CN114916910B

CN114916910B - Pulse classification method, classification model training method, classification device, and storage medium

Info

Publication number: CN114916910B
Application number: CN202210475427.1A
Authority: CN
Inventors: 史心群; 段晓东
Original assignee: Wuxi Huazhuo Optoelectronics Technology Co ltd
Current assignee: Wuxi Huazhuo Optoelectronics Technology Co ltd
Priority date: 2022-04-29
Filing date: 2022-04-29
Publication date: 2024-04-09
Anticipated expiration: 2042-04-29
Also published as: CN114916910A

Abstract

The invention discloses a pulse classification method, a classification model training method, classification equipment and a storage medium, wherein the pulse classification method comprises the following steps: receiving a pulse basic data set; comparing the pulse condition detection data with the preset value of at least two pulse intervals in the corresponding pulse dimension to obtain a target pulse interval; comparing the pulse condition operation data with the numerical values of at least two pulse type intervals in preset corresponding pulse type dimensions in the range of the target pulse type to obtain a target pulse type interval; and judging the pulse condition corresponding to the pulse basic data set as the target pulse type in the target pulse types. The pulse classification method provided by the invention can form classification logic for two-step judgment, and more pulse classification results can be obtained by subdivision.

Description

Pulse classification method, classification model training method, classification device, and storage medium

Technical Field

The invention relates to the field of traditional Chinese medicine pulse diagnosis, in particular to a pulse condition classification method, a classification model training method, classification equipment and a storage medium.

Background

The four diagnostic methods of "looking at, smelling, asking and cutting" are the methods of diagnosing patients in Chinese national traditional medicine (abbreviated as traditional Chinese medicine, hereinafter the same), wherein "cutting" generally represents pulse diagnosis, a doctor of traditional Chinese medicine touches three parts of the cun, guan and chi of the radial artery of the patient by fingers, applies different pressures such as floating, middle and sinking respectively, senses the fluctuation of the artery of the lung meridian of the hand taiyin in traditional Chinese medicine, and analyzes the pulse data containing the information of the position, intensity, trend, shape, width, rhythm and the like of the pulse to know the pulse condition of the patient in various dimensions so as to analyze and judge the current physiological state of the patient. Because the acquisition of pulse data does not need to carry out invasive operation on a patient, and does not need to acquire body fluid or other secretions of a human body to carry out high-precision analysis, a doctor of traditional Chinese medicine can quickly grasp the condition of the patient and take medicine according to symptoms, and therefore, the method has extremely strong development requirements.

In the prior art, after calculation is performed on the acquired pulse data, data capable of preliminarily reflecting the pulse characteristics are obtained, and then the data are compared with standard data mastered in advance, so that the pulse type is preliminarily obtained, but only pulse types with obvious partial characteristics can be classified in the prior art so as to master the rough condition of the pulse, so that the accuracy of pulse classification by using a computer program in the prior art is far less than the result obtained by actual operation of traditional Chinese medicine, the effect of comprehensively processing and analyzing the acquired data to obtain more accurate and specific pulse condition types cannot be exerted, and pulse condition classification information is more difficult to accurately provide for doctors.

Disclosure of Invention

The invention aims to provide a pulse classification method, which aims to solve the technical problems that the pulse classification precision is low, the accuracy is poor, and specific pulse types which are enough for doctors to judge cannot be obtained in the prior art.

The invention aims at providing a pulse classification model training method.

One of the purposes of the present invention is to provide a pulse classification device.

It is an object of the present invention to provide a storage medium.

In order to achieve one of the above objects, an embodiment of the present invention provides a pulse classification method, including: receiving a pulse basic data set, wherein the pulse basic data set comprises pulse condition detection data and pulse condition operation data; comparing the pulse condition detection data with the numerical values of at least two pulse intervals in preset corresponding pulse dimensions to obtain a target pulse interval; the pulse condition detection data only fall into the target pulse class interval, and the target pulse class interval represents the data range of the target pulse class under the pulse class dimension; comparing the pulse condition operation data with the numerical values of at least two pulse type intervals in preset corresponding pulse type dimensions in the range of the target pulse type to obtain a target pulse type interval; the pulse condition operation data only fall into the target pulse type interval, and the target pulse type interval represents the data range of the target pulse type under the pulse type dimension; and judging that the pulse condition corresponding to the pulse basic data set is the target pulse type in the target pulse types.

As a further improvement of an embodiment of the invention, the pulse condition detection data is at least one of periodic data, wave crest data, wave trough data and normalized maximum average data; the pulse condition operation data is at least one of rising branch slope data, resistance coefficient data, rising slope data and falling slope data; the target pulse is one of floating pulse, flood pulse, soft pulse, loose pulse, hollow pulse, leather pulse, deep pulse, volt pulse, firm pulse, weak pulse, slow pulse, astringency pulse, knot pulse, rapid pulse, promoting pulse, disease pulse, artery, deficiency pulse, micro pulse, thin pulse, short pulse, real pulse, slippery pulse, tight pulse, long pulse and wiry pulse; the target pulse is one of floating, sinking, slow, rapid, weak and real.

As a further improvement of an embodiment of the present invention, the pulse condition detection data includes first detection data and second detection data, and the pulse dimension includes a first pulse dimension and a second pulse dimension; the method specifically comprises the following steps: comparing the first detection data with the numerical values of at least two pulse intervals in the corresponding first pulse dimension; if the first detection data simultaneously falls into a first pulse interval and a second pulse interval in the first pulse dimension, comparing other pulse detection data in the pulse basic data set with at least two pulse intervals in the corresponding pulse dimension, searching to obtain a second pulse dimension, a third pulse interval in the second pulse dimension, and second detection data only falling into the third pulse interval, wherein the third pulse interval is used as the target pulse interval.

As a further improvement of an embodiment of the present invention, the first detection data is period detection data, and the other pulse condition detection data includes at least one of peak detection data, trough detection data and normalized maximum average detection data; the method specifically comprises the following steps: if the period detection data fall into two pulse period intervals at the same time, comparing the peak detection data, the trough detection data and/or the normalized maximum average detection data with at least two pulse intervals in the corresponding pulse peak dimension, pulse trough dimension and/or pulse normalized maximum average dimension.

As a further improvement of an embodiment of the present invention, the pulse condition operation data includes first operation data and second operation data, and the pulse type dimension includes a first pulse type dimension and a second pulse type dimension; the method specifically comprises the following steps: comparing the first operation data with the numerical values of at least two pulse intervals in the corresponding first pulse dimension; if the first operation data falls into the first pulse type interval and the second pulse type interval in the first pulse type dimension at the same time, comparing other pulse type operation data in the pulse basic data set with at least two pulse type intervals in the corresponding pulse type dimension, searching to obtain a third pulse type interval in the second pulse type dimension and the second pulse type dimension, and only falling into the second operation data of the third pulse type interval, wherein the third pulse type interval is used as the target pulse type interval.

As a further improvement of an embodiment of the present invention, the first operation data is resistance coefficient operation data, and the other pulse condition operation data includes at least one of rising branch slope operation data, rising slope operation data and falling slope operation data; the method specifically comprises the following steps: if the resistance coefficient operation data falls into two pulse type resistance coefficient intervals at the same time, comparing the rising branch slope operation data, the rising slope operation data and/or the falling slope operation data with at least two pulse type intervals in the corresponding pulse type rising branch slope dimension, pulse type rising slope dimension and/or pulse type falling slope dimension.

As a further improvement of an embodiment of the present invention, the pulse basis data set includes cycle detection data; the method specifically comprises the following steps: comparing the period detection data with the numerical values of at least two pulse period intervals in the preset pulse period dimension; if the period detection data fall into a plurality of types of period intervals, the plurality of types of period intervals are used as the target pulse intervals, and the pulse condition corresponding to the pulse basic data set is judged to belong to a plurality of types of pulses.

As a further improvement of an embodiment of the present invention, the pulse basic data set includes rising branch slope operation data and/or resistance coefficient operation data; the method specifically comprises the following steps: comparing the rising branch slope operation data with the numerical values of at least two pulse type rising branch slope intervals in the preset pulse type rising branch slope dimension; if the rising branch slope operation data fall into a rising branch slope interval of the pulse, taking the rising branch slope interval of the pulse as the target pulse type interval; judging that the pulse condition corresponding to the pulse basic data set is a disease pulse in a class; and/or comparing the resistance coefficient operation data with the numerical values of at least two pulse type resistance coefficient intervals in the preset pulse type resistance coefficient dimension; if the resistance coefficient operation data fall into an arterial resistance coefficient interval, taking the arterial resistance coefficient interval as the target pulse type interval; and judging the artery of the pulse basic data set corresponding to the pulse condition in the class.

As a further improvement of an embodiment of the present invention, the pulse basic data set includes rising branch slope operation data and resistance coefficient operation data; the method specifically comprises the following steps: if the resistance coefficient operation data simultaneously fall into a pulse counting resistance coefficient interval and a pulse promoting resistance coefficient interval, comparing the rising branch slope operation data with the numerical values of the pulse counting rising branch slope interval and the pulse promoting rising branch slope interval; if the rising branch slope operation data fall into a pulse-promoting rising branch slope interval, taking the pulse-promoting rising branch slope interval as the target pulse type interval; and judging that the pulse condition corresponding to the pulse basic data set is the pulse promotion in the class.

As a further improvement of an embodiment of the present invention, the pulse basic data further includes peak detection data; the method specifically comprises the following steps: if the period detection data simultaneously fall into a late period interval and a floating period interval, comparing the peak detection data with the values of the late peak interval and the floating peak interval; if the peak detection data fall into the delay peak section, the delay peak section is taken as the target pulse section, and the pulse condition corresponding to the pulse basic data set is judged to belong to the delay pulse.

As a further improvement of an embodiment of the present invention, the pulse basic data set includes resistance coefficient operation data and/or rising slope operation data; the method specifically comprises the following steps: comparing the resistance coefficient operation data with the numerical values of at least two pulse type resistance coefficient intervals in preset pulse type resistance coefficient dimensions; if the resistance coefficient operation data fall into a delayed pulse resistance coefficient section, the delayed pulse resistance coefficient section is used as the target pulse type section; judging that the pulse condition corresponding to the pulse basic data set is a delayed pulse in a delayed type; and/or comparing the rising slope calculation data with the numerical value of at least two pulse rising slope intervals in the preset pulse rising slope dimension; if the rising slope calculation data falls into an astringency rising slope section, taking the astringency rising slope section as the target pulse type section; and judging that the pulse condition corresponding to the pulse basic data set is an astringency pulse in the late class.

