CN114366047A - Multitask neural network pulse condition data processing method, system and terminal - Google Patents
Multitask neural network pulse condition data processing method, system and terminal Download PDFInfo
- Publication number
- CN114366047A CN114366047A CN202210102125.XA CN202210102125A CN114366047A CN 114366047 A CN114366047 A CN 114366047A CN 202210102125 A CN202210102125 A CN 202210102125A CN 114366047 A CN114366047 A CN 114366047A
- Authority
- CN
- China
- Prior art keywords
- pulse
- pulse condition
- data
- identification
- identification result
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- 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
-
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- 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
Landscapes
- Health & Medical Sciences (AREA)
- Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Physics & Mathematics (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Theoretical Computer Science (AREA)
- Medical Informatics (AREA)
- Animal Behavior & Ethology (AREA)
- Public Health (AREA)
- Heart & Thoracic Surgery (AREA)
- Artificial Intelligence (AREA)
- Pathology (AREA)
- Surgery (AREA)
- Veterinary Medicine (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Physiology (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Psychiatry (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Cardiology (AREA)
- Signal Processing (AREA)
- Alternative & Traditional Medicine (AREA)
- Image Analysis (AREA)
Abstract
According to the multitask neural network pulse condition data processing method, the multitask neural network pulse condition data processing system and the multitask neural network pulse condition data processing terminal, a pulse rate identification result, a rhythm identification result, a pulse fluency identification result and a pulse tension identification result corresponding to a pulse condition data segment are obtained according to the constructed multitask pulse condition signal identification model. The scheme of the invention can subdivide the result of the pulse condition signal from a single element, not only output the pulse condition name, more accord with the theory of traditional Chinese medicine, and can obtain the high-accuracy pulse rate identification result, rhythm identification result, pulse condition fluency identification result and pulse condition tension identification result.
Description
Technical Field
The application relates to the technical field of data processing, in particular to a method, a system and a terminal for processing pulse condition data of a multitask neural network.
Background
Regarding the types of pulse conditions, different factors in the medical works such as Nei Jing and frequent lake pulmonology are classified into different categories. Among the pulse conditions listed by doctors in all ages, the pulse conditions are divided into four aspects of position, number, shape and potential by Zhou scholarly of late Qing physicians, which is a way to explain the high acceptance at present. The concrete description is that for the "bit", the floating and sinking size is also; slow and rapid in the case of several counts; the shape is just as the linear face of an arithmetic physician; for those with the tendency of contraction, expansion and contraction, there are preponderance and decline. Many simple pulses are not really a simple pulse attribute, but a plurality of single elements (bit attribute, number attribute, shape attribute, potential attribute) are recombined according to different proportions.
For the identification of pulse condition data, the current widely processing mode is to identify the time domain, frequency domain or time-frequency domain characteristic points, and classify by means of the characteristic value judgment mode of an expert system. Meanwhile, the time and frequency domain feature points of some pulse condition signals are not obvious, or there are multiple expression modes, so the identification mode of the feature points is not suitable for all signals. And the obtained classification result is only the name of the pulse condition, such as the flat pulse, the smooth pulse, the string pulse and the like, and the subdivision content of the pulse condition on a single pulse condition element is lacked.
Disclosure of Invention
In view of the above drawbacks of the prior art, the present application aims to provide a method, a system and a terminal for processing pulse condition data of a multi-tasking neural network, so as to solve the problem that the pulse condition data identification processing method in the prior art lacks the content of pulse condition subdivision on a single pulse condition element.
To achieve the above and other related objects, the present application provides a pulse data processing method for a multi-tasking neural network, comprising: acquiring a pulse condition data segment to be identified; based on the constructed multi-task pulse condition signal identification model, obtaining a multi-element pulse condition identification result corresponding to the pulse condition data segment according to the pulse condition data segment; wherein the multi-element recognition result comprises: a pulse rate recognition result, a rhythm recognition result, a pulse condition fluency recognition result and a pulse condition tension recognition result; the construction method of the multitask pulse condition signal identification model comprises the following steps: constructing a pulse signal identification frame model; training the pulse condition signal identification framework model by utilizing a training data set to obtain a multi-task pulse condition signal identification model; and wherein the training data set comprises: a plurality of pulse condition data segment samples and a multi-element pulse condition identification result corresponding to each pulse condition data segment sample.
