CN114469017A - Pulse condition data detection method and system - Google Patents
Pulse condition data detection method and system Download PDFInfo
- Publication number
- CN114469017A CN114469017A CN202210102117.5A CN202210102117A CN114469017A CN 114469017 A CN114469017 A CN 114469017A CN 202210102117 A CN202210102117 A CN 202210102117A CN 114469017 A CN114469017 A CN 114469017A
- Authority
- CN
- China
- Prior art keywords
- pulse condition
- data
- pulse
- result
- valid
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 111
- 238000000034 method Methods 0.000 claims abstract description 37
- 238000012544 monitoring process Methods 0.000 claims abstract description 17
- 238000000605 extraction Methods 0.000 claims description 65
- 238000012549 training Methods 0.000 claims description 64
- 238000004364 calculation method Methods 0.000 claims description 24
- 238000012545 processing Methods 0.000 claims description 19
- 230000004913 activation Effects 0.000 claims description 10
- 230000004927 fusion Effects 0.000 claims description 5
- 238000011176 pooling Methods 0.000 claims description 4
- 238000004804 winding Methods 0.000 claims description 4
- 238000002759 z-score normalization Methods 0.000 claims description 4
- 230000008569 process Effects 0.000 abstract description 18
- 230000008092 positive effect Effects 0.000 abstract description 5
- 230000006870 function Effects 0.000 description 25
- 238000010586 diagram Methods 0.000 description 8
- 238000013528 artificial neural network Methods 0.000 description 5
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 3
- 239000000470 constituent Substances 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 230000008034 disappearance Effects 0.000 description 2
- 239000012634 fragment Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 235000019800 disodium phosphate Nutrition 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000010223 real-time analysis Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
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
- 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 pulse condition data detection method and the pulse condition data detection system obtain a pulse condition judgment result and a pulse rate monitoring result corresponding to pulse condition data to be detected through the constructed pulse condition detection model. The scheme of the invention can analyze whether the pulse condition is valid data or not while detecting the pulse condition, and predict the numerical value of the pulse rate under the condition of judging the correct pulse condition signal; and the invalid acquisition process can be quitted in advance, so that the invalid acquisition time is greatly shortened, and the method has a positive effect on improving the user experience of pulse condition detection products.
Description
Technical Field
The present application relates to the field of data processing technologies, and in particular, to a pulse condition data detection method and system.
Background
In the field of professional traditional Chinese medicine pulse manifestation appearance detection, a long time is usually needed for carrying out a pulse manifestation detection process, the time often exceeds 5 minutes, but because the positioning requirement of equipment to the detection position is extremely high, the problems of incorrect wearing and the like often exist, and effective pulse manifestation signals cannot be obtained. The conventional pulse condition signal processing method needs to analyze after the signal acquisition is completed, and if the correct signal is not successfully detected by wearing, a large amount of invalid experience time is wasted.
Most of the previous researches and patents focus on the processing from pulse signals to the features of pulse, such as position, number, shape, and potential, and few mention is made about how to determine valid data from raw data. For example, in the dynamic three-part pulse signal continuous monitoring and real-time analysis system disclosed in CN106859608A, the signals detected by the default devices are all correct and valid pulse signals, and the processing procedure of unsuccessful acquisition of signals is ignored. In a pulse condition recognition method based on two-way convolutional neural network fusion disclosed in CN112487945A, although an artificial intelligence technique is used to process pulse condition signals, the network structure cannot process abnormal signals, which is also only for the pulse condition signals that are deemed to be correct and effective by preprocessing.
Disclosure of Invention
In view of the above-mentioned shortcomings of the prior art, the present application is directed to a method and system for detecting pulse condition data, which solves the above-mentioned problems of the prior art.
To achieve the above and other related objects, the present application provides a pulse condition data detecting method, comprising: acquiring pulse condition data to be detected; based on the deployed pulse condition detection model, obtaining a pulse condition detection result corresponding to the data according to the pulse condition data; wherein, the pulse condition detection result comprises: pulse condition discrimination result and pulse rate monitoring result; and wherein the pulse condition discrimination result comprises: a valid or invalid pulse condition result; the pulse condition detection model is constructed in a mode comprising the following steps: constructing a pulse condition detection framework model; training the pulse condition detection framework model by utilizing a training data set to obtain a pulse condition detection model; and wherein the training data set comprises: a valid training data set and an invalid training data set.
In one or more embodiments of the present application, the valid training data set includes: a plurality of valid pulse condition data samples with valid classification labels and pulse rates corresponding to the samples; the invalid training data set comprises: a plurality of invalid pulse condition data samples having invalid classification labels.
In one or more embodiments of the present application, the training data set is obtained by a Z-Score normalization process.