As a further improvement of an embodiment of the present invention, the pulse basic data set includes rising slope operation data and falling slope operation data; the method specifically comprises the following steps: if the rising slope operation data simultaneously falls into a slow pulse rising slope section and a knot pulse rising slope section, comparing the descending slope operation data with the numerical values of the slow pulse descending slope section and the knot pulse descending slope section; if the descending slope operation data falls into a knot pulse descending slope interval, taking the knot pulse descending slope interval as the target pulse type interval; and judging that the pulse condition corresponding to the pulse basic data set is a knot pulse in the late class.

As a further improvement of an embodiment of the present invention, the pulse basis data set further includes trough detection data; the method specifically comprises the following steps: if the period detection data simultaneously fall into a floating period interval and a virtual period interval, comparing the value of the trough detection data with the values of the floating trough interval and the virtual trough interval; if the trough detection data fall into the virtual trough section, the virtual trough section is taken as the target pulse section, and the pulse condition corresponding to the pulse basic data set is judged to belong to the virtual pulse.

As a further improvement of an embodiment of the present invention, the pulse basic data set includes rising branch slope operation data; the method specifically comprises the following steps: comparing the rising branch slope operation data with the numerical values of at least two pulse type rising branch slope intervals in the preset pulse type rising branch slope dimension; if the rising branch slope operation data fall into a pulse-generation rising branch slope interval, taking the pulse-generation rising branch slope interval as the target pulse type interval; and judging that the pulse condition corresponding to the pulse basic data set is a pulse generation in the deficiency type.

As a further improvement of an embodiment of the present invention, the pulse basic data further includes rising slope operation data; the method specifically comprises the following steps: if the rising branch slope operation data falls into the micro-pulse rising branch slope interval and the fine pulse rising branch slope interval at the same time, comparing the rising slope operation data with the numerical values of the micro-pulse rising slope interval and the fine pulse rising slope interval; if the rising slope calculation data falls into a micro-pulse rising slope interval, taking the micro-pulse rising slope interval as the target pulse type interval; and judging that the pulse condition corresponding to the pulse basic data set is a micro pulse in the deficiency type.

As a further improvement of an embodiment of the present invention, the pulse basis data set further comprises normalized maximum average detection data; the method specifically comprises the following steps: if the period detection data simultaneously fall into a floating period interval and a real period interval, comparing the normalized maximum average detection data with the numerical values of the floating normalized maximum average interval and the real normalized maximum average interval; and if the normalized maximum average detection data falls into the real normalized maximum average interval, taking the real normalized maximum average interval as the target pulse interval, and judging that the pulse condition corresponding to the pulse basic data set belongs to real pulse.

As a further improvement of an embodiment of the present invention, the pulse basic data set includes rising branch slope operation data; the method specifically comprises the following steps: comparing the rising branch slope operation data with the numerical values of at least two pulse type rising branch slope intervals in the preset pulse type rising branch slope dimension; if the rising branch slope operation data fall into a tight pulse rising branch slope interval, taking the tight pulse rising branch slope interval as the target pulse type interval; and judging that the pulse condition corresponding to the pulse basic data set is the tight pulse in the real type.

As a further improvement of an embodiment of the present invention, the pulse basic data set further includes a falling slope calculation data; the method specifically comprises the following steps: if the rising branch slope operation data simultaneously falls into a real pulse rising branch slope interval and a long pulse rising branch slope interval, comparing the descending slope operation data with the numerical values of the real pulse descending slope interval and the long pulse descending slope interval; if the descending slope operation data falls into a real pulse descending slope interval, taking the real pulse descending slope interval as the target pulse type interval; and judging that the pulse condition corresponding to the pulse basic data set is a real pulse in real types.

As a further improvement of an embodiment of the present invention, the method further includes: acquiring pulse time data and a pulse space data set which correspond to each other in a first period, and extracting a first maximum time value corresponding to a time pulse maximum value in the pulse time data and a first minimum time value corresponding to a time pulse minimum value in the pulse time data in the first period; traversing and acquiring a first space data set and a second space data set which respectively correspond to the first maximum time value and the first minimum time value according to the pulse space data set; screening data in the first space data set according to an analysis sliding window with a preset step length to obtain the pulse intermediate data, and calculating to form the pulse basic data set according to the pulse time data, the first space data set, the second space data set and the pulse intermediate data; the pulse time data represent the overall pulse change condition of the region to be measured at least two different moments, and the pulse space data represent the pulse condition of the region to be measured at least two different positions.

As a further improvement of an embodiment of the present invention, the method further includes: traversing the pulse time data, analyzing to obtain cycle data, calculating to obtain peak data according to the time pulse maximum value, and calculating to obtain trough data according to the time pulse maximum value; traversing the first space data set, extracting to obtain maximum peak data, normalizing the first space data set by taking the maximum peak data as a basis, calculating to obtain a maximum normalized average value, and calculating to obtain rising branch slope data according to the maximum peak data and the first maximum time value; traversing the second space data set, extracting to obtain wave trough minimum data, and calculating to obtain resistance coefficient data according to the wave trough minimum data and the first minimum time value.

As a further improvement of an embodiment of the present invention, the method further includes: fitting according to the first space data set to generate a space pressure curved surface model, traversing and obtaining a space maximum value in the space pressure curved surface model, and extracting a partial region in the space pressure curved surface model according to the analysis sliding window to obtain a pulse analysis curved surface containing the pulse intermediate data; dividing the pulse analysis curved surface into a curved surface ascending side and a curved surface descending side, and respectively calculating the slopes of the curved surface ascending side and the curved surface descending side relative to the space maximum point corresponding to the space maximum value to obtain ascending slope data and descending slope data.

As a further improvement of an embodiment of the present invention, the method further includes: calculating the preset step length according to the space maximum value, and determining a plurality of pulse length reference points by taking the space maximum value as a starting point according to an analysis sliding window configured as the preset step length; extracting a region surrounded by the pulse length reference points in the space pressure curved surface model to obtain the pulse analysis curved surface; wherein the preset step length is one fifth of the maximum value of the space.

In order to achieve one of the above objects, an embodiment of the present invention provides a training method for a pulse classification model, which is used for training the pulse classification model, and includes: receiving pulse preset data and corresponding pulse type preset labels, sequentially performing baseline drift removal and wavelet transformation denoising on the pulse preset data, and dividing according to preset proportion to obtain pulse training data, pulse verification data, and corresponding pulse type training labels and pulse type verification labels; the pulse classification model is called, the pulse training data is used as the pulse basic data set to carry out traversal prediction, and a plurality of groups of model training parameters are obtained; respectively carrying the model training parameters in the pulse classification model, and carrying out traversal verification by taking the pulse verification data as the pulse basic data set to correspondingly obtain verification result data; calculating to obtain model evaluation scores according to the verification result data and the pulse type verification labels, and traversing to obtain standard model parameters corresponding to the highest model evaluation scores; wherein the pulse classification model is configured to execute the steps of the pulse classification method according to any one of the above technical schemes.

As a further improvement of an embodiment of the present invention, the data amount of the preset proportion setting pulse training data is 80% of the data amount of the pulse preset data; the method specifically comprises the following steps: calculating to obtain accuracy, precision, recall and F1 fraction according to the verification result data and the pulse type verification tag, and calculating to obtain the model evaluation fraction according to a preset weight coefficient; the preset weight coefficient sets the weight of the accuracy rate, the weight of the accuracy rate and the weight of the recall rate to be equal and larger than the weight of the F1 fraction.

In order to achieve one of the above objects, an embodiment of the present invention provides a pulse classification device, including a processor, a memory, and a communication bus, wherein the processor and the memory complete communication with each other through the communication bus; the memory is used for storing application programs; the processor is used for realizing the steps of the pulse classification method according to any one of the technical schemes when executing the application program stored on the memory.

In order to achieve one of the above objects, an embodiment of the present invention provides a storage medium having stored thereon an application program, which when executed, implements the steps of the pulse classification method according to any one of the above aspects.

Compared with the prior art, the pulse classification method provided by the invention has the advantages that the pulse type which initially represents the pulse change trend is obtained by comparing the detection data in the pulse basic data with the values of the two pulse type intervals in the pulse type dimension, and then the pulse type which finally represents the pulse type is obtained by comparing the operation data in the pulse basic data with the values of the two pulse type intervals in the pulse type dimension, so that the classification logic of two-step judgment is formed.

Drawings

Fig. 1 is a schematic diagram of a pulse classification apparatus according to an embodiment of the present invention.

FIG. 2 is a schematic diagram showing the steps of a pulse classification method according to an embodiment of the invention.

FIG. 3 is a schematic diagram showing the steps of a pulse classification method according to another embodiment of the present invention.

FIG. 4 is a schematic diagram showing the steps of a pulse classification method according to another embodiment of the present invention.

FIG. 5 is a schematic diagram showing steps of an example of a pulse classification method according to another embodiment of the present invention.

Fig. 6 is a schematic structural diagram of a space pressure curved surface model according to an example of a pulse classification method according to another embodiment of the present invention.

FIG. 7 is a schematic diagram of a space pressure surface model projected onto a W1-W2 plane according to an embodiment of a pulse classification method according to the present invention.

FIG. 8 is a schematic diagram illustrating the steps of a training method for a pulse classification model according to an embodiment of the present invention.

Detailed Description

The present invention will be described in detail below with reference to specific embodiments shown in the drawings. These embodiments are not intended to limit the invention and structural, methodological, or functional modifications of these embodiments that may be made by one of ordinary skill in the art are included within the scope of the invention.

It should be noted that the term "comprises," "comprising," or any other variation thereof is intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Furthermore, the terms "first," "second," "third," "fourth," and the like are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.

Along with the development of medical technology, how to simulate the traditional Chinese medicine technology, the method is a technical problem to be solved in the field by acquiring external signs of a patient, particularly acquiring fluctuation conditions of the pulse of the patient within a period of time, sensing the pulse condition of the patient and analyzing to obtain information such as symptoms of the patient, and alternatively, acquiring, processing and calculating the information and the data by using a modern detection technology.