In one or more embodiments of the present application, the pulse condition signal recognition framework model includes: the characteristic crude extraction structure is used for carrying out crude characteristic extraction on the input pulse condition data fragments so as to output 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 crude characteristic extraction data to output fine characteristic extraction data; the multi-element result output structure is connected with the fine feature extraction structure and is used for obtaining a multi-element identification result corresponding to the pulse condition data segment based on the input fine feature extraction data; wherein the multi-element result output structure comprises: the pulse rate identification result output structure is used for extracting data and outputting a pulse rate identification result based on the input fine characteristics; a rhythm recognition result output structure for outputting a rhythm recognition result based on the input fine feature extraction data; the pulse fluency identification result output structure is used for extracting data and outputting a pulse fluency identification result based on the input fine characteristics; and the pulse tension recognition result output structure is used for extracting data based on the input fine characteristics and outputting a pulse tension recognition result.
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 ResBlock1D structure is used for carrying out three times of residual error calculation on 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 pulse rate recognition result output structure is a relu function; the activation function adopted by the rhythm identification result output structure is a sigmoid function; the activation function adopted by the pulse fluency identification result output structure is a softmax function; and the activation function adopted by the pulse tension recognition result output structure is a sigmoid function.
In one or more embodiments of the present application, the multi-element result output structure includes: a pulse rate recognition result output structure for extracting data based on the input fine features and outputting pulse rate values in units of sub-minute; the rhythm identification result output structure is used for extracting data and outputting rhythm two-classification results based on the input fine characteristics; wherein the rhythm binary classification result comprises: corresponding to the result of regular or irregular rhythm; the pulse fluency identification result output structure is used for extracting data and outputting pulse fluency three-classification results based on the input fine characteristics; wherein the three classification results of pulse fluency degree comprise: corresponding to one of the slippery pulse, the unsmooth pulse and the unsmooth pulse; the pulse tension recognition result output structure is used for extracting data and outputting pulse tension two-classification results based on the input fine characteristics; wherein, the pulse tension binary classification result comprises: corresponding to a wiry or even pulse.
In one or more embodiments of the present application, the pulse signal recognition framework model further includes: a loss weighting calculation layer connected to the multi-factor result output structure, comprising: the first loss weighting calculation module is used for performing loss weighting calculation on the pulse rate identification result by adopting an MSE function; and the second loss weighting calculation module is used for performing loss weighting calculation on the rhythm identification result, the pulse condition fluency identification result and the pulse condition nervousness identification result by adopting a weighted cross entropy loss function.
In one or more embodiments of the present application, the pulse data segments are obtained by resampling the pulse signal.
To achieve the above and other related objects, the present application provides a pulse data processing system of a multitasking neural network, comprising: the data acquisition module is used for acquiring pulse condition data segments to be identified; the identification module is connected with the data acquisition module and used for acquiring a multi-element pulse condition identification result corresponding to the pulse condition data segment based on the constructed multi-task pulse condition signal identification model; wherein the multi-element recognition result comprises: a pulse rate recognition result, a rhythm recognition result, a pulse condition fluency recognition result and a pulse condition tension recognition result; the construction method of the multitask pulse condition signal identification model comprises the following steps: constructing a pulse signal identification frame model; training the pulse condition signal identification framework model by utilizing a training data set to obtain a multi-task pulse condition signal identification model; and wherein the training data set comprises: a plurality of pulse condition data segment samples and a multi-element pulse condition identification result corresponding to each pulse condition data segment sample.
To achieve the above and other related objects, the present application provides a multitask neural network pulse data processing terminal, including: one or more memories and one or more processors; the one or more memories for storing a computer program; the one or more processors are connected with the memory and used for operating the computer program to execute the multitask neural network pulse condition data processing method.
As described above, according to the multitask neural network pulse condition data processing method, the multitask neural network pulse condition data processing system and the terminal, based on the constructed multitask pulse condition signal identification model, a pulse rate identification result, a rhythm identification result, a pulse fluency identification result and a pulse tension identification result corresponding to the segment are obtained according to the pulse condition data segment. The scheme of the invention can subdivide the result of the pulse condition signal from a single element, not only output the pulse condition name, more accord with the theory of traditional Chinese medicine, and can obtain the high-accuracy pulse rate identification result, rhythm identification result, pulse condition fluency identification result and pulse condition tension identification result.