In one or more embodiments of the present application, the pulse condition detection framework model includes: the characteristic crude extraction structure is used for carrying out crude characteristic extraction on the input pulse condition data so as to output pulse condition crude characteristic extraction data; the characteristic fine extraction structure is connected with the characteristic crude extraction structure and is used for performing fine characteristic extraction on the pulse condition coarse characteristic extraction data to output pulse condition fine characteristic extraction data; the pulse condition detection result output structure is connected with the characteristic fine extraction structure and is used for obtaining a pulse condition detection result corresponding to the pulse condition data based on the input fine characteristic extraction data; wherein, the pulse condition detection result output structure comprises: the data validity judging output structure is used for extracting data based on the input fine pulse characteristics and outputting a valid pulse condition result or an invalid pulse condition result; and the pulse rate output structure is used for extracting data and outputting a pulse rate monitoring result based on the input fine pulse condition characteristics.
In one or more embodiments of the present application, the data valid decision output structure includes: two layers of connected Dense structures; and/or, the pulse rate output structure comprises: two layers of connected Dense structure.
In one or more embodiments of the present application, the characteristic crude extraction structure comprises: two connected feature rough extraction structures; wherein, each coarse extraction structure of characteristics includes: the multilayer structure comprises a winding layer, a Dropout layer connected with the winding layer and a BatchNormalization layer connected with the Dropout layer.
In one or more embodiments of the present application, the feature refinement extraction structure includes: a branch feature extraction structure comprising: three branch structures corresponding to different convolution kernels and used for respectively outputting corresponding branch characteristic data according to input coarse characteristic extraction data; wherein each branch structure comprises: the three-layer connected ConvBNRelu _ x structure is used for carrying out three times of residual error calculation on the input coarse feature extraction data; the global average pooling layer is used for carrying out average processing on the data subjected to the three-time residual error processing; and the fusion structure is connected with the branch feature extraction structure and used for fusing the branch feature data output by the three branch structures and outputting the fine feature extraction data.
In one or more embodiments of the present application, the activation function adopted by the data valid determination output structure is a sigmoid function; the activation function adopted by the pulse rate output structure is a Relu function.
In one or more embodiments of the present application, the pulse detection framework model further includes: the loss weighting calculation layer is connected with the pulse condition detection result output structure and comprises: the first loss weighting calculation module is used for performing loss weighting calculation on the pulse condition judgment result by adopting a binary cross entropy loss function; and the second loss weighting calculation module is used for performing loss weighting calculation on the pulse rate monitoring result by adopting an MSE loss function.
To achieve the above and other related objects, the present application provides a pulse condition data detecting system, comprising: the data acquisition module is used for acquiring pulse condition data to be detected; the pulse condition detection module is connected with the data acquisition module and used for acquiring a pulse condition detection result corresponding to the data according to the pulse condition data based on the deployed pulse condition detection model; wherein, the pulse condition detection result comprises: pulse condition discrimination result and pulse rate monitoring result; and wherein the pulse condition discrimination result comprises: a valid or invalid pulse condition result; the pulse condition detection model is constructed in a mode comprising the following steps: constructing a pulse condition detection framework model; training the pulse condition detection framework model by utilizing a training data set to obtain a pulse condition detection model; and wherein the training data set comprises: a valid training data set and an invalid training data set.
As described above, the pulse condition data detection method and system of the present application obtain the pulse condition discrimination result and the pulse rate monitoring result corresponding to the pulse condition data to be detected through the constructed pulse condition detection model. The scheme of the invention can analyze whether the pulse condition is valid data or not while detecting the pulse condition, and predict the numerical value of the pulse rate under the condition of judging the correct pulse condition signal; and the invalid acquisition process can be quitted in advance, so that the invalid acquisition time is greatly shortened, and the method has a positive effect on improving the user experience of pulse condition detection products.
Drawings
Fig. 1 is a schematic flow chart illustrating a pulse condition data detection method according to an embodiment of the present application.
Fig. 2 is a schematic structural diagram of a pulse condition detection framework model in an embodiment of the present application.
Fig. 3 is a schematic structural diagram of a coarse feature extraction structure in an embodiment of the present application.
Fig. 4 is a schematic structural diagram of a feature fine extraction structure in an embodiment of the present application.
Fig. 5 is a schematic structural diagram of a ConvBNRelu _ x structure in the embodiment of the present application.
FIG. 6 is a flowchart illustrating a pulse data detection method according to an embodiment of the present application.
Fig. 7 is a schematic structural diagram of a pulse condition detection framework model in an embodiment of the present application.
Fig. 8 is a schematic structural diagram of a pulse data detection system according to an embodiment of the present application.
Detailed Description
The following description of the embodiments of the present application is provided by way of specific examples, and other advantages and effects of the present application will be readily apparent to those skilled in the art from the disclosure herein. The present application is capable of other and different embodiments and of being practiced or of being carried out in various ways, and its several details are capable of modification in various respects, all without departing from the spirit and scope of the present application. It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict.
Embodiments of the present application will be described in detail below with reference to the accompanying drawings so that those skilled in the art to which the present application pertains can easily carry out the present application. The present application may be embodied in many different forms and is not limited to the embodiments described herein.
In order to clearly explain the present application, components that are not related to the description are omitted, and the same reference numerals are given to the same or similar components throughout the specification.
Throughout the specification, when a component is referred to as being "connected" to another component, this includes not only the case of being "directly connected" but also the case of being "indirectly connected" with another element interposed therebetween. In addition, when a component is referred to as "including" a certain constituent element, unless otherwise stated, it means that the component may include other constituent elements, without excluding other constituent elements.