In order to further simulate the actual operation of traditional Chinese medicine, the output classification result is refined to be specific to the pulse type, and the medical staff is assisted to accurately grasp the condition of a patient, the invention provides pulse classification equipment shown in figure 1 and a pulse classification method shown in figure 2.

The invention firstly provides a storage medium, an application program is stored on the storage medium, and when the application program is executed, the invention realizes a pulse classification method, so that two layers of comparison screening can be carried out by utilizing data with different dimensions in a pulse basic data set, the pulse range is gradually narrowed, and finally, the target pulse corresponding to the pulse basic data set is determined, thereby assisting medical workers in analyzing diseases.

In addition, the storage medium may be provided in a processing device in a pulse data detection system, and the storage medium may be any available medium that can be accessed by the device, or may be a storage device such as a server, a data center, or the like that includes one or more integration of the available media. Usable media may be magnetic media such as floppy disks, hard disks, magnetic tapes, or optical media such as DVDs (Digital Video Disc, high-density digital video discs), or semiconductor media such as SSDs (Solid State disks).

An embodiment of the present invention provides a pulse classification device 100 as shown in fig. 1, which includes a processor 11, a communication interface 12, a memory 13, and a communication bus 14. The processor 11, the communication interface 12, and the memory 13 perform communication with each other via the communication bus 14.

Wherein the memory 13 is used for storing application programs; the processor 11 is configured to execute an application program stored on the memory 13, which may be an application program stored on a storage medium as described above, i.e. the storage medium may be contained in the memory 13. When the application is executed, functions and steps such as those described above can be realized as well, and corresponding technical effects can be achieved. Other structural features may be realized by referring to the pulse data processing device, and the possible functional partitions and modules may be adjusted according to the application program installed in the device.

An embodiment of the present invention provides a pulse classification method as shown in fig. 2, where a program or an instruction applied by the method may be loaded on the storage medium and/or the pulse classification device and/or the pulse data detection system, so as to achieve a technical effect of pulse classification. The pulse classification method specifically comprises the following steps.

Step 21, a pulse basis data set is received.

Step 22, comparing the pulse condition detection data with the preset value of at least two pulse intervals in the corresponding pulse dimension to obtain a target pulse interval.

Step 23, comparing the pulse condition operation data with the numerical values of at least two pulse type intervals in the preset corresponding pulse type dimension within the range of the target pulse type to obtain the target pulse type interval.

Step 24, determining the pulse condition corresponding to the pulse basic data as the target pulse type in the target pulse types.

The pulse basic data set comprises pulse condition detection data and pulse condition operation data. The pulse detection data only fall into a target pulse interval, and the target pulse interval represents the data range of the target pulse under the pulse dimension. The pulse condition operation data only falls into a target pulse type interval, and the target pulse type interval characterizes the data range of the target pulse type under the pulse type dimension.

Therefore, the pulse basic data can be utilized, and the first layer of judgment logic formed by the pulse dimension and the pulse interval is used for primarily judging the pulse, so that the pulse condition representing the pulse potential is obtained; the above preliminary judgment can further reduce the range to be determined of the pulse type to which the pulse condition corresponding to the pulse basic data belongs, thereby facilitating the subsequent determination of the pulse type. And continuing to perform pulse type final judgment through a second layer of judgment logic formed by pulse type dimension and pulse type interval to obtain specific pulse type corresponding to the pulse basic data.

The pulse basic data set may include or be equally understood as data detected by a time sensor and a space sensor which are matched, or may be defined as a data set generated by processing and calculating the detected data. The pulse condition detection data and the pulse condition operation data are contained in the pulse basic data set, and are mutually compared and verified to determine the pulse type of the pulse condition corresponding to the pulse basic data set.

In one embodiment, the pulse condition detection data is configured to characterize a macroscopically static characteristic of a pulse condition corresponding to the pulse basis data set. Specifically, in one embodiment, the time sensor may be configured to, after outputting the pressure with the value of the pressure applying time data, define the difference between the pressure applying time data and the pressure applying time data as the pulse time data, corresponding to receiving the signal with the value of the pressure applying time data. Preferably, the pulse time data characterizes the overall pulse change of the region under test at least two different moments. The space sensor may be configured to define, after outputting the pressure having the value of the pressure application space data, a difference between the pressure application space data and the pressure application time data as the pulse space data, corresponding to the signal having the value of the pressure application space data. Preferably, the pulse space data characterizes pulse conditions at least two different locations in the region to be measured. Therefore, pulse condition detection data meeting preset characteristic conditions can be directly extracted by traversing pulse time data and/or pulse space data.

Correspondingly, the pulse condition operation data is configured to represent microscopic dynamic characteristics of the pulse condition corresponding to the pulse basic data. Specifically, the pulse condition operation data includes first operation data which can be directly used as or calculated from a peak and its neighboring region, a trough and its neighboring region, a time when the peak occurs, a time when the trough occurs, and trend data generated from the above data. Further, the trend data may be slope data characterizing the speed of change of the distribution data, slope change data characterizing the acceleration of change of the distribution data, or the like.

Here, the manner of determining whether the pulse condition detection data falls into the pulse type section and whether the pulse condition operation data falls into the pulse type section may be determined by, for example, determining the minimum value and the maximum value of the pulse condition detection data and the pulse type section respectively, that is, when the pulse condition detection data simultaneously satisfies two conditions that the value of the pulse condition detection data is greater than (or greater than or equal to) the minimum value of the section and the value of the pulse condition detection data is less than (or less than or equal to) the maximum value of the section, that is, determining that the pulse condition detection data falls into the pulse type section. Of course, in order to clearly determine whether the criterion is "greater than" or "greater than or equal to", a determination step of opening/closing the pulse section is required to be provided before determining whether the criterion falls.

In addition to the simple steps, the preset instruction implementation can be conveniently invoked based on the difference of the operation platform and the software foundation. For example, in the Python language-based implementation, the binary search may be implemented using the binary_left or the binary_right, the traversal search may be performed using the enable characteristic and the for … … else loop, or the variable res may be set as a result to perform the step-by-step judgment. For example, in an embodiment based on the Java language, an inter-valutil/isinthenterval (String data_value) sentence may be directly called, and the result of falling or not falling is obtained by using the character strings formed by the pulse detection data and the pulse intervals as the variables of the sentence, respectively. The judgment of whether the pulse condition operation data falls into the pulse type interval can be alternatively implemented by applying any one of the above technical schemes, and will not be repeated here.

And for the judgment of 'falling only', other pulse intervals can be simply traversed, and when the obtained pulse detection data fall into a certain pulse interval and the pulse detection data do not fall into other pulse intervals, the pulse interval is judged to be the target pulse interval. Correspondingly, the pulse condition operation data and the pulse type interval can also be judged by the traversing operation mode. In one embodiment, due to the specificity of some pulses, the pulse detection data are obviously different from or even exclusive of other pulses in some pulse dimensions, so that some pulse intervals in some pulse dimensions exist, and once the pulse detection data are judged to fall into the intervals, no traversal is needed for other pulse intervals to verify whether the pulse detection data fall into the intervals. For example, when the period detection data in the pulse detection data falls into a number of types of period intervals in the pulse period dimension, the number of types of period intervals is determined as the target pulse interval. Of course, the present invention is not limited to this configuration.

Similarly, in one embodiment, due to the specificity of some pulse types, the pulse types have obvious differences or even exclusion from other pulse types in some pulse type dimensions, so that some pulse type intervals in some pulse type dimensions exist, and once the pulse type operation data are judged to fall into the intervals, no traversal is needed to verify whether the other pulse type intervals fall into the intervals. For example, when the rising branch slope calculation data in the pulse condition calculation data falls into the rising branch slope section of the pulse type in the rising branch slope dimension, the rising branch slope section of the flood pulse, the rising branch slope section of the soft pulse, the rising branch slope section of the disease pulse, the rising branch slope section of the false pulse, the rising branch slope section of the short pulse, the rising branch slope section of the slip pulse, the rising branch slope section of the tight pulse, or the rising branch slope section of the chord pulse, or when the resistance coefficient calculation data in the pulse condition calculation data falls into the leather resistance coefficient section, the delay resistance coefficient section, or the arterial resistance coefficient section in the pulse type in the resistance coefficient dimension, or when the rising slope calculation data in the pulse condition calculation data falls into the rising slope section of the astringency pulse in the rising slope dimension, it is determined that the first calculation data falls into the corresponding section only.

Of course, the present invention may not have the above-mentioned judgment conditions (corresponding to the types of the pulse condition operation data) at the same time, and either of the above-mentioned judgment conditions for the dimension and the judgment conditions for the section may be selected to achieve the corresponding intended effects, and the present invention is not limited to this configuration.

In addition, after judging that the pulse condition detection data only falls into the target pulse class interval, the pulse condition detection method can also output the judging result that the pulse condition corresponding to the pulse basic data set belongs to the target pulse class, so that the pulse condition detection method can be used for reference by medical workers, and even the execution of the subsequent steps can be selectively stopped according to the requirement. Similar variations are included in the scope of the invention.

Preferably, in a specific embodiment, the pulse condition detection data may be one of periodic data, peak data, trough data, and normalized maximum average data, so that the pulse basic data may be determined by performing preliminary screening on the pulse according to the static data. In contrast, the pulse dimension may be one of pulse period dimension, pulse peak dimension, pulse trough dimension, and pulse normalized maximum average dimension. The pulse condition operation data can be one of rising branch slope data, resistance coefficient data, rising slope data and falling slope data, so that pulse types can be further screened from the trend data to determine the pulse types corresponding to the pulse basic data. In contrast, the pulse type dimension may be one of a pulse type rising branch slope dimension, a pulse type drag coefficient dimension, a pulse type rising slope dimension, and a pulse type falling slope dimension.

Further, the target pulse can be one of floating pulse, flood pulse, soft pulse, loose pulse, hollow pulse, leather pulse, deep pulse, weak pulse, fast pulse, weak pulse, slow pulse, astringent pulse, knot pulse, rapid pulse, fast pulse, disease pulse, artery, weak pulse, micro pulse, thin pulse, short pulse, real pulse, slippery pulse, tight pulse, long pulse and wiry pulse. Each of the above-listed pulse types may have a plurality of pulse type intervals corresponding to the pulse type rising branch slope dimension, the pulse type resistance coefficient dimension, the pulse type rising slope dimension, and the pulse type falling slope dimension, respectively, for example, the above-mentioned floating pulse rising branch slope interval, and the definition of other pulse type intervals is anticipated by those skilled in the art, and will not be repeated here. The target pulse can be one of floating, sinking, late, numbered, deficient and real. Each of the above-listed pulses may have a plurality of pulse intervals corresponding to the pulse period dimension, the pulse peak dimension, the pulse trough dimension, and the pulse normalized maximum average dimension, for example, the above-mentioned several pulse period intervals, and the definition of other pulse intervals is anticipated by those skilled in the art and will not be repeated here.