Drawings
Fig. 1 is a schematic flowchart illustrating a method for processing pulse data of a multitasking neural network according to an embodiment of the present application.
Fig. 2 is a schematic structural diagram of a pulse condition signal recognition 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 the structure of ResBlock1D in the embodiment of the present application.
Fig. 6 is a schematic structural diagram of a pulse condition signal recognition framework model in an embodiment of the present application.
FIG. 7 is a block diagram of a pulse data processing system of a multitasking neural network according to an embodiment of the present application.
Fig. 8 is a schematic structural diagram of a terminal of a pulse data processing side of a multitasking neural network in 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 referred to as being "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 term "comprises/comprising" when used in this specification is taken to specify the presence of stated features, regions, integers, steps, operations, elements, and/or components, but does not exclude the presence or addition of other features, regions, integers, steps, operations, elements, and/or components.
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 defects of the prior art, based on the constructed multitask pulse condition signal identification model, a pulse rate identification result, a rhythm identification result, a pulse fluency identification result and a pulse tension identification result corresponding to the segment are obtained according to the pulse condition data segment. The scheme of the invention can subdivide the result of the pulse condition signal from a single element, not only output the pulse condition name, more accord with the theory of traditional Chinese medicine, and can obtain the high-accuracy pulse rate identification result, rhythm identification result, pulse condition fluency identification result and pulse condition tension identification result.
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 is a schematic flow chart showing a pulse data processing method of a multitask neural network according to an embodiment of the present invention.
The method comprises the following steps:
step S11: and acquiring a pulse condition data segment to be identified.
Optionally, the pulse condition data segment is data with a certain length; for example, since a plurality of pulse periods are required for determining the rhythm of the pulse signal, the input length is set to 18 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 dimensionality of the acquired data is (1800, 1).
Optionally, since the pulse rate of the pulse condition signal is a specific value in units of sub-minutes, the value is between 40 and 200, and the pulse rate of the detected data is generally distributed in a more central position, and pulse rate results at two ends are less obtained and are distributed unevenly; therefore, it is necessary to resample the pulse signal to obtain the pulse data segment. Preferably, the resampling parameter is set to 1-1.3, and the length of 1800samples is randomly truncated after resampling is completed. By adopting the resampling mode, the robustness of pulse rate output can be greatly improved, and the judgment results of rhythm and pulse condition can not be influenced, so that the multi-task pulse condition signal identification model can simultaneously output a pulse rate identification result, a rhythm identification result, a pulse condition fluency identification result and a pulse condition tensity identification result.
Step S12: and based on the constructed multi-task pulse condition signal identification model, obtaining a multi-element pulse condition identification result corresponding to the segment according to the pulse condition data segment.
In detail, the multi-element recognition result includes: a pulse rate recognition result, a rhythm recognition result, a pulse condition fluency recognition result and a pulse condition tension recognition result. The construction method of the multitask pulse condition signal identification model comprises the following steps: constructing a pulse signal identification frame model; training the pulse condition signal identification framework model by utilizing a training data set to obtain a multi-task pulse condition signal identification model; and wherein the training data set comprises: a plurality of pulse condition data segment samples and a multi-element pulse condition identification result corresponding to each pulse condition data segment sample.
Optionally, as shown in fig. 2, the pulse condition signal recognition framework model includes: a characteristic crude extraction structure 21 for performing crude characteristic extraction on the input pulse condition data segment to output crude characteristic extraction data; the characteristic fine extraction structure 22 is connected with the characteristic coarse extraction structure and is used for performing fine characteristic extraction on the coarse characteristic extraction data to output fine characteristic extraction data; a multi-element result output structure 23 connected to the fine feature extraction structure 22 for obtaining a multi-element recognition result corresponding to the pulse condition data segment based on the input fine feature extraction data; wherein, the multi-element result output structure 23 includes: a pulse rate recognition result output structure 231 for outputting a pulse rate recognition result based on the input fine feature extraction data; a rhythm recognition result output structure 232 for extracting data based on the input fine features and outputting a rhythm recognition result; a pulse fluency identification result output structure 233 for outputting a pulse fluency identification result based on the input fine feature extraction data; and a pulse tension recognition result output structure 234 for outputting a pulse tension recognition result based on the input fine feature extraction data.