When an element is referred to as being "on" another element, it can be directly on the other element, or intervening elements may also be present. When a component is said to be "directly on" another component, there are no intervening components present.
Although the terms first, second, etc. may be used herein to describe various elements in some instances, these elements should not be limited by these terms. These terms are only used to distinguish one element from another. For example, the first interface and the second interface, etc. are described. Also, as used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context indicates otherwise. It will be further understood that the terms "comprises," "comprising," "includes" and/or "including," when used in this specification, specify the presence of stated features, steps, operations, elements, components, items, species, and/or groups, but do not preclude the presence, or addition of one or more other features, steps, operations, elements, components, species, and/or groups thereof. The terms "or" and/or "as used herein are to be construed as inclusive or meaning any one or any combination. Thus, "A, B or C" or "A, B and/or C" means "any of the following: a; b; c; a and B; a and C; b and C; A. b and C ". An exception to this definition will occur only when a combination of elements, functions, steps or operations are inherently mutually exclusive in some way.
The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. As used herein, the singular forms "a", "an" and "the" include plural forms as long as the words do not expressly indicate a contrary meaning. The terms "comprises" and/or "comprising," when used in this specification, specify the presence of stated features, regions, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of other features, regions, integers, steps, operations, elements, components, and/or groups thereof.
Terms indicating "lower", "upper", and the like relative to space may be used to more easily describe a relationship of one component with respect to another component illustrated in the drawings. Such terms are intended to include not only the meanings indicated in the drawings, but also other meanings or operations of the device in use. For example, if the device in the figures is turned over, elements described as "below" other elements would then be oriented "above" the other elements. Thus, the exemplary terms "under" and "beneath" all include above and below. The device may be rotated 90 or other angles and the terminology representing relative space is also to be interpreted accordingly.
Although not defined differently, including technical and scientific terms used herein, all terms have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. Terms defined in commonly used dictionaries are to be additionally interpreted as having meanings consistent with those of related art documents and the contents of the present prompts, and must not be excessively interpreted as having ideal or very formulaic meanings unless defined.
In view of the deficiency of the prior art, a pulse condition data detection method is provided, and a pulse condition discrimination result and a pulse rate monitoring result corresponding to pulse condition data to be detected are obtained through a constructed pulse condition detection model. The scheme of the invention can analyze whether the pulse condition is valid data or not while detecting the pulse condition, and predict the numerical value of the pulse rate under the condition of judging the correct pulse condition signal; and the invalid acquisition process can be quitted in advance, so that the invalid acquisition time is greatly shortened, and the method has a positive effect on improving the user experience of pulse condition detection products.
Embodiments of the present invention will be described in detail below with reference to the accompanying drawings so that those skilled in the art can easily implement the embodiments of the present invention. The present invention may be embodied in many different forms and is not limited to the embodiments described herein.
Fig. 1 shows a schematic flow chart of a pulse data detection method according to an embodiment of the invention.
The method comprises the following steps:
step S11: and acquiring pulse condition data to be detected.
Optionally, the pulse condition data is fixed-length data; for example, in order to quickly recognize data, it is set to input raw data having a length of 4 seconds.
Optionally, a device with a sampling rate of 100sps and a channel number of 1 is used to sample the pulse condition signal, so that the dimension of the acquired data is (400, 1).
Step S12: and obtaining a pulse condition detection result corresponding to the data according to the pulse condition data based on the deployed pulse condition detection model.
In detail, the pulse condition detection result includes: pulse condition discrimination result and pulse rate monitoring result; and wherein the pulse condition discrimination result comprises: a valid pulse condition or a null pulse condition. The pulse condition detection model is constructed in a mode comprising the following steps: constructing a pulse condition detection framework model; training the pulse condition detection framework model by utilizing a training data set to obtain a pulse condition detection model; and wherein the training data set comprises: a valid training data set and an invalid training data set.
It should be noted that when obtaining invalid pulse condition results, the corresponding pulse rate monitoring results should be ignored, and the values thereof have no practical meaning.
Optionally, the valid training data set includes: a plurality of valid pulse condition data samples with valid classification labels and pulse rates corresponding to the valid pulse condition data samples; the invalid training data set comprises: a plurality of invalid pulse condition data samples having invalid classification labels.
Preferably, the valid pulse condition data and the invalid non-pulse condition data acquired in the past are marked separately, the valid pulse condition data mark classification label is 1, and the invalid data mark classification label is 0, for training the classification result. Meanwhile, in order to output the pulse rate in the model, the pulse rate in the data can be marked, but the non-pulse data has no pulse rate result, and the value of the data can be set as a value, such as 0, by default when the data is marked.
Optionally, since the valid pulse condition data may be at different pulse taking pressures and exhibit different amplitude heights, in order to ignore the influence of the pulse taking pressure and the valid peak-to-peak value and to make the trained model more robust, the pulse condition data is subjected to Z-Score normalization. That is, the effective pulse condition data and the ineffective pulse condition data are both subjected to Z-Score standardization processing to obtain data values after the training data set is processed as follows:
where mean (x) represents the mean of the input data x sequence, std (x) represents the standard deviation of the input data x sequence.