In one embodiment, 28 pulses can be classified as follows: superficial, surging, soft, scattered, hollow, and leather veins are considered to belong to the superficial category, deep, superficial, firm, and weak veins may be considered to belong to the deep category, delayed, slow, astringent, and knotted veins may be considered to belong to the slow category, rapid, and arterial veins may be considered to belong to the rapid category, deficient, micro, thin, and short veins may be considered to belong to the deficient category, and/or real, slippery, tight, long, and wiry veins may be considered to belong to the real category. Thus, the pulse basic data sets are sequentially subjected to distinguishing judgment of pulse types and pulse types, and the corresponding pulse types are finally locked.

The present invention provides a first example of a pulse classification method based on the above embodiment, which specifically includes the following steps.

Step 21, a pulse basis data set is received.

Step 221, comparing the first detection data with the values of at least two pulse intervals in the corresponding first pulse dimension.

Step 222, if the first detection data falls into the first pulse interval and the second pulse interval in the first pulse dimension at the same time, comparing the other pulse detection data in the pulse basic data set with at least two pulse intervals in the corresponding pulse dimension, searching to obtain a second pulse dimension, a third pulse interval in the second pulse dimension, and second detection data falling into the third pulse interval only, and taking the third pulse interval as a target pulse interval.

And step 23, comparing the pulse condition operation data with the numerical values of at least two pulse type intervals in the preset corresponding pulse type dimension to obtain a target pulse type interval.

Step 24, determining the pulse condition corresponding to the pulse basic data set as the target pulse type in the target pulse types.

The pulse condition detection data comprise first detection data and second detection data, and the pulse dimension comprises a first pulse dimension and a second pulse dimension. Therefore, the first embodiment provides a case that the pulse condition detection data may fall into two pulse intervals in the corresponding pulse dimensions in the calculation process of the target pulse interval, based on this, the complex cases that the pulse condition detection data includes the plurality of types can be dealt with, the calculation and the judgment of the plurality of pulse intervals can be performed on a pulse dimension-by-pulse dimension basis, and finally the target pulse interval meeting the condition is obtained.

For different kinds of pulse condition detection data, the detection rate of the pulse period in the target pulse interval relative to the corresponding pulse dimension is different, so that it is preferable that the period detection data be set as first detection data, and at least one of the peak detection data, the trough detection data and the normalized maximum average detection data be set as other pulse condition detection data. Based on this, the above step 222 may be further modified into:

Step 222', if the period detection data falls into two pulse period intervals, comparing the peak detection data, the trough detection data and/or the normalized maximum average data with at least two pulse intervals in the corresponding pulse peak dimension, pulse trough dimension and/or pulse normalized maximum average dimension, searching to obtain a third pulse interval in the second pulse dimension and the second pulse dimension, and the second detection data falling into the third pulse interval only, and taking the third pulse interval as the target pulse interval.

Through the operation judgment of the pulse period dimension, at least the pulse detection data corresponding to the number type and other pulse types, the sinking type and the late type, the sinking type and the weak type and the late type and the weak type can be distinguished; through the operation judgment of the pulse crest dimension, at least the floating type and the slow type, the sinking type and the real type, the slow type and the deficient type, the slow type and the real type, and the deficient type and the real type respectively correspond to the pulse detection data in pairs; through the operation and judgment of the wave trough dimension of the pulse, at least the pulse detection data corresponding to the floating type, the sinking type, the floating type, the virtual type, the sinking type and the late type can be distinguished from each other; through the operation and judgment of the maximum average dimension of pulse normalization, at least pulse detection data corresponding to floating type and real type can be distinguished from each other.

Of course, the above summary outlines a preferred configuration manner, in other embodiments, other pulse condition detection data may include peak detection data, trough detection data and normalized maximum average detection data at the same time, and further sets the order of pulse period dimension, pulse peak dimension, pulse trough dimension, and pulse normalized maximum average dimension to perform dimension-by-dimension judgment, so as to accelerate the speed of detecting the target pulse interval. In the embodiment where the other pulse condition detection data includes only peak detection data and normalized maximum average detection data, the above order relation may still be applied to achieve the effect of improving the detection speed. In addition, during iterative judgment, one of the peak detection data, the trough detection data and the normalized maximum average detection data or other combinations can be selected in a targeted manner based on the judgment result of the period detection data or any other pulse condition detection data to perform operation judgment, and the generated embodiment will be described in detail below.

Furthermore, the present invention provides a second example of a pulse classification method based on the above embodiment, which specifically includes the following steps.

Step 21, a pulse basis data set is received.

Step 231, comparing the first operation data with the values of at least two pulse intervals in the corresponding first pulse dimension.

Step 232, if the first operation data falls into the first pulse type interval and the second pulse type interval in the first pulse type dimension at the same time, comparing the other pulse type operation data in the pulse basic data set with at least two pulse type intervals in the corresponding pulse type dimension, searching to obtain a third pulse type interval in the second pulse type dimension and the second pulse type dimension, and taking the third pulse type interval as the target pulse type interval.

The pulse condition operation data comprise first operation data and second operation data, and the pulse type dimension comprises a first pulse type dimension and a second pulse type dimension. Thus, the second embodiment provides a case where the possible pulse condition operation data falls into two pulse type intervals in the corresponding pulse type dimension simultaneously in the calculation process of the target pulse type interval. Based on this, it is possible to perform the calculation and judgment of the plurality of pulse type sections in the dimension of each pulse type in response to the complex situation in which the pulse condition calculation data includes the plurality of types, and finally obtain the target pulse type section satisfying the condition.

The detection rate of the target pulse type section with respect to the pulse type period in the corresponding pulse type dimension is different for different kinds of pulse type operation data, and thus it is preferable that resistance coefficient operation data be set as first operation data and at least one of rising slope operation data, and falling slope operation data be set as other pulse type detection data. Based on this, the above step 232 may be further modified as follows:

step 232', if the drag coefficient operation data falls into two kinds of pulse drag coefficient intervals, comparing the rising branch slope operation data, the rising slope operation data and/or the falling slope operation data with at least two pulse intervals in the corresponding pulse rising branch slope dimension, pulse rising slope dimension and/or pulse falling slope dimension, searching to obtain a second pulse dimension, a third pulse interval in the second pulse dimension, and the second operation data falling into the third pulse interval only, and taking the third pulse interval as the target pulse interval.

Therefore, through the operation judgment of the dimension of the pulse type resistance coefficient, at least the pulse condition operation data corresponding to the leather pulse, other floating pulse types, the delayed pulse, other late pulse types, the artery, other rapid pulse types, the deep pulse, the weak pulse, the volt pulse, the weak pulse, the slow pulse, the knot pulse, the real pulse and the long pulse can be distinguished;

Through the operation judgment of the ascending branch slope dimension of the pulse, at least the floating pulse and other floating pulse types, the flood pulse and other floating pulse types, the soft pulse and other floating pulse types, the disease pulse and other rapid pulse types, the deficiency pulse and other weak pulse types, the pulse generation and other weak pulse types, the short pulse and other weak pulse types, the slippery pulse and other real pulse types, the tight pulse and other real pulse types, the wiry pulse and other real pulse types, the deep pulse and firm pulse, the deep pulse and weak pulse, the firm pulse and weak pulse, the slow pulse and knot pulse, and the rapid pulse and fast pulse can be respectively distinguished from the corresponding pulse condition operation data;

through the operation judgment of the ascending slope dimension of the pulse type, at least the astringency pulse can be distinguished from the pulse condition operation data corresponding to other late pulse types, micro pulses and thin pulses respectively;

through the operation judgment of the descending slope dimension of the pulse, the pulse condition operation data corresponding to the scattered pulse, the volt pulse, the firm pulse, the volt pulse, the weak pulse, the slow pulse, the knot pulse, the micro pulse, the thin pulse, the real pulse and the long pulse can be distinguished from each other at least.

Of course, the above description outlines a preferred configuration, and in other embodiments, other pulse condition operation data may include rising branch slope operation data, rising slope operation data, and falling slope operation data at the same time, and further sets the order of pulse type resistance coefficient dimension, pulse type rising branch slope dimension, pulse type rising slope dimension, and pulse type falling slope dimension to perform the dimension-by-dimension judgment, so as to accelerate the speed of detecting the target pulse type section. In the embodiment where the other pulse condition operation data includes only the rising branch slope operation data and the falling slope operation data, the above sequence relation may be still applicable to achieve the effect of improving the detection speed. Meanwhile, during iterative judgment, one of the rising slope operation data, the rising slope operation data and the falling slope operation data or other combinations can be selected in a targeted manner based on the judgment result of the resistance coefficient operation data or any other pulse condition operation data to carry out operation judgment, and the generated implementation mode is described in detail below.

In combination with the first embodiment and the second embodiment, another embodiment of the present invention provides a pulse classification method, as shown in fig. 3, specifically including the following steps.

Step 21, a pulse basis data set is received.

The definition of the relevant data and the explanation of the steps are referred to above and will not be repeated here. It should be emphasized that this alternative embodiment can increase the overall speed of pulse type judgment and ensure the accuracy of the target pulse type interval and the target pulse type interval retrieval.

It will be appreciated that the above-mentioned setting of the order of dimensions is not necessarily fixed, and those skilled in the art can make sequential adjustments in combination with actual needs, whether the above-mentioned determination of pulses or pulse types. Typically, the order of the pulse type dimension may be also set as the pulse type rising branch slope dimension, the pulse type drag coefficient dimension, the pulse type falling slope dimension, and the pulse type rising slope dimension. Of course, after locking the pulses, the above-described sequence may be specifically formulated according to the actual situation of the plurality of pulses described below.