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 segments 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 ResBlock1D structure is used for carrying out three times of residual error calculation on 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 ResBlock1D 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 characteristic 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, a residual structure adopted by each ResBlock1D structure includes: two interconnected structures with sequentially connected convolution layer, BatchNormal layer, relu function layer, respectively, and a skip-connected residual structure layer that directly adds input to output.
Optionally, the activation function adopted by the pulse rate recognition result output structure is a relu function; the activation function adopted by the rhythm identification result output structure is a sigmoid function; the activation function adopted by the pulse fluency identification result output structure is a softmax function; and the activation function adopted by the pulse tension recognition result output structure is a sigmoid function.
Optionally, the multi-element result output structure includes: a pulse rate recognition result output structure for extracting data based on the input fine features and outputting pulse rate values in units of sub-minute; the rhythm identification result output structure is used for extracting data and outputting rhythm two-classification results based on the input fine characteristics; wherein the rhythm binary classification result comprises: corresponding to the result of regular or irregular rhythm; for example, output is uniform or non-uniform; the pulse fluency identification result output structure is used for extracting data and outputting pulse fluency three-classification results based on the input fine characteristics; wherein the three classification results of pulse fluency degree comprise: corresponding to one of the slippery pulse, the unsmooth pulse and the unsmooth pulse; for example, outputting one of a slip, astringent, and no-slip feature; the pulse tension recognition result output structure is used for extracting data and outputting pulse tension two-classification results based on the input fine characteristics; wherein, the pulse tension binary classification result comprises: corresponding to a wiry or even pulse; for example, output chord or flat.
Optionally, the prepared data set is used for training a pulse condition signal recognition 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 the acquisition of the training data set, the pulse condition signal with the time length of 20 seconds can be intercepted each time, and the data of the middle 18 seconds can be randomly intercepted to be used as the training data to prepare the training set. For the setting of the model training parameters, an Adam optimizer can be used as an optimizer for model training to train 700epoch, the initial setting learning rate is 0.0001, and the reduction is 0.1 times after 300 epoch.
Optionally, the pulse signal recognition framework model further includes: the loss calculation layer is connected with the multi-element result output structure and is used for carrying out weighted calculation on the loss of the output multi-element recognition result;
the loss calculation layer includes:
the first loss calculation module is used for performing loss calculation on the pulse rate identification result by adopting an MSE function;
because the actual pulse condition data has the problem of data imbalance, for example, in the dimension of rhythm, the data with regular rhythm is much more than the data with irregular rhythm, in order to enable the model to focus attention on the category of few samples on the basis of accurate output, the sample size weighting is carried out on the output loss value to form a loss function of weighted cross entropy; namely, the loss calculation layer further includes: the second loss calculation module is used for performing loss calculation on the rhythm identification result, the pulse condition fluency identification result and the pulse condition nervousness identification result by adopting a weighted cross entropy loss function;
i.e. the weighted cross entropy loss function is:
wherein, ω is(i)The weight is the inverse of the class ratio.
In addition, control parameters may also be added to the loss term of the model loss function in order to balance the loss values of the various sections.
I.e. the model loss function:
wherein alpha isiFor each control parameter.
In the preferred embodiment, the control parameters of the pulse rate recognition result, the rhythm recognition result, the pulse fluency recognition result and the pulse tension recognition result are 0.5, 1, 8 and 4, respectively.
It should be noted that, if the loss weighting calculation is combined with the signal resampling, the robustness of the model can be further improved, and the over-fitting phenomenon can be prevented. In order to better describe the pulse condition data processing method of the multitask neural network, a specific embodiment is provided;
example 1: a multitask neural network pulse condition data processing method comprises the following steps:
1. building a model: constructing a pulse signal identification frame model, and as shown in fig. 6, showing a structural schematic diagram of the pulse signal identification frame model;
wherein, the pulse signal identification frame 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 ResBlock1D structures side by side, with a residual structure used in each ResBlock1D 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 four tasks of the model are respectively the output of the pulse rate, and because the pulse rate is a specific numerical value taking each minute as a unit, relu is suitable to be used as an activation function of an output layer; the rhythm classification task outputs uniform/irregular classification results and applies a sigmoid function as an activation function of an output layer; classifying the pulse conditions (fluency), outputting three classification results with smooth/astringent/no characteristics, and applying a softmax function as an activation function of an output layer; the output of the pulse condition (tensity) is a chord/flat binary classification result, and a sigmoid function is applied as an activation function of an output layer.