Optionally, as shown in fig. 2, the pulse condition detection framework model includes: a characteristic crude extraction structure 21 for performing crude characteristic extraction on the input pulse condition data to output pulse condition crude characteristic extraction data; the fine characteristic extraction structure 22 is connected with the coarse characteristic extraction structure and is used for performing fine characteristic extraction on the pulse condition coarse characteristic extraction data to output pulse condition fine characteristic extraction data; a pulse detection result output structure 23 connected to the fine feature extraction structure 22 for obtaining a pulse detection result corresponding to the pulse data based on the input fine feature extraction data; wherein, the pulse condition detection result output structure 23 includes: a data valid determination output structure 231 for extracting data based on the inputted fine pulse characteristics and outputting a valid pulse result or an invalid pulse result; and the pulse rate output structure 232 is used for extracting data and outputting a pulse rate monitoring result based on the input fine pulse condition characteristics.
Optionally, as shown in fig. 3, the coarse feature extraction structure includes: two connected feature coarse sub-extraction structures 311 and 312; inputting pulse condition data to be detected into the characteristic coarse sub-extraction structure 311 to output first coarse characteristic extraction data, and inputting the first coarse characteristic extraction data into the characteristic coarse sub-extraction structure 312 to output coarse characteristic extraction data; wherein, each coarse extraction structure of characteristics includes: a convolutional layer Conv for mapping data onto a feature space; a Dropout layer connected with the convolution layer and used for removing the training unit of the neural network from the network according to a certain probability, so that the activation value of the neuron pauses working, and the model generalization is stronger; preferably, the probability value for its removal is set to 0.4. And the Batchnormalization layer is connected with the Dropout layer and is used for normalizing the output value of the upper layer and solving the influence of transmission offset and increase of the input data.
Optionally, as shown in fig. 4, the feature fine extraction structure includes:
a branch feature extraction structure comprising: three branch structures 411, 412 and 413 corresponding to different convolution kernels, for respectively outputting corresponding branch feature data according to the input coarse feature extraction data; different convolution kernel sizes are set for extracting features on different scales.
Wherein each branch structure comprises: the three-layer connected ConvBNRelu _ x structure is used for carrying out three times of residual error calculation on the input coarse feature extraction data; the residual error structure is used for preventing the gradient disappearance problem after the network becomes deep; the global average pooling layer is connected with the ConvBNRelu _ x structure and is used for carrying out average processing on the data subjected to the three-time residual error processing and reducing the dimension of feature output;
and the fusion structure 42 is connected with the branch feature extraction structure and is used for fusing the branch feature data output by the three branch structures and outputting the fine feature extraction data.
Optionally, the sizes of convolution kernels adopted by the three branch structures are 3, 5 and 7 respectively.
Optionally, as shown in fig. 5, the residual structure adopted by each ConvBNRelu _ x structure includes: respectively, two interconnected structures with sequentially connected convolution layer, BatchNormalization layer, relu function layer, and a skip-connection residual structure layer that directly adds the input to the output.
Optionally, the activation function adopted by the data effective judgment output structure is a sigmoid function; the activation function adopted by the pulse rate output structure is a Relu function.
Optionally, the data valid determination output structure includes: two layers of connected Dense structures; and/or, the pulse rate output structure comprises: two layers of connected Dense structure.
Optionally, a prepared training data set is used for training the pulse condition detection framework model, so that parameters of the model are automatically calculated by a computer, an output result is similar to a marked result as much as possible, and an Adam optimizer is used as the optimizer for model training to train the model;
preferably, for setting the model training parameters, an Adam optimizer can be used as the optimizer for model training to train 300 epoch. The initial learning rate was set at 0.0001 and the shrinkage was 0.1 times after 200 epochs.
Optionally, the pulse condition detection framework model further includes: the loss calculation layer is connected with the pulse condition detection result output structure and is used for carrying out loss weighted calculation on the output pulse condition detection result;
the lossy computation layer includes: the first loss weighting calculation module is used for performing loss weighting calculation on the effective pulse condition result or the ineffective pulse condition result by adopting a binary cross entropy loss function; and the second loss weighting calculation module is used for performing loss weighting calculation on the effective pulse rate or the ineffective pulse rate by adopting a Mask-MSE loss function.
Optionally, the MSE loss function is a Mask-MSE loss function, and the Mask-MSE loss function includes:
the binary cross entropy loss function is:
considering that the pulse rate value marked in the invalid data is 0, great disturbance can be caused to the loss value in the training process, and during calculation, the classification label is used as a mask to further calculate the loss value, so that the pulse rate output of the invalid signal does not influence the model training process. Namely, the parameter alpha is used for controlling the correlation between classification loss and pulse rate loss; namely, the model loss calculation formula of the loss weighting calculation layer is as follows:
loss＝αloss(cls)+(1-α)loss(hr)； (4)
optionally, the pulse condition detection model is deployed to an application end, that is, the trained pulse condition detection model is deployed to the application end; although the training of the artificial neural network model needs to be carried out on GPU equipment, the application reasoning process is simpler than the training process, the requirement on the equipment is lower than the training process, the calculated amount of the model parameters designed at this time can be reasoned on general CPU equipment, and the operation efficiency of the model parameters at the application end is better.