For example, in one embodiment, the present invention takes advantage of the fact that several types of cycle intervals have data ranges that are exclusive of other types of pulse cycle intervals, e.g., in one case the cycle interval is (0.46,0.67), to perform the arrangement of the first stage operational decision step of step 22. Thus, the pulse basis data set may include cycle detection data, and the operation determination step of the first stage of step 22 may include the following.

In step 221A, the cycle detection data is compared with the values of at least two pulse cycles in the predetermined dimension of pulse cycles.

Step 220A, if the period detection data falls into the period intervals of the several types, the period intervals of the several types are used as the target pulse intervals, and it is determined that the pulse condition corresponding to the pulse basic data set belongs to the pulses of the several types.

Additionally, the at least two pulse period intervals at least comprise a plurality of period intervals, and one or more of a floating period interval, a sinking period interval, a delay period interval, a virtual period interval and a real period interval can be further included. For this, the person skilled in the art can adapt as desired. Meanwhile, the above definition may be alternatively implemented for the pulse crest dimension, the pulse trough dimension, and the pulse normalized maximum average dimension, which will not be described in detail below.

Based on the steps 221A and 220A in the first stage, the period detection data belonging to the period intervals of several types can be quickly locked, so that the pulse condition corresponding to the pulse basic data can be quickly classified into several types of pulses.

Further, in one embodiment, the present invention utilizes the fact that in several types of pulses, the rising branch slope interval has a data range that is exclusive from the rising branch slope interval of other types of pulses, such as in one case the rising branch slope interval is (651,745), and/or utilizes the fact that the arterial resistance coefficient interval has a data range that is exclusive from the resistance coefficient interval of other types of pulses, such as in one case the resistance coefficient interval is (0.5,0.71), to perform the first stage operation determination step of step 23 and the arrangement of the corresponding step 24. Thus, the pulse basic data set may include the rising branch slope calculation data and/or the resistance coefficient calculation data, and the calculation determination step in the first stage of step 23 and the corresponding step 24 may specifically include the following.

Step 231A, comparing the ascending branch slope calculation data with the values of at least two ascending branch slope sections in the preset ascending branch slope dimension.

In step 230A, if the rising branch slope calculation data falls into the rising branch slope interval, the rising branch slope interval is used as the target pulse type interval.

Step 24A, determining the pulse condition corresponding to the pulse basic data set as the disease pulse in the class. And/or the number of the groups of groups,

step 231A', comparing the resistance coefficient operation data with the values of at least two pulse type resistance coefficient intervals in the preset pulse type resistance coefficient dimension.

In step 230A', if the resistance coefficient calculation data falls into the arterial resistance coefficient interval, the arterial resistance coefficient interval is taken as the target pulse type interval.

Step 24A', determine the pulse condition corresponding to the pulse basic data set as an artery in the class.

For other pulse types in the number types, the comparison result between the two pulse condition operation data and the pulse type interval in the two pulse type dimensions corresponding to the two pulse type operation data can be mutually compared, so that a final pulse type judgment result can be obtained. The above-described procedure is used to form another part of the first-stage operation judgment step of step 23, and the pulse basis data may include both the rising branch slope data and the resistance coefficient operation data, and the other part of step 23 may specifically have the following configuration.

In step 2321A, if the resistance coefficient operation data falls into the pulse count resistance coefficient section and the pulse promotion resistance coefficient section at the same time, the rising branch slope operation data is compared with the values of the pulse count rising branch slope section and the pulse promotion rising branch slope section.

In step 2322A, if the ascending branch slope calculation data falls into the ascending branch slope interval, the ascending branch slope interval is taken as the target pulse type interval.

Step 24A ", the pulse condition corresponding to the basic pulse data is determined as the pulse acceleration in the category.

Correspondingly, the method can further comprise the steps of: if the rising branch slope operation data fall into the rising branch slope interval of the number pulse, taking the rising branch slope interval of the number pulse as a target pulse type interval; and judging the pulse condition corresponding to the pulse basic data as a number pulse in the category. Based on the above, no matter what pulse types in the pulse basic data the corresponding pulse conditions belong to, the rapid and accurate judgment can be carried out.

In addition, in the first stage, in consideration of the situation that the period detection data may fall into two pulse period intervals at the same time, the invention further introduces the peak detection data as the content of the pulse basic data, and combines the judgment results of the period detection data to form a contrast with each other, so as to distinguish the pulses. The above procedure is used to form the operation judgment step of the second stage of step 22, and the pulse basic data may include both the period detection data and the peak detection data, and step 22 may have the following configuration.

In step 2221B, if the period detection data falls into the late period interval and the float period interval at the same time, the peak detection data is compared with the values of the late peak interval and the float peak interval.

Step 2222B, if the peak detection data falls into the late peak section, taking the late peak section as the target pulse section, and determining that the pulse condition corresponding to the pulse basic data set belongs to the late pulse.

Of course, the step 2221B and the step 2222B described above may be provided as two execution paths in parallel with the step 220A, that is, the step 220A may be executed when it is judged that the number of types of cycle sections are fallen, and the step 2221B and the step 2222B may be checked and executed when it is judged that the number of types of cycle sections are not fallen, as a complement to the step 220A.

In addition, although step 2222B only exemplifies one case of judging a late pulse, those skilled in the art can understand that the technical scheme corresponding to the case of judging a floating pulse can be formed under the teaching of the present invention, and will not be repeated here. Further, in response to the determination of the floating pulse, the determination of the specific pulse under the floating pulse may be performed according to the above-described feature record, for example, by distinguishing the floating pulse, the surging pulse, the soft pulse by the rising slope region of the pulse, and/or by distinguishing the leather pulse by the coefficient of resistance region of the pulse, and/or by dispersing the pulse and the light pulse by the falling slope region of the pulse.

Similarly, the above steps may alternatively be applied to the following cases: the period detection data simultaneously falls into a sinking period section and a real period section, the period detection data simultaneously falls into a late period section and a virtual period section, the period detection data simultaneously falls into a late period section and a real period section, and the period detection data simultaneously falls into a virtual period section and a real period section. The formed technical scheme for judging and determining various pulses is within the protection scope of the invention.

Further, the present invention makes use of the fact that in a late pulse, the late pulse resistance coefficient interval has a data range that is exclusive of other pulse type resistance coefficient intervals, for example, in one case, the resistance coefficient interval is the advantage of (0.027,0.09), and/or the fact that the astringency rise slope interval has a data range that is exclusive of other pulse type rise slope intervals, for example, in one case, the rising slope interval is the advantage of (11.1,13.1), the second stage operation judgment step of step 23 and the arrangement of the corresponding step 24 are performed. Thus, the pulse basic data set may include the resistance coefficient operation data and/or the rising slope operation data, and the operation determination step of the second stage of step 23 and the corresponding step 24 may specifically include the following.

Step 231B, comparing the resistance coefficient operation data with the values of at least two pulse type resistance coefficient intervals in the preset pulse type resistance coefficient dimension.

In step 230B, if the resistance coefficient calculation data falls within the delay pulse resistance coefficient interval, the delay pulse resistance coefficient interval is used as the target pulse type interval.

Step 24B, determining that the pulse condition corresponding to the pulse basic data set is a delayed pulse in the delayed type. And/or the number of the groups of groups,

step 231B', comparing the rising slope calculation data with the values of at least two pulse rising slope sections in the predetermined pulse rising slope dimension.

In step 232B', if the rising slope calculation data falls into the rising slope region of the astringency, the rising slope region of the astringency is used as the target pulse type region.

Step 24B', determining that the pulse condition corresponding to the pulse basic data set is an astringency pulse in the late class.

For other pulse types in the late type, the comparison result between the two pulse condition operation data and the pulse type interval in the two pulse type dimensions corresponding to the two pulse type operation data can be mutually compared, so that a final pulse type judgment result can be obtained. The above-described procedure is used to form another part of the second-stage operation judgment step of step 23, and the pulse base data may include both the rising slope operation data and the falling slope operation data, and the other part of step 23 may specifically have the following configuration.

In step 2321B, if the rising slope calculation data falls into the ramp rising slope section and the junction rising slope section at the same time, the descending slope calculation data is compared with the values of the ramp descending slope section and the junction descending slope section.

In step 2322B, if the falling slope calculation data falls into the node falling slope interval, the node falling slope interval is used as the target pulse type interval.

Step 24B ", the pulse condition corresponding to the pulse basic data set is determined as the astringency pulse in the late class.

Correspondingly, the method can further comprise the steps of: if the descending slope calculation data falls into the descending slope interval of the slow pulse, taking the descending slope interval of the slow pulse as a target pulse type interval; and judging that the pulse condition corresponding to the pulse basic data is slow pulse in the slow type. Based on the above, no matter what kind of pulse in the pulse basic data the corresponding pulse condition belongs to, the rapid and accurate judgment can be carried out.

In addition, in the second stage, in consideration of the situation that the period detection data may fall into two pulse period intervals at the same time and the peak detection data may fall into two pulse peak intervals at the same time, the present invention further introduces the trough detection data as the content of the pulse basic data, and combines the judgment results of the period detection data to form a contrast with each other, so as to distinguish the pulses. The above procedure is used to form the operation judging step of the third stage of step 22, and the pulse basic data may include both the period detection data and the trough detection data, and step 22 may have the following configuration.

Step 2221C, if the period detection data falls into the floating period interval and the virtual period interval at the same time, comparing the values of the trough detection data with the values of the floating trough interval and the virtual trough interval.

Step 2222C, if the trough detection data falls into the virtual trough interval, taking the virtual trough interval as the target pulse interval, and determining that the pulse condition corresponding to the pulse basic data set belongs to the virtual pulse.

Of course, the step 2221C and the step 2222C may be provided with a plurality of execution paths in parallel with the step 220A and/or the step 2221B, and the step 2222B, in addition to the step 220A and/or the step 2221B, and the step 2222B.

In addition, although step 2222C only exemplifies one case of judging a pulse of the deficiency type, those skilled in the art will understand that the technical scheme corresponding to the case of judging a pulse of the float type and the technical scheme of judging a specific pulse of the float type may be formed under the teaching of the present invention, and will not be repeated here.

Similarly, the above steps may alternatively be applied to the following cases: the period detection data simultaneously falls into a sinking period section and a floating period section, and the period detection data simultaneously falls into a late period section and a sinking period section. The formed technical scheme for judging and determining various pulses is within the protection scope of the invention.