A loss weighting calculation layer connected to the multi-factor result output structure, comprising: the first loss weighting calculation module is used for performing loss weighting calculation on the pulse rate by adopting an MSE function; and the second loss weighting calculation module is used for performing loss weighting calculation on the rhythm two classification result, the pulse vigor fluency three-classification result and the pulse vigor tensity two-classification result by adopting a weighted cross entropy loss function.
2. Resampling the signal;
since the pulse rate of the pulse signal is a specific value in terms of sub-minute, which is between 40 and 200, the pulse rate of the detected data is generally distributed in a more central position, the pulse rate results at two ends are less obtained, and the distribution is uneven. Therefore, the signal is resampled, the resampled parameter is 1-1.3, the robustness of pulse rate output can be greatly improved, and the length of 1800samples is randomly intercepted after resampling is completed for training. Meanwhile, the resampling mode does not influence the judgment results of rhythm and pulse condition, so that the multi-task model can simultaneously obtain the output of four tasks. If a mode of combining loss weighting calculation and signal resampling is utilized, the robustness of the model can be further improved, and the over-fitting phenomenon is prevented.
3. Training a network model;
and (3) carrying out model training on the pulse condition signal recognition framework model by using the prepared data set, so that the parameters of the model are automatically calculated by a computer, and the output result is close to the marked result as much as possible. Pulse signals with the time length of 20 seconds are intercepted each time, and the data of the middle 18 seconds are randomly intercepted and taken as training data to prepare a training set. In the model training, an Adam optimizer is used as an optimizer for model training to train 700 epoch. The initial learning rate was set at 0.0001 and the shrinkage was 0.1 fold after 300 epochs.
4. Recognizing pulse conditions;
acquiring a pulse condition data segment to be identified; based on the constructed multi-task pulse condition signal identification model, obtaining a multi-element pulse condition identification result corresponding to the pulse condition data segment according to the pulse condition data segment;
the model of the embodiment can subdivide the result of the pulse condition signal from a single element, rather than only outputting the pulse condition name, and is more consistent with the theory of traditional Chinese medicine. By using the model obtained by data training, the average pulse rate error is lower than 1 time/minute, the classification accuracy of the rhythm is 99%, the classification accuracy of the pulse condition (fluency) is 93%, and the classification accuracy of the pulse condition (nervousness) is 92%.
Similar to the principle of the above embodiments, the invention provides a multitask neural network pulse data processing system.
Specific embodiments are provided below in conjunction with the attached figures:
fig. 7 is a schematic structural diagram of a pulse data processing system of a multitasking neural network according to an embodiment of the present invention.
The system comprises:
a data obtaining module 71, configured to obtain a pulse condition data segment to be identified;
the identification module 72 is connected with the data acquisition module 71 and is used for acquiring a multi-element pulse condition identification result corresponding to the pulse condition data segment based on the constructed multi-task pulse condition signal identification model; wherein the multi-element recognition result comprises: a pulse rate recognition result, a rhythm recognition result, a pulse condition fluency recognition result and a pulse condition tension recognition result;
the construction method of the multitask pulse condition signal identification model comprises the following steps:
constructing a pulse signal identification frame model;
training the pulse condition signal identification framework model by utilizing a training data set to obtain a multi-task pulse condition signal identification model; and wherein the training data set comprises: a plurality of pulse condition data segment samples and a multi-element pulse condition identification result corresponding to each pulse condition data segment sample.
It should be noted that the division of each module in the system embodiment of fig. 7 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. For 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 multitask neural network pulse data processing system and the multitask pulse signal identification model have been described in the foregoing embodiments, repeated descriptions are omitted here.
Fig. 8 shows a schematic structural diagram of a pulse data processing terminal 80 of the multitask neural network in the embodiment of the present invention.
The multitask neural network pulse data processing terminal 80 includes: a memory 81 and a processor 82, the memory 81 being for storing computer programs; the processor 82 runs a computer program to implement the pulse data processing method of the multitask neural network as described in fig. 1.