Therefore, the pulse condition detection model is light and convenient, can be deployed on mobile terminal equipment, does not need to use a highly-configured cloud server, does not need network support, and is not limited by the network and other reasons in a use scene.
Optionally, model inference is performed on the pulse condition detection model deployed to the application end, that is, the pulse condition signal acquisition device of the application end acquires an element signal, when the data length satisfies the model input length, a newly acquired data segment can be intercepted, Z-Score standardization is performed on the data according to the same way of processing a data set, and then the data segment is sent to the model for inference, so that a classification result and a pulse rate result are respectively acquired at two output nodes of the model. Firstly, it should be determined whether the result at the classification node is a valid pulse signal, if so, the result at the pulse rate node is continuously read, otherwise, the output value of the pulse rate node should be ignored, and the value has no actual meaning.
In order to better explain the pulse condition data detection method, the following specific examples are provided for illustration.
Example 1: fig. 6 is a schematic flow chart of the pulse condition data detection method.
The method comprises the following steps:
step 1: processing the data set;
the main framework of the technical scheme is carried out by using a mode of training a model by an artificial neural network, the training process can not be separated from the support of a data set, and a successful neural network model can not be separated from a good data set. Therefore, the correct pulse condition data and the incorrect non-pulse condition data acquired in the past are marked separately, the correct pulse condition data mark classification label is 1, the incorrect data mark classification label is 0, and the training is used for training the classification result. Meanwhile, in order to output the pulse rate in the model, the pulse rate result in the data segment should also be marked, and the non-pulse condition data has no pulse rate result, and the default value is set to 0 when the data is marked. For valid pulse condition data, the data are subjected to Z-Score normalization in order to ignore the influence of the pulse taking pressure and the valid peak-to-peak value and to make the trained model more robust, since the data may exhibit different amplitude heights under different pulse taking pressures. The calculation formula is as follows:
where mean (x) represents the mean of the input data x sequence, std (x) represents the standard deviation of the input data x sequence.
Step 2: building a model: constructing a pulse condition detection framework model, and as shown in fig. 7, showing a structural schematic diagram of the pulse condition detection framework model;
wherein, the pulse condition detection framework model comprises:
and the Conv + Dropout + BN structures are used for carrying out characteristic rough extraction, and each Conv + Dropout + BN structure comprises a convolution layer + a Dropout layer + a BatchNormalization layer. The data are mapped to the feature space through the convolutional layer, and the training unit of the neural network is removed from the network according to a certain probability by using the Dropout layer, so that the activation value of the neuron of the training unit is suspended, the generalization of the model can be stronger, and the probability value of the removal is set to be 0.4. And the Batchnormalization normalizes the output value of the upper layer, and solves the influence of the transmission offset and increase of the input data.
Three branched ConvBNRelu _ x structures side by side, wherein a residual structure is used in each ConvBNRelu _ x structure. Each branch has 3 levels, the sizes of convolution kernels in the branches are 3, 5 and 7 respectively, and different sizes of convolution kernels are set so as to extract features on different scales. The residual structure is used to prevent the gradient disappearance problem after the network has deepened. And after three times of residual processing, averaging the output elements by using a global average pooling layer GlobavalePool, and reducing the dimensionality of characteristic output.
And the fusion structure fuses the branch characteristic data output by the three branch structures.
The pulse condition detection result output structure has two tasks, one is used as a classification result to judge whether data is effective pulse condition data, and the other is used as an output node for performing a pulse rate identification task to output a specific pulse rate value; namely, the method comprises the following steps:
the data valid judging output structure comprises: the two layers of connected Dense structures are used for extracting data and outputting effective pulse condition results or ineffective pulse condition results based on the input fine pulse condition features; a pulse rate output structure comprising: the two layers of connected Dense structures are used for extracting data and outputting effective pulse rate or ineffective pulse rate based on the input fine pulse characteristic; the output structure of the pulse condition detection result uses a Sigmoid function as an activation function, and the result can be regarded as a probability value of whether the pulse condition data is the pulse condition data or not. The pulse rate output structure uses the Relu function as an activation function.
And a loss weighting calculation layer, wherein the output layer of the pulse condition signal classification result uses binary cross entropy as a loss function, and the pulse rate identification result uses Mask-MSE to calculate an error value.
And step 3: training a network model;
the training of the model is performed using the prepared data set, so that the computer automatically calculates the parameters of the model to make the output result as close as possible to the labeled result. In the model training, an Adam optimizer is used as an optimizer for model training to train 300 epoch. The initial learning rate was set at 0.0001 and the shrinkage was 0.1 times after 200 epochs. The output layer of the pulse condition signal classification result uses binary cross entropy as a loss function, and the pulse rate identification result uses Mask-MSE to calculate an error value.