Further, the present invention utilizes the advantage that in a pulse of a weak type, the pulse-generating rising branch slope interval has a data range that is exclusive to rising branch slope intervals of other pulse types, such as in one case the rising branch slope interval is (403,435.6), and/or utilizes the advantage that the pulse-generating rising branch slope interval has a data range that is exclusive to rising branch slope intervals of other pulse types, such as in one case the rising branch slope interval is (688,716), and/or utilizes the advantage that the short pulse rising branch slope interval has a data range that is exclusive to rising branch slope intervals of other pulse types, such as in one case the rising branch slope interval is (227.1,272.5), to perform the third stage operation determining step of step 23 and the arrangement of the corresponding step 24. Thus, the pulse basic data set may include the rising branch slope calculation data, and the calculation and judgment step in the third stage of step 23 and the corresponding step 24, which may include at least the following, for example, pulse generation.

Step 231C, comparing the ascending and descending slope calculation data with the values of at least two ascending and descending slope sections in the preset ascending and descending slope dimension of the pulse.

In step 230C, if the rising branch slope calculation data falls into the rising branch slope interval, the rising branch slope interval is used as the target pulse type interval.

Step 24C, determining the pulse condition corresponding to the pulse basic data set as the pulse generation in the deficiency type.

The above steps may alternatively be applied to distinguish between deficient pulses and short pulses, and will not be repeated here. For other pulse types in the deficiency type, the comparison result between the two pulse condition operation data and the pulse type interval in the two pulse type dimensions corresponding to the two pulse type operation data can be mutually compared, so that a final pulse type judgment result can be obtained. The above-described procedure is used to form another part of the third-stage operation judgment step of step 23, and the pulse base data may include both the rising-edge slope operation data and the rising-edge slope operation data, and the other part of step 23 may specifically have the following configuration.

In step 2321C, if the rising branch slope calculation data falls into the micro-pulse rising branch slope section and the fine pulse rising branch slope section at the same time, comparing the rising slope calculation data with the values of the micro-pulse rising slope section and the fine pulse rising slope section.

In step 2322C, if the rising slope calculation data falls into the micro-pulse rising slope section, the micro-pulse rising slope section is used as the target pulse type section.

Step 24C', determining that the pulse condition corresponding to the pulse basic data set is a micro pulse in the deficiency type.

Correspondingly, the method can further comprise the steps of: if the rising slope calculation data falls into the fine pulse rising slope interval, taking the fine pulse rising slope interval as a target pulse type interval; and judging that the pulse condition corresponding to the pulse basic data is a thin pulse in the deficiency type. Based on the above, no matter which pulse types in the deficiency type the corresponding pulse conditions in the pulse basic data belong to, the rapid and accurate judgment can be carried out.

In addition, in the third stage, in consideration of the situation that the period detection data may fall into two pulse period intervals at the same time, the peak detection data may fall into two pulse peak intervals at the same time, and the trough detection data may fall into two pulse trough intervals at the same time, the invention further introduces the normalized maximum average detection data as the content in the pulse basic data, and combines the judging results of the period detection data to form a contrast with each other, thereby distinguishing the pulses. The above procedure is used to form the fourth stage of the operation determining step of step 22, and the pulse basic data may include both the period detection data and the normalized maximum average detection data, and step 22 may have the following configuration.

Step 2221D, if the period detection data falls into the floating period interval and the real period interval at the same time, comparing the value of the normalized maximum average detection data with the value of the floating normalized maximum average interval and the real normalized maximum average interval.

Step 2222D, if the normalized maximum average detection data falls into the real normalized maximum average interval, taking the real normalized maximum average interval as the target pulse interval, and determining that the pulse condition corresponding to the pulse basic data set belongs to the real pulse.

Of course, the steps 2221D and 2222D may be implemented as complements of the steps 220A and/or 2221B, 2222B, and/or steps 2221C and 2222C, and the steps 220A and/or 2221B, 2222B, and/or steps 2221C and 2222C may be implemented in multiple parallel execution ways.

In addition, although step 2222D only exemplifies one case of judging a real pulse, those skilled in the art can understand that the technical scheme corresponding to the case of judging a floating pulse and the technical scheme of judging a specific pulse under the floating condition can be formed under the teaching of the present invention, and will not be repeated here.

Further, the present invention utilizes that in a real pulse, the tight pulse rising branch slope interval has a data range that is exclusive to other pulse species rising branch slope intervals, such as in one case the rising branch slope interval is (222.5,238), and/or utilizes that the slippery pulse rising branch slope interval has a data range that is exclusive to other pulse species rising branch slope intervals, such as in one case the rising branch slope interval is (348,383), and/or utilizes that the chordal pulse rising branch slope interval has a data range that is exclusive to other pulse species rising branch slope intervals, such as in one case the rising branch slope interval is (279,334), to perform the fourth stage operational decision step of step 23 and the arrangement of the corresponding step 24. Thus, the pulse basic data set may include the rising branch slope calculation data, and the calculation determining step in the fourth stage of step 23 and the corresponding step 24, which may include at least the following, for example, a pulse tightening.

Step 231D, comparing the ascending and descending slope calculation data with the values of at least two ascending and descending slope sections in the preset ascending and descending slope dimension of the pulse.

In step 230D, if the rising branch slope calculation data falls into the tight pulse rising branch slope interval, the tight pulse rising branch slope interval is used as the target pulse type interval.

Step 24D, determining the pulse condition corresponding to the pulse basic data set as the tight pulse in the real type.

The above steps may alternatively be applied to distinguish between the slippery pulse and the chord pulse, and will not be repeated here. For other pulse types in the real type, the comparison result between the two pulse condition operation data and the pulse type interval in the two pulse type dimensions corresponding to the two pulse type operation data can be mutually compared, so that the final pulse type judgment result can be obtained. The above-described procedure is used to form another part of the fourth-stage operation judgment step of step 23, and the pulse base data may include both the rising slope operation data and the falling slope operation data, and the above-described another part of step 23 may specifically have the following configuration.

In step 2321D, if the rising branch slope calculation data falls into the real pulse rising branch slope section and the long pulse rising branch slope section at the same time, comparing the descending slope calculation data with the values of the real pulse descending slope section and the long pulse descending slope section.

In step 2322D, if the falling slope calculation data falls into the real pulse falling slope interval, the real pulse falling slope interval is used as the target pulse type interval.

In step 24D', the pulse condition corresponding to the pulse basic data set is determined as the real pulse in the real class.

Correspondingly, the method can further comprise the steps of: if the descending slope calculation data falls into the long pulse descending slope interval, taking the long pulse descending slope interval as a target pulse type interval; and judging that the pulse condition corresponding to the pulse basic data is a long pulse in the real type. Based on the above, no matter which pulse types in the real type the corresponding pulse conditions in the pulse basic data belong to, the rapid and accurate judgment can be carried out.

Six kinds of pulses and the corresponding twenty-eight kinds of pulses can be calculated, judged and distinguished. It should be noted that, the above-mentioned floating and descending pulse types, and the distinction between sinking and sinking pulse types may be implemented alternatively by any of the above-mentioned technical schemes, which are not described herein, but the formed technical scheme is still within the scope of the present invention.

In another embodiment of the present invention, as shown in fig. 4, a pulse classification method is provided, which specifically includes the following steps.

Step 201, acquiring a pulse time data set and a pulse space data set corresponding to each other in a first period, and extracting a first maximum time value corresponding to a time pulse maximum value in the pulse time data in the first period and a first minimum time value corresponding to a time pulse minimum value in the pulse time data.

Step 202, traversing and acquiring a first space data set and a second space data set corresponding to a first maximum time value and a first minimum time value respectively according to the pulse space data set.

Step 203, screening the data in the first space data set according to the analysis sliding window with the preset step length to obtain the pulse intermediate data, and calculating to form a pulse basic data set according to the pulse time data, the first space data set, the second space data set and the pulse intermediate data.

Step 21, a pulse basis data set is received.

The pulse time data represent the overall pulse change condition of the region to be measured at least two different moments, and the pulse space data represent the pulse condition of the region to be measured at least two different positions. In addition, the pulse basic data set of the present invention is not necessarily formed by using all of the pulse time data, the first space data set, the second space data set and the pulse intermediate data, and at least one of the pulse time data, the first space data set, the second space data set and the pulse intermediate data may be selected for analysis operation in view of the various configuration options of the pulse basic data set provided in the foregoing. The pulse detection data may be selected from only the first detection data, or may be selected from both the first detection data and the second detection data.

Specifically, for the pulse condition detection data and the pulse condition calculation data possibly included in the pulse basic data set, another embodiment of the present invention provides an example of a pulse condition classification method, as shown in fig. 5, which specifically includes the following steps.

And 2011, traversing pulse time data, analyzing to obtain cycle data, calculating to obtain peak data according to the maximum value of the time pulse, and calculating to obtain trough data according to the maximum value of the time pulse.

Step 2021, traversing the first spatial data set, extracting to obtain peak maximum data, normalizing the first spatial data set based on the peak maximum data, calculating to obtain a normalized maximum average value, and calculating to obtain rising branch slope data according to the peak maximum data and the first maximum time value.

Step 2022, traversing the second spatial data set, extracting to obtain minimum data of the trough, and calculating to obtain resistance coefficient data according to the minimum data of the trough and the first minimum time value.

It should be reiterated that the above steps are not necessarily included in this embodiment, and those skilled in the art may selectively omit a part of the steps or a part of the details of the steps in view of different configurations of the pulse basis data set, and thus the resulting embodiment is not exhaustive.

Peak maximum data may be defined as the dominant wave height and may be obtained by truncating the vertical distance of the peak to the baseline. The normalization method may be to traverse all data in the first spatial data set, and make the data in the first spatial data set and the peak maximum data as a quotient, so as to converge the data in the first spatial data set in the interval range of [0,1], and further calculate the average value of all normalized data in the first spatial data set, so as to obtain the normalized maximum average value. In embodiments where the spatial sensors that collect pulse spatial data are configured in three, each including twenty spatial sensors thereon, the normalized maximum average may be specifically defined as an average of sixty data after normalization.