Alternatively, the number of the memories 81 may be one or more, the number of the processors 82 may be one or more, and fig. 8 illustrates one example.
Optionally, the processor 82 in the terminal 80 for processing pulse condition data based on a multitasking neural network loads one or more instructions corresponding to the processes of the application program into the memory 81 according to the steps described in fig. 1, and the processor 82 runs the application program stored in the first memory 81, so as to implement various functions in the method for processing pulse condition data based on a multitasking neural network described in fig. 1.
Optionally, the memory 81 may include, but is not limited to, a high speed random access memory, a non-volatile memory. Such as one or more magnetic disk storage devices, flash memory devices, or other non-volatile solid-state storage devices; the Processor 82 may include, but is not limited to, a Central Processing Unit (CPU), a Network Processor (NP), and the like; the Integrated Circuit may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other Programmable logic device, a discrete Gate or transistor logic device, or a discrete hardware component.
Optionally, the Processor 82 may be a general-purpose Processor, and includes a Central Processing Unit (CPU), a Network Processor (NP), and the like; the Integrated Circuit may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other Programmable logic device, a discrete Gate or transistor logic device, or a discrete hardware component.
The present invention also provides a computer-readable storage medium storing a computer program, which when executed implements the method for processing pulse condition data of a multitask neural network as shown in fig. 1. The computer-readable storage medium may include, but is not limited to, floppy diskettes, optical disks, CD-ROMs (compact disc-read only memories), magneto-optical disks, ROMs (read-only memories), RAMs (random access memories), EPROMs (erasable programmable read only memories), EEPROMs (electrically erasable programmable read only memories), magnetic or optical cards, flash memory, or other type of media/machine-readable medium suitable for storing machine-executable instructions. The computer readable storage medium may be a product that is not accessed by the computer device or may be a component that is used by an accessed computer device.
In summary, according to the multitask neural network pulse condition data processing method, the multitask neural network pulse condition data processing system and the multitask neural network pulse condition data processing terminal, a pulse rate identification result, a rhythm identification result, a pulse condition fluency identification result and a pulse condition tension identification result corresponding to a segment are obtained according to the constructed multitask pulse condition signal identification model. The scheme of the invention can subdivide the result of the pulse condition signal from a single element, not only output the pulse condition name, more accord with the theory of traditional Chinese medicine, and can obtain the high-accuracy pulse rate identification result, rhythm identification result, pulse condition fluency identification result and pulse condition tension identification result.
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 multitask neural network pulse condition data processing method is characterized by comprising the following steps:
acquiring a pulse condition data segment to be identified;
based on the constructed multi-task pulse condition signal identification model, obtaining a multi-element pulse condition identification result corresponding to the pulse condition data segment according to the pulse condition data segment; wherein the multi-element recognition result comprises: a pulse rate recognition result, a rhythm recognition result, a pulse condition fluency recognition result and a pulse condition tension recognition result;
the construction method of the multitask pulse condition signal identification model comprises the following steps:
constructing a pulse signal identification frame model;
training the pulse condition signal identification framework model by utilizing a training data set to obtain a multi-task pulse condition signal identification model; and wherein the training data set comprises: a plurality of pulse condition data segment samples and a multi-element pulse condition identification result corresponding to each pulse condition data segment sample.
2. The method for processing the pulse condition data of the multitasking neural network according to claim 1, wherein the pulse condition signal recognition framework model comprises:
the characteristic crude extraction structure is used for carrying out crude characteristic extraction on the input pulse condition data fragments so as to output 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 crude characteristic extraction data to output fine characteristic extraction data;
the multi-element result output structure is connected with the fine feature extraction structure and is used for obtaining a multi-element identification result corresponding to the pulse condition data segment based on the input fine feature extraction data;
wherein the multi-element result output structure comprises:
the pulse rate identification result output structure is used for extracting data and outputting a pulse rate identification result based on the input fine characteristics;
a rhythm recognition result output structure for outputting a rhythm recognition result based on the input fine feature extraction data;
the pulse fluency identification result output structure is used for extracting data and outputting a pulse fluency identification result based on the input fine characteristics;
and the pulse tension recognition result output structure is used for extracting data based on the input fine characteristics and outputting a pulse tension recognition result.
3. The method for processing the pulse condition data of the multitask neural network according to claim 2, wherein the characteristic crude extracting 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.