4. Deploying the model to an application end;
deploying the trained network model to an application end; although the training of the artificial neural network model needs to be carried out on GPU equipment, the application reasoning process is simpler than the training process, and the requirement on the equipment is not as high as the training process. The calculated amount of the model parameters designed at this time can be inferred on common CPU equipment, and the operation efficiency is better at the mobile end;
5. model reasoning;
the pulse signal acquisition equipment at the application end acquires element signals, can intercept newly acquired data fragments when the data length meets the input length of the model, standardizes the data in a Z-Score mode according to the same mode of processing a data set, then sends the data fragments to the model for reasoning, and respectively acquires a classification result and a pulse rate result on two output nodes of the model. Firstly, it should be determined whether the result at the classification node is a valid pulse signal, if so, the result at the pulse rate node is continuously read, otherwise, the output value of the pulse rate node should be ignored, and the value has no actual meaning.
6. Detecting pulse condition data;
and inputting the acquired pulse condition data to be detected into the constructed model and acquiring a pulse condition detection result corresponding to the data.
The model of this embodiment can carry out the analysis whether for valid data when pulse condition check out test set detects, can withdraw from invalid acquisition process in advance, shortens invalid acquisition time greatly, has the positive effect on improving the user experience of pulse condition detection class product. The model is light and convenient, can be deployed on mobile terminal equipment, does not need to use a highly-configured cloud server, does not need network support, and is not limited by the network and other reasons in a use scene. After the model provided by the invention is trained by data, the classification accuracy rate exceeds 97%, the average absolute pulse rate error is 0.18bpm, and samples with pulse rate errors exceeding +/-3 bpm only account for 6.5 ten thousandths.
Similar to the principle of the above embodiments, the present invention provides a pulse condition data detection system.
Specific embodiments are provided below in conjunction with the attached figures:
fig. 8 is a schematic structural diagram of a pulse data detection system according to an embodiment of the present invention.
The system comprises:
the data acquisition module 81 is used for acquiring pulse condition data to be detected;
the pulse condition detection module 82 is connected with the data acquisition module 81 and is used for acquiring a pulse condition detection result corresponding to the data according to the pulse condition data based on the deployed pulse condition detection model; wherein, the pulse condition detection result comprises: pulse condition discrimination result and pulse rate monitoring result; and wherein the pulse condition discrimination result comprises: a valid or invalid pulse condition result;
the pulse condition detection model is constructed in a mode comprising the following steps:
constructing a pulse condition detection framework model;
training the pulse condition detection framework model by utilizing a training data set to obtain a pulse condition detection model; and wherein the training data set comprises: a valid training data set and an invalid training data set.
It should be noted that the division of each module in the system embodiment of fig. 8 is only a division of a logical function, and all or part of the actual implementation may be integrated into one physical entity or may be physically separated. And these modules can be realized in the form of software called by processing element; or may be implemented entirely in hardware; part of the modules can be realized in a software calling mode through a processing element, and part of the modules can be realized in a hardware mode;
for example, the modules may be one or more integrated circuits configured to implement the above methods, such as: one or more Application Specific Integrated Circuits (ASICs), or one or more microprocessors (DSPs), or one or more Field Programmable Gate Arrays (FPGAs), among others. For another example, when one of the above modules is implemented in the form of a Processing element scheduler code, the Processing element may be a general-purpose processor, such as a Central Processing Unit (CPU) or other processor capable of calling program code. As another example, these modules may be integrated together and implemented in the form of a system-on-a-chip (SOC).
It should be noted that, since the structures and the implementation principles of the pulse condition data detection system and the pulse condition detection model have been described in the foregoing embodiments, repeated descriptions are omitted here.
In summary, the pulse condition data detection method and system of the application obtain the pulse condition discrimination result and the pulse rate monitoring result corresponding to the pulse condition data to be detected through the constructed pulse condition detection model. The scheme of the invention can analyze whether the pulse condition is valid data or not while detecting the pulse condition, and predict the numerical value of the pulse rate under the condition of judging the correct pulse condition signal; and the invalid acquisition process can be quitted in advance, so that the invalid acquisition time is greatly shortened, and the method has a positive effect on improving the user experience of pulse condition detection products.
The above embodiments are merely illustrative of the principles and utilities of the present application and are not intended to limit the application. Any person skilled in the art can modify or change the above-described embodiments without departing from the spirit and scope of the present application. Accordingly, it is intended that all equivalent modifications or changes which can be made by those skilled in the art without departing from the spirit and technical concepts disclosed in the present application shall be covered by the claims of the present application.
Claims (10)
1. A pulse condition data detection method is characterized by comprising the following steps:
acquiring pulse condition data to be detected;
based on the deployed pulse condition detection model, obtaining a pulse condition detection result corresponding to the data according to the pulse condition data; wherein, the pulse condition detection result comprises: pulse condition discrimination result and pulse rate monitoring result; and wherein the pulse condition discrimination result comprises: a valid or invalid pulse condition result;
the pulse condition detection model is constructed in a mode comprising the following steps:
constructing a pulse condition detection framework model;
training the pulse condition detection framework model by utilizing a training data set to obtain a pulse condition detection model; and wherein the training data set comprises: a valid training data set and an invalid training data set.
2. The method of claim 1, wherein the valid training data set comprises: a plurality of valid pulse condition data samples with valid classification labels and pulse rates corresponding to the samples; the invalid training data set comprises: a plurality of invalid pulse condition data samples having invalid classification labels.