Further, the second space data set corresponds to the time point when the time sensor collects the minimum pulse time data, and the space sensor collects all the sets of pulse space data, so that the extracted minimum data of the trough can reflect the condition that the pulse fluctuation is weakest in the first period. Trough minimum data may be specifically defined as the down-isthmic height, which may be obtained by intercepting the vertical distance of the trough from the baseline. Based on the definition of the peak maximum data and the trough minimum data, the rising branch slope data can be specifically configured as a quotient of the peak maximum data and the first maximum time value, and the resistance coefficient data can be specifically configured as a quotient of the trough minimum data and the first minimum time value, so as to reflect microscopic dynamic characteristics of the pulse condition. In an embodiment in which the spatial sensing portion is configured as sixty, the rising branch slope data and the resistance coefficient data may also be specifically defined as average values of sixty corresponding operation results.

As further shown in fig. 5, in an example provided in the above further embodiment, the following steps may be further included.

Step 2031, fitting according to the first space data set to generate a space pressure curved surface model, traversing and obtaining a space maximum value in the space pressure curved surface model, and extracting a partial region in the space pressure curved surface model according to an analysis sliding window to obtain a pulse analysis curved surface containing pulse intermediate data.

Step 2032, dividing the pulse analysis curved surface into a curved surface ascending side and a curved surface descending side, and calculating the slopes of the curved surface ascending side and the curved surface descending side relative to the spatial maximum point corresponding to the spatial maximum value respectively to obtain ascending slope data and descending slope data.

Also, the above steps are not necessarily included in this embodiment, and in view of the different composition of the pulse basis data sets, one skilled in the art may selectively omit some steps or some details of steps, and thus the resulting embodiment is not exhaustive.

As in fig. 6, a spatial pressure surface model obtained based on the above embodiment is shown. As in fig. 7, a projection of the spatial pressure surface model obtained based on the above embodiment with respect to the W1-W2 plane is shown. Therefore, based on the foregoing steps, the spatial maximum point P corresponding to the spatial maximum value can be correspondingly formed, the curved surface S can be analyzed by pulse conditions, and the curved surface ascending side S1 and the curved surface descending side S2 after being divided, and the selection of the curved surface ascending side S1 and the curved surface descending side S2 can be randomly selected.

Preferably, the area ratio of the rising side S1 of the curved surface to the pulse analysis curved surface is configured to be equal to the area ratio of the falling side S2 of the curved surface to the pulse analysis curved surface, slope reference points are selected on the edges of the rising side S1 of the curved surface and the falling side S2 of the curved surface at equal intervals respectively, and rising slope reference vectors and falling slope reference vectors which are equal to the number of the generated space maximum points P are correspondingly calculated to finally calculate rising slope data and falling slope data. In an embodiment in which a plurality of rising slope reference vectors and falling slope reference vectors are arranged, the rising slope data and the falling slope data may be average values of slope data corresponding to the plurality of vectors.

Further, in an example provided in the foregoing still another embodiment, the method may further include the steps of: calculating a preset step length according to the space maximum value, and determining a plurality of pulse length reference points by taking the space maximum value as a starting point according to an analysis sliding window configured as the preset step length; and extracting an area formed by surrounding pulse length reference points in the space pressure curved surface model to obtain a pulse analysis curved surface. Wherein the preset step size is one fifth of the maximum value of the space.

An embodiment of the present invention provides a training method for a pulse classification model, which is used for training the pulse classification model, as shown in fig. 8, and specifically includes the following steps.

Step 31, receiving pulse preset data and corresponding pulse type preset labels, sequentially performing baseline drift removal and wavelet transformation denoising on the pulse preset data, and dividing according to preset proportion to obtain pulse training data, pulse verification data, and corresponding pulse type training labels and pulse type verification labels.

And step 32, invoking a pulse classification model, and performing traversal prediction by taking the pulse training data as a pulse basic data set to obtain a plurality of groups of model training parameters.

And 33, respectively carrying model training parameters in the pulse classification model, performing traversal verification by taking the pulse verification data as a pulse basic data set, and correspondingly obtaining verification result data.

And step 34, calculating to obtain model evaluation scores according to the verification result data and the pulse type verification labels, and traversing to obtain standard model parameters corresponding to the highest model evaluation scores.

The pulse classification model is configured to perform the pulse classification method provided by any one of the foregoing embodiments or examples to perform the screening and classification of pulse types. Specifically, the pulse classification model may be bayesian, neural network, decision tree, support vector machine, etc. Considering the step configuration of the pulse classification method, the pulse classification model is preferably loaded with a neural network and is specifically constructed as an LSTM (Long Short-Term Memory) artificial neural network architecture. Training of the model can be performed autonomously by carrying a sklearn model.

Preferably, the preset proportion may be configured such that the data amount of the set pulse training data occupies 80% of the data amount of the pulse preset data.

Further, step 34 may specifically further include: according to the verification result data and the pulse type verification label, calculating to obtain accuracy, precision, recall rate and F1 score (F1 score is an index used for measuring the precision of the two classification models in statistics), and calculating to obtain model evaluation score according to a preset weight coefficient.

The preset weight coefficient sets the weight of the accuracy rate, the weight of the accuracy rate and the weight of the recall rate to be equal and larger than the weight of the F1 score. Preferably, the weights of the accuracy rate, the precision rate and the recall rate are all 30%, and the weight of the F1 score is 10%, so that the correlation of the F1 score and the accuracy rate and the recall rate can be prevented from affecting the performance of the model evaluation score.

In summary, according to the pulse classification method provided by the invention, the pulse class which initially represents the pulse variation trend is obtained by comparing the detection data in the pulse basic data with the values of the two pulse class intervals in the pulse class dimension, and then the pulse class which finally represents the pulse class is obtained by comparing the operation data in the pulse basic data with the values of the two pulse class intervals in the pulse class dimension, so as to form a classification logic for two-step judgment.

It should be understood that although the present disclosure describes embodiments, not every embodiment is provided with a separate embodiment, and that this description is for clarity only, and that the skilled artisan should recognize that the embodiments may be combined as appropriate to form other embodiments that will be understood by those skilled in the art.

The above list of detailed descriptions is only specific to practical embodiments of the present invention, and they are not intended to limit the scope of the present invention, and all equivalent embodiments or modifications that do not depart from the spirit of the present invention should be included in the scope of the present invention.

Claims

1. A pulse classification method, comprising:

receiving a pulse basic data set, wherein the pulse basic data set comprises pulse condition detection data and pulse condition operation data; the pulse condition detection data is at least one of periodic data, wave crest data, wave trough data and normalized maximum average data; the pulse condition operation data is at least one of rising branch slope data, resistance coefficient data, rising slope data and falling slope data;

Comparing the pulse condition detection data with the numerical values of at least two pulse intervals in preset corresponding pulse dimensions to obtain a target pulse interval; the pulse condition detection data only fall into the target pulse class interval, and the target pulse class interval represents the data range of the target pulse class under the pulse class dimension; the pulse dimension is at least one of pulse period dimension, pulse wave crest dimension, pulse wave trough dimension and pulse normalization maximum average dimension corresponding to the pulse detection data; the target pulse is one of a floating type, a sinking type, a slow type, a rapid type, a weak type and a real type;

comparing the pulse condition operation data with the numerical values of at least two pulse type intervals in preset corresponding pulse type dimensions in the range of the target pulse type to obtain a target pulse type interval; the pulse condition operation data only fall into the target pulse type interval, and the target pulse type interval represents the data range of the target pulse type under the pulse type dimension; the pulse type dimension is at least one of pulse type rising branch slope dimension, pulse type resistance coefficient dimension, pulse type rising slope dimension and pulse type falling slope dimension corresponding to the pulse condition operation data; the target pulse is one of floating pulse, flood pulse, soft pulse, loose pulse, hollow pulse, leather pulse, deep pulse, volt pulse, firm pulse, weak pulse, slow pulse, astringency pulse, knot pulse, rapid pulse, promoting pulse, disease pulse, artery, deficiency pulse, micro pulse, thin pulse, short pulse, real pulse, slippery pulse, tight pulse, long pulse and wiry pulse;

Judging that the pulse condition corresponding to the pulse basic data set is the target pulse type in the target pulse types;

the pulse condition detection data comprise first detection data and second detection data, and the pulse dimension comprises a first pulse dimension and a second pulse dimension; comparing the pulse condition detection data with the preset value of at least two pulse intervals in the corresponding pulse dimension to obtain a target pulse interval, wherein the method specifically comprises the following steps of:

comparing the first detection data with the numerical values of at least two pulse intervals in the corresponding first pulse dimension;

if the first detection data simultaneously fall into a first pulse interval and a second pulse interval in the first pulse dimension, comparing other pulse detection data in the pulse basic data set with at least two pulse intervals in the corresponding pulse dimension, searching to obtain a second pulse dimension, a third pulse interval in the second pulse dimension, and second detection data only falling into the third pulse interval, wherein the third pulse interval is used as the target pulse interval; the first detection data are periodic detection data, and the other pulse condition detection data comprise at least one of wave crest detection data, wave trough detection data and normalized maximum average detection data;

The pulse condition operation data comprise first operation data and second operation data, and the pulse type dimension comprises a first pulse type dimension and a second pulse type dimension; comparing the pulse condition operation data with the numerical values of at least two pulse type intervals in preset corresponding pulse type dimensions to obtain a target pulse type interval, wherein the method specifically comprises the following steps of:

comparing the first operation data with the numerical values of at least two pulse intervals in the corresponding first pulse dimension;

if the first operation data simultaneously falls into a first pulse type interval and a second pulse type interval in the first pulse type dimension, comparing other pulse type operation data in the pulse basic data set with at least two pulse type intervals in the corresponding pulse type dimension, searching to obtain a second pulse type dimension, a third pulse type interval in the second pulse type dimension, and second operation data only falling into the third pulse type interval, wherein the third pulse type interval is used as the target pulse type interval; the first operation data is resistance coefficient operation data, and the other pulse condition operation data comprises at least one of rising branch slope operation data, rising slope operation data and falling slope operation data.

2. The pulse classification method of claim 1, wherein the pulse basis data set comprises cycle detection data; the method specifically comprises the following steps:

comparing the period detection data with the numerical values of at least two pulse period intervals in the preset pulse period dimension;

if the period detection data fall into a plurality of types of period intervals, the plurality of types of period intervals are used as the target pulse intervals, and the pulse condition corresponding to the pulse basic data set is judged to belong to a plurality of types of pulses.

3. The pulse classification method according to claim 2, wherein the pulse basis data set includes rising branch slope calculation data and/or resistance coefficient calculation data; the method specifically comprises the following steps:

comparing the rising branch slope operation data with the numerical values of at least two pulse type rising branch slope intervals in the preset pulse type rising branch slope dimension;

if the rising branch slope operation data fall into a rising branch slope interval of the pulse, taking the rising branch slope interval of the pulse as the target pulse type interval;

judging that the pulse condition corresponding to the pulse basic data set is a disease pulse in a class; and/or the number of the groups of groups,

comparing the resistance coefficient operation data with the numerical values of at least two pulse type resistance coefficient intervals in preset pulse type resistance coefficient dimensions;

If the resistance coefficient operation data fall into an arterial resistance coefficient interval, taking the arterial resistance coefficient interval as the target pulse type interval;

and judging the artery of the pulse basic data set corresponding to the pulse condition in the class.

4. The pulse classification method of claim 3, wherein the pulse basis data set comprises rising branch slope calculation data and resistance coefficient calculation data; the method specifically comprises the following steps:

if the resistance coefficient operation data simultaneously fall into a pulse counting resistance coefficient interval and a pulse promoting resistance coefficient interval, comparing the rising branch slope operation data with the numerical values of the pulse counting rising branch slope interval and the pulse promoting rising branch slope interval;

if the rising branch slope operation data fall into a pulse-promoting rising branch slope interval, taking the pulse-promoting rising branch slope interval as the target pulse type interval;

and judging that the pulse condition corresponding to the pulse basic data set is the pulse promotion in the class.

5. The pulse classification method of claim 2, wherein the pulse basis data further comprises peak detection data; the method specifically comprises the following steps:

if the period detection data simultaneously fall into a late period interval and a floating period interval, comparing the peak detection data with the values of the late peak interval and the floating peak interval;

If the peak detection data fall into the delay peak section, the delay peak section is taken as the target pulse section, and the pulse condition corresponding to the pulse basic data set is judged to belong to the delay pulse.

6. The pulse classification method according to claim 5, wherein the pulse basis data set includes resistance coefficient operation data and/or rising slope operation data; the method specifically comprises the following steps:

if the resistance coefficient operation data fall into a delayed pulse resistance coefficient section, the delayed pulse resistance coefficient section is used as the target pulse type section;

judging that the pulse condition corresponding to the pulse basic data set is a delayed pulse in a delayed type; and/or the number of the groups of groups,

comparing the rising slope operation data with the numerical values of at least two pulse rising slope intervals in a preset pulse rising slope dimension;

if the rising slope calculation data falls into an astringency rising slope section, taking the astringency rising slope section as the target pulse type section;

and judging that the pulse condition corresponding to the pulse basic data set is an astringency pulse in the late class.

7. The pulse classification method of claim 6, wherein the pulse basis data set comprises rising slope calculation data and falling slope calculation data; the method specifically comprises the following steps:

if the rising slope operation data simultaneously falls into a slow pulse rising slope section and a knot pulse rising slope section, comparing the descending slope operation data with the numerical values of the slow pulse descending slope section and the knot pulse descending slope section;

if the descending slope operation data falls into a knot pulse descending slope interval, taking the knot pulse descending slope interval as the target pulse type interval;

and judging that the pulse condition corresponding to the pulse basic data set is a knot pulse in the late class.

8. The pulse classification method of claim 2, wherein the pulse basis data set further comprises trough detection data; the method specifically comprises the following steps:

if the period detection data simultaneously fall into a floating period interval and a virtual period interval, comparing the value of the trough detection data with the values of the floating trough interval and the virtual trough interval;

if the trough detection data fall into the virtual trough section, the virtual trough section is taken as the target pulse section, and the pulse condition corresponding to the pulse basic data set is judged to belong to the virtual pulse.

9. The pulse classification method of claim 8, wherein the pulse basis data set comprises rising branch slope calculation data; the method specifically comprises the following steps:

if the rising branch slope operation data fall into a pulse-generation rising branch slope interval, taking the pulse-generation rising branch slope interval as the target pulse type interval;

and judging that the pulse condition corresponding to the pulse basic data set is a pulse generation in the deficiency type.

10. The pulse classification method of claim 9, wherein the pulse basis data further comprises rising slope calculation data; the method specifically comprises the following steps:

if the rising branch slope operation data falls into the micro-pulse rising branch slope interval and the fine pulse rising branch slope interval at the same time, comparing the rising slope operation data with the numerical values of the micro-pulse rising slope interval and the fine pulse rising slope interval;

if the rising slope calculation data falls into a micro-pulse rising slope interval, taking the micro-pulse rising slope interval as the target pulse type interval;

and judging that the pulse condition corresponding to the pulse basic data set is a micro pulse in the deficiency type.

11. The pulse classification method of claim 2, wherein the pulse basis data set further comprises normalized maximum average detection data; the method specifically comprises the following steps:

if the period detection data simultaneously fall into a floating period interval and a real period interval, comparing the normalized maximum average detection data with the numerical values of the floating normalized maximum average interval and the real normalized maximum average interval;

and if the normalized maximum average detection data falls into the real normalized maximum average interval, taking the real normalized maximum average interval as the target pulse interval, and judging that the pulse condition corresponding to the pulse basic data set belongs to real pulse.

12. The pulse classification method of claim 11, wherein the pulse basis data set comprises rising branch slope calculation data; the method specifically comprises the following steps:

if the rising branch slope operation data fall into a tight pulse rising branch slope interval, taking the tight pulse rising branch slope interval as the target pulse type interval;

And judging that the pulse condition corresponding to the pulse basic data set is the tight pulse in the real type.

13. The pulse classification method of claim 12, wherein the pulse basis data set further comprises falling slope calculation data; the method specifically comprises the following steps:

if the rising branch slope operation data simultaneously falls into a real pulse rising branch slope interval and a long pulse rising branch slope interval, comparing the descending slope operation data with the numerical values of the real pulse descending slope interval and the long pulse descending slope interval;

if the descending slope operation data falls into a real pulse descending slope interval, taking the real pulse descending slope interval as the target pulse type interval;

and judging that the pulse condition corresponding to the pulse basic data set is a real pulse in real types.

14. The pulse classification method of claim 1, wherein prior to receiving the pulse basis data set, the method further comprises:

acquiring pulse time data and a pulse space data set which correspond to each other in a first period, and extracting a first maximum time value corresponding to a time pulse maximum value in the pulse time data and a first minimum time value corresponding to a time pulse minimum value in the pulse time data in the first period;

Traversing and acquiring a first space data set and a second space data set which respectively correspond to the first maximum time value and the first minimum time value according to the pulse space data set;

screening data in the first space data set according to an analysis sliding window with a preset step length to obtain pulse intermediate data, and calculating to form the pulse basic data set according to the pulse time data, the first space data set, the second space data set and the pulse intermediate data;

the pulse time data represent the overall pulse change condition of the region to be measured at least two different moments, and the pulse space data represent the pulse condition of the region to be measured at least two different positions.

15. The method of pulse classification as defined in claim 14, further comprising:

traversing the pulse time data, analyzing to obtain cycle data, calculating to obtain peak data according to the time pulse maximum value, and calculating to obtain trough data according to the time pulse maximum value;

traversing the first space data set, extracting to obtain maximum peak data, normalizing the first space data set by taking the maximum peak data as a basis, calculating to obtain a maximum normalized average value, and calculating to obtain rising branch slope data according to the maximum peak data and the first maximum time value;

Traversing the second space data set, extracting to obtain wave trough minimum data, and calculating to obtain resistance coefficient data according to the wave trough minimum data and the first minimum time value.

16. The method of pulse classification as defined in claim 14, further comprising:

fitting according to the first space data set to generate a space pressure curved surface model, traversing and obtaining a space maximum value in the space pressure curved surface model, and extracting a partial region in the space pressure curved surface model according to the analysis sliding window to obtain a pulse analysis curved surface containing the pulse intermediate data;

dividing the pulse analysis curved surface into a curved surface ascending side and a curved surface descending side, and respectively calculating the slopes of the curved surface ascending side and the curved surface descending side relative to the space maximum point corresponding to the space maximum value to obtain ascending slope data and descending slope data.

17. The method of pulse classification as defined in claim 16, further comprising:

calculating the preset step length according to the space maximum value, and determining a plurality of pulse length reference points by taking the space maximum value as a starting point according to an analysis sliding window configured as the preset step length;

Extracting a region surrounded by the pulse length reference points in the space pressure curved surface model to obtain the pulse analysis curved surface;

wherein the preset step length is one fifth of the maximum value of the space.

18. A method for training a pulse classification model, comprising:

receiving pulse preset data and corresponding pulse type preset labels, preprocessing the pulse preset data, and dividing the pulse preset data according to preset proportions to obtain pulse training data, pulse verification data, and corresponding pulse type training labels and pulse type verification labels;

the pulse classification model is called, the pulse training data is used as the pulse basic data set to carry out traversal prediction, and a plurality of groups of model training parameters are obtained;

respectively carrying the model training parameters in the pulse classification model, and carrying out traversal verification by taking the pulse verification data as the pulse basic data set to correspondingly obtain verification result data;

calculating to obtain model evaluation scores according to the verification result data and the pulse type verification labels, and traversing to obtain standard model parameters corresponding to the highest model evaluation scores;

Wherein the pulse classification model is configured to perform the steps of the pulse classification method of any of claims 1-17.

19. The pulse classification model training method of claim 18, wherein the preset ratio sets a data amount of pulse training data to 80% of the data amount of the pulse preset data; according to the verification result data and the pulse type verification tag, calculating to obtain a model evaluation score, traversing to obtain a standard model parameter corresponding to the highest model evaluation score, and specifically comprising the following steps:

calculating to obtain accuracy, precision, recall and F1 fraction according to the verification result data and the pulse type verification tag, and calculating to obtain the model evaluation fraction according to a preset weight coefficient;

the preset weight coefficient sets the weight of the accuracy rate, the weight of the accuracy rate and the weight of the recall rate to be equal and larger than the weight of the F1 fraction.

20. The pulse classification equipment is characterized by comprising a processor, a memory and a communication bus, wherein the processor and the memory are communicated with each other through the communication bus;

the memory is used for storing application programs;

The processor is configured to implement the steps of the pulse classification method according to any one of claims 1-17 when executing an application stored on the memory.

21. A storage medium having stored thereon an application program, wherein the application program, when executed, implements the steps of the pulse classification method of any of claims 1-17.