4. The method for processing pulse data of a multitasking neural network according to claim 1, wherein said feature detail extracting 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 ResBlock1D structure is used for carrying out three times of residual error calculation on 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.
5. The method for processing the pulse condition data of the multitask neural network according to claim 2, wherein an activation function adopted by the pulse rate identification result output structure is a relu function; the activation function adopted by the rhythm identification result output structure is a sigmoid function; the activation function adopted by the pulse fluency identification result output structure is a softmax function; and the activation function adopted by the pulse tension recognition result output structure is a sigmoid function.
6. The method for processing the pulse condition data of the multitask neural network according to claim 1 or 5, wherein the multi-element result output structure comprises:
a pulse rate recognition result output structure for extracting data based on the input fine features and outputting pulse rate values in units of sub-minute;
the rhythm identification result output structure is used for extracting data and outputting rhythm two-classification results based on the input fine characteristics; wherein the rhythm binary classification result comprises: corresponding to the result of regular or irregular rhythm;
the pulse fluency identification result output structure is used for extracting data and outputting pulse fluency three-classification results based on the input fine characteristics; wherein the three classification results of pulse fluency degree comprise: corresponding to one of the slippery pulse, the unsmooth pulse and the unsmooth pulse;
the pulse tension recognition result output structure is used for extracting data and outputting pulse tension two-classification results based on the input fine characteristics; wherein, the pulse tension binary classification result comprises: corresponding to a wiry or even pulse.
7. The method for processing pulse data of a multitasking neural network according to claim 1, wherein said pulse signal identification framework model further comprises:
a loss weighting calculation layer connected to the multi-factor result output structure, comprising:
the first loss weighting calculation module is used for performing loss weighting calculation on the pulse rate identification result by adopting an MSE function;
and the second loss weighting calculation module is used for performing loss weighting calculation on the rhythm identification result, the pulse condition fluency identification result and the pulse condition nervousness identification result by adopting a weighted cross entropy loss function.
8. The method of claim 1, wherein the segments of pulse data are obtained by resampling pulse signals.
9. A multitasking neural network pulse data processing system, comprising:
the data acquisition module is used for acquiring pulse condition data segments to be identified;
the identification module is connected with the data acquisition module and used for acquiring a multi-element pulse condition identification result corresponding to the pulse condition data segment based on the constructed multi-task pulse condition signal identification model; wherein the multi-element recognition result comprises: a pulse rate recognition result, a rhythm recognition result, a pulse condition fluency recognition result and a pulse condition tension recognition result;
the construction method of the multitask pulse condition signal identification model comprises the following steps:
constructing a pulse signal identification frame model;
training the pulse condition signal identification framework model by utilizing a training data set to obtain a multi-task pulse condition signal identification model; and wherein the training data set comprises: a plurality of pulse condition data segment samples and a multi-element pulse condition identification result corresponding to each pulse condition data segment sample.
10. A multitask neural network pulse condition data processing terminal is characterized by comprising: one or more memories and one or more processors;
the one or more memories for storing a computer program;
the one or more processors, coupled to the memory, to execute the computer program to perform the method of any of claims 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210102125.XA CN114366047B (en) | 2022-01-27 | 2022-01-27 | Multi-task neural network pulse condition data processing method, system and terminal |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210102125.XA CN114366047B (en) | 2022-01-27 | 2022-01-27 | Multi-task neural network pulse condition data processing method, system and terminal |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114366047A true CN114366047A (en) | 2022-04-19 |
| CN114366047B CN114366047B (en) | 2023-05-09 |
Family
ID=81146290