3. The method of claim 2, wherein the training data set is obtained by Z-Score normalization.
4. The method for detecting pulse condition data according to claim 1, wherein said pulse condition detection framework model comprises:
the characteristic crude extraction structure is used for carrying out crude characteristic extraction on the input pulse condition data so as to output pulse condition crude characteristic extraction data;
the characteristic fine extraction structure is connected with the characteristic crude extraction structure and is used for performing fine characteristic extraction on the pulse condition coarse characteristic extraction data to output pulse condition fine characteristic extraction data;
the pulse condition detection result output structure is connected with the characteristic fine extraction structure and is used for obtaining a pulse condition detection result corresponding to the pulse condition data based on the input fine characteristic extraction data;
wherein, the pulse condition detection result output structure comprises:
the data validity judging output structure is used for extracting data based on the input fine pulse characteristics and outputting a valid pulse condition result or an invalid pulse condition result;
and the pulse rate output structure is used for extracting data and outputting a pulse rate monitoring result based on the input fine pulse condition characteristics.
5. The pulse condition data detecting method according to claim 4, wherein the data valid decision output structure comprises: two layers of connected Dense structures; and/or, the pulse rate output structure comprises: two layers of connected Dense structure.
6. The pulse condition data detection method according to claim 4, wherein the characteristic crude extraction structure comprises: two connected feature rough extraction structures;
wherein, each coarse extraction structure of characteristics includes: the multilayer structure comprises a winding layer, a Dropout layer connected with the winding layer and a BatchNormalization layer connected with the Dropout layer.
7. The pulse condition data detecting method as claimed in claim 4, wherein said feature extraction structure comprises:
a branch feature extraction structure comprising: three branch structures corresponding to different convolution kernels and used for respectively outputting corresponding branch characteristic data according to input coarse characteristic extraction data; wherein each branch structure comprises: the three-layer connected ConvBNRelu _ x structure is used for carrying out three times of residual error calculation on the input coarse feature extraction data; the global average pooling layer is used for carrying out average processing on the data subjected to the three-time residual error processing;
and the fusion structure is connected with the branch feature extraction structure and used for fusing the branch feature data output by the three branch structures and outputting the fine feature extraction data.
8. The pulse condition data detection method according to claim 4, wherein the activation function adopted by the data valid judgment output structure is a sigmoid function; the activation function adopted by the pulse rate output structure is a Relu function.
9. The method of claim 1, wherein the pulse detection framework model further comprises:
the loss weighting calculation layer is connected with the pulse condition detection result output structure and comprises:
the first loss weighting calculation module is used for performing loss weighting calculation on the pulse condition judgment result by adopting a binary cross entropy loss function;
and the second loss weighting calculation module is used for performing loss weighting calculation on the pulse rate monitoring result by adopting an MSE loss function.
10. A pulse condition data detection system, comprising:
the data acquisition module is used for acquiring pulse condition data to be detected;
the pulse condition detection module is connected with the data acquisition module and used for acquiring a pulse condition detection result corresponding to the data according to the pulse condition data based on the deployed pulse condition detection model; wherein, the pulse condition detection result comprises: pulse condition discrimination result and pulse rate monitoring result; and wherein the pulse condition discrimination result comprises: a valid pulse condition result or a non-valid pulse condition result;
the pulse condition detection model is constructed in a mode comprising the following steps:
constructing a pulse condition detection framework model;
training the pulse condition detection framework model by utilizing a training data set to obtain a pulse condition detection model; and wherein the training data set comprises: a valid training data set and an invalid training data set.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210102117.5A CN114469017A (en) | 2022-01-27 | 2022-01-27 | Pulse condition data detection method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210102117.5A CN114469017A (en) | 2022-01-27 | 2022-01-27 | Pulse condition data detection method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114469017A true CN114469017A (en) | 2022-05-13 |
Family
ID=81477280
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210102117.5A Pending CN114469017A (en) | 2022-01-27 | 2022-01-27 | Pulse condition data detection method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114469017A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114818838A (en) * | 2022-06-30 | 2022-07-29 | 中国科学院国家空间科学中心 | Low signal-to-noise ratio moving point target detection method based on pixel time domain distribution learning |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107468221A (en) * | 2017-09-12 | 2017-12-15 | 济南大学 | Pulse wave signal processing method and lung cancer detection system based on Jin Shi pulse studies |
| CN109313937A (en) * | 2017-09-11 | 2019-02-05 | 深圳市得道健康管理有限公司 | Chinese medicine pulse assistant diagnosis system and method based on cloud computing platform |
| CN109567763A (en) * | 2018-11-29 | 2019-04-05 | 中国科学院长春光学精密机械与物理研究所 | A kind of pulse condition detection device |
| US20190110703A1 (en) * | 2017-10-12 | 2019-04-18 | Kwangjae LEE | Pulse diagnosis apparatus and pulse diagnosis method thereof |
| CN109730656A (en) * | 2019-01-09 | 2019-05-10 | 中国科学院苏州纳米技术与纳米仿生研究所 | Nerve network system, computer equipment for pulse wave signal classification |
| CN110956197A (en) * | 2019-10-28 | 2020-04-03 | 新绎健康科技有限公司 | Method and system for establishing pulse wave noise signal identification model based on convolutional neural network |
| CN112183354A (en) * | 2020-09-29 | 2021-01-05 | 上海中医药大学 | Single-period pulse wave signal quality evaluation method based on support vector machine |
| CN112472048A (en) * | 2021-01-15 | 2021-03-12 | 郑州大学 | Neural network structure for pulse condition recognition of cardiovascular disease patient |
| CN113558584A (en) * | 2021-06-22 | 2021-10-29 | 深圳市大数据研究院 | Pulse wave preprocessing method based on signal quality evaluation |
-
2022
- 2022-01-27 CN CN202210102117.5A patent/CN114469017A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109313937A (en) * | 2017-09-11 | 2019-02-05 | 深圳市得道健康管理有限公司 | Chinese medicine pulse assistant diagnosis system and method based on cloud computing platform |
| CN107468221A (en) * | 2017-09-12 | 2017-12-15 | 济南大学 | Pulse wave signal processing method and lung cancer detection system based on Jin Shi pulse studies |
| US20190110703A1 (en) * | 2017-10-12 | 2019-04-18 | Kwangjae LEE | Pulse diagnosis apparatus and pulse diagnosis method thereof |
| CN109567763A (en) * | 2018-11-29 | 2019-04-05 | 中国科学院长春光学精密机械与物理研究所 | A kind of pulse condition detection device |
| CN109730656A (en) * | 2019-01-09 | 2019-05-10 | 中国科学院苏州纳米技术与纳米仿生研究所 | Nerve network system, computer equipment for pulse wave signal classification |
| CN110956197A (en) * | 2019-10-28 | 2020-04-03 | 新绎健康科技有限公司 | Method and system for establishing pulse wave noise signal identification model based on convolutional neural network |
| CN112183354A (en) * | 2020-09-29 | 2021-01-05 | 上海中医药大学 | Single-period pulse wave signal quality evaluation method based on support vector machine |
| CN112472048A (en) * | 2021-01-15 | 2021-03-12 | 郑州大学 | Neural network structure for pulse condition recognition of cardiovascular disease patient |
| CN113558584A (en) * | 2021-06-22 | 2021-10-29 | 深圳市大数据研究院 | Pulse wave preprocessing method based on signal quality evaluation |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114818838A (en) * | 2022-06-30 | 2022-07-29 | 中国科学院国家空间科学中心 | Low signal-to-noise ratio moving point target detection method based on pixel time domain distribution learning |
| CN114818838B (en) * | 2022-06-30 | 2022-09-13 | 中国科学院国家空间科学中心 | Low signal-to-noise ratio moving point target detection method based on pixel time domain distribution learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Oo et al. | Plant leaf disease detection and classification using image processing | |
| CN109003279B (en) | Fundus retina blood vessel segmentation method and system based on K-Means clustering labeling and naive Bayes model | |
| Aquino et al. | A new methodology for estimating the grapevine-berry number per cluster using image analysis | |
| CN108932479A (en) | A kind of human body anomaly detection method | |
| CN112294341B (en) | Sleep electroencephalogram spindle wave identification method and system based on light convolutional neural network | |
| US11435719B2 (en) | System and method for identifying manufacturing defects | |
| CN111738174B (en) | Human body example analysis method and system based on depth decoupling | |
| CN104809450B (en) | Wrist vena identification system based on online extreme learning machine | |
| Ganeshbabu | Glaucoma image classification using discrete orthogonal stockwell transform | |
| Prasetio et al. | Support vector slant binary tree architecture for facial stress recognition based on Gabor and HOG Feature | |
| Ahmadi et al. | Iris recognition system based on canny and LoG edge detection methods | |
| Wang et al. | SIFT based vein recognition models: analysis and improvement | |
| CN114469017A (en) | Pulse condition data detection method and system | |
| Bhagya et al. | Analysis of image segmentation algorithms for the effective detection of leukemic cells | |
| Sankaran et al. | Plant disease detection and recognition using K means clustering | |
| Ponni Bala et al. | Early detection and classification of microaneurysms in retinal fundus images using sequential learning methods | |
| KR20080079798A (en) | Method of face detection and recognition | |
| CN116821790A (en) | Freezing gait prediction method based on self-adaptive enhanced integrated learning strategy | |
| Gajanan et al. | Android based plant disease identification system using feature extraction technique | |
| Thiyaneswaran et al. | Iris Recognition using Left and Right Iris Feature of the Human Eye for Biometric Security System | |
| Zabihi et al. | Vessel extraction of conjunctival images using LBPs and ANFIS | |
| CN115424093A (en) | Method and device for identifying cells in fundus image | |
| Sadreazami et al. | EEGCAPs: Brain activity recognition using modified common spatial patterns and capsule network | |
| Mitrokhin et al. | Deep learning approach for QRS wave detection in ECG monitoring | |
| Al-Otaibi et al. | Statistical Image Analysis Based Automated Leaves Classification. |
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 |