A multiple linear regression model for carotid-to-femoral pulse wave velocity estimation based on schrodinger spectrum characterization | |
| CN114159079A (en) | Multi-type muscle fatigue detection method based on feature extraction and GRU deep learning model | |
| CN117281479A (en) | Human lower limb chronic pain distinguishing method, storage medium and device based on surface electromyographic signal multi-dimensional feature fusion | |
| CN114145725B (en) | PPG sampling rate estimation method based on noninvasive continuous blood pressure measurement | |
| CN116451129A (en) | Pulse classification and identification method and system | |
| CN116451110A (en) | Blood glucose prediction model construction method based on signal energy characteristics and pulse period | |
| CN113925495B (en) | Arterial and venous fistula abnormal tremor signal identification system and method combining statistical learning and time-frequency analysis | |
| Ayushi et al. | A survey of ECG classification for arrhythmia diagnoses using SVM | |
| CN115067912A (en) | Non-invasive blood viscosity prediction method and device based on ELM model | |
| CN115177260A (en) | Intelligent electrocardio-heart sound diagnosis method and device based on artificial neural network |
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 |