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210102125.XA Active CN114366047B (en) | 2022-01-27 | 2022-01-27 | Multi-task neural network pulse condition data processing method, system and terminal |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114366047B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115374125A (en) * | 2022-09-01 | 2022-11-22 | 无锡市华焯光电科技有限公司 | Pulse condition diagnosis and classification method, database construction method, device and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103034837A (en) * | 2011-09-30 | 2013-04-10 | Ge医疗系统环球技术有限公司 | Relevancy of characteristic parameters and pulse condition elements |
| CN110929697A (en) * | 2019-12-17 | 2020-03-27 | 中国人民解放军海军航空大学 | Neural network target identification method and system based on residual error structure |
| CN111202512A (en) * | 2020-03-05 | 2020-05-29 | 齐鲁工业大学 | Electrocardiogram classification method and device based on wavelet transformation and DCNN |
| CN112561863A (en) * | 2020-12-03 | 2021-03-26 | 吉林大学 | Medical image multi-classification recognition system based on improved ResNet |
| CN112613581A (en) * | 2020-12-31 | 2021-04-06 | 广州大学华软软件学院 | Image recognition method, system, computer equipment and storage medium |
| CN113397555A (en) * | 2021-07-21 | 2021-09-17 | 南通大学附属医院 | Arrhythmia classification algorithm of C-LSTM for physiological parameter monitoring |
-
2022
- 2022-01-27 CN CN202210102125.XA patent/CN114366047B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103034837A (en) * | 2011-09-30 | 2013-04-10 | Ge医疗系统环球技术有限公司 | Relevancy of characteristic parameters and pulse condition elements |
| CN110929697A (en) * | 2019-12-17 | 2020-03-27 | 中国人民解放军海军航空大学 | Neural network target identification method and system based on residual error structure |
| CN111202512A (en) * | 2020-03-05 | 2020-05-29 | 齐鲁工业大学 | Electrocardiogram classification method and device based on wavelet transformation and DCNN |
| CN112561863A (en) * | 2020-12-03 | 2021-03-26 | 吉林大学 | Medical image multi-classification recognition system based on improved ResNet |
| CN112613581A (en) * | 2020-12-31 | 2021-04-06 | 广州大学华软软件学院 | Image recognition method, system, computer equipment and storage medium |
| CN113397555A (en) * | 2021-07-21 | 2021-09-17 | 南通大学附属医院 | Arrhythmia classification algorithm of C-LSTM for physiological parameter monitoring |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115374125A (en) * | 2022-09-01 | 2022-11-22 | 无锡市华焯光电科技有限公司 | Pulse condition diagnosis and classification method, database construction method, device and storage medium |
| CN115374125B (en) * | 2022-09-01 | 2024-05-10 | 无锡市华焯光电科技有限公司 | Pulse diagnosis classification method, database construction method, device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114366047B (en) | 2023-05-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20210398287A1 (en) | Image processing method and image processing device | |
| JP6794593B2 (en) | Methods and devices for optimizing and applying multi-layer neural network models, and storage media | |
| Lynch | Development of a raster database of annual, monthly and daily rainfall for southern Africa | |
| CN111275175B (en) | Neural network training method, device, image classification method, device and medium | |
| Goceri | Skin disease diagnosis from photographs using deep learning | |
| Konstantopoulos et al. | Analyzing effect sizes: Fixed-effects models | |
| JP7087825B2 (en) | Learning device and learning method | |
| CN111967495A (en) | Classification recognition model construction method | |
| CN109146891B (en) | Hippocampus segmentation method and device applied to MRI and electronic equipment | |
| CN111354338B (en) | Parkinson speech recognition system based on PSO convolution kernel optimization sparse transfer learning | |
| Zhang et al. | Applying sparse coding to surface multivariate tensor-based morphometry to predict future cognitive decline | |
| CN114366047A (en) | Multitask neural network pulse condition data processing method, system and terminal | |
| US20220108180A1 (en) | Method and apparatus for compressing artificial neural network | |
| CN111696119A (en) | Nuclear magnetic resonance image hippocampus segmentation method and device | |
| CN113538359A (en) | System and method for finger vein image segmentation | |
| Kaul et al. | Focusnet++: Attentive aggregated transformations for efficient and accurate medical image segmentation | |
| CN107977976A (en) | A kind of image partition method | |
| Xi et al. | Boosting with spatial regularization | |
| CN112233742B (en) | Medical record document classification system, equipment and storage medium based on clustering | |
| CN111680132A (en) | Noise filtering and automatic classifying method for internet text information | |
| CN116522282A (en) | Attention defect and hyperactivity disorder detection method based on BOLD signal | |
| CN113782121B (en) | Random grouping method, device, computer equipment and storage medium | |
| CN114469017A (en) | Pulse condition data detection method and system | |
| CN115153496A (en) | Early cognitive dysfunction classification system, method and terminal | |
| Attia et al. | Brain seizures detection using machine learning classifiers based on electroencephalography signals: a comparative study |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |