Intelligent tremor monitoring method and device based on holographic phase Hilbert spectrum analysis

Abstract

The invention provides an intelligent tremor monitoring method based on holographic phase Hilbert spectrum analysis. The method comprises the following steps: acquiring a signal to be analyzed; synchronously acquiring EMG and ACC; performing holographic phase Hilbert spectrum analysis: performing first-layer decomposition on the EMG by adopting a mode decomposition method to obtain a first-layer mode component which comprises a plurality of first-layer mode functions; calculating an envelope of the first-layer modal function to obtain an envelope line; performing second-layer decomposition on the envelope line by adopting a mode decomposition method to obtain a second-layer mode component which comprises a plurality of second-layer mode functions; decomposing the ACC by adopting a mode decomposition method to obtain an ACC mode component which comprises a plurality of ACC mode functions; and constructing a holographic phase Hilbert spectrum for tremor analysis.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of medical data monitoring, and relates to an intelligent tremor monitoring method and device based on holographic phase Hilbert spectrum analysis. BACKGROUND

[0002] Electromyography tremor analysis is an objective method for evaluating the characteristics of patient tremor in clinical practice. It records the information of tremor frequency, amplitude and harmonic resonance through an accelerometer and / or electromyography to reflect the changes of the basal ganglia-thalamus-cerebellum-cortex and electromyography signals. At present, the general method for data analysis and information integration of tremor signals is Fourier Transform (FT), which abstracts the analysis object into a linear time-invariant system, i.e. the input and output signals satisfy the linear relationship, and the system parameters do not change with time. FT also needs to make the above signal assumption when performing tremor analysis. From the pathogenesis and clinical observation, the tremor signals collected by the accelerometer and surface electromyography, especially the pathological tremor signals, have high time-varying and nonlinear characteristics. Therefore, there are certain limitations in applying FT to tremor analysis, mainly including (1) inaccurate frequency in complex cases, (2) prominent harmonic phenomenon, and (3) unable to effectively distinguish the motor unit potential firing frequency. In recent years, the main methods for analyzing nonlinear time-varying signals include Short Time Fourier Transform (STFT), Wavelet Transform (WT), Hilbert-Huang Transform based on Empirical Mode Decomposition (EMD), and further development of Holo-Hilbert Spectral Analysis (HHSA).

[0003] The introduction of holographic Hilbert spectrum analysis solves the modulation problem and reveals all possible coupling frequency pairs involved in cross-frequency coupling (CFC). From the definition of holographic Hilbert spectrum, it can be known that the core is the modulation frequency and the amplitude modulation frequency. The modulation frequency is the frequency of the intrinsic mode function itself, and the amplitude modulation frequency is the frequency of the intrinsic mode function of the envelope of the intrinsic mode function. It is an adaptive data analysis method for analyzing the modulation state of the signal itself, but it cannot be used to analyze the relationship between two signals. In tremor analysis, there is a strong causal relationship between the tremor discharge of the muscle and the fluctuation change of the accelerator. Because of the tremor of the muscle, the fluctuation of the center of gravity of the limb is caused, and this fluctuation of the center of gravity is recorded by the accelerator. However, how to establish the relationship between the electromyogram and the accelerator becomes a problem in analysis. In addition, in tremor analysis, it is also necessary to consider whether the two antagonistic muscles are synchronous contraction or alternate contraction. In the face of many problems in tremor analysis, the holographic Hilbert spectrum analysis in the prior art cannot solve them. SUMMARY

[0004] In order to overcome some problems in the prior art, the present application provides an intelligent tremor monitoring method and device based on holographic phase Hilbert spectrum analysis, which can establish the relationship between the electromyogram and the accelerator signal.

[0005] The first aspect of the present application provides an intelligent tremor monitoring method based on holographic phase Hilbert spectrum analysis, comprising the following steps:

[0006] S100, obtaining a signal to be analyzed; synchronously collecting an electromyogram (EMG) and an accelerator signal (ACC);

[0007] S200, holographic phase Hilbert spectrum analysis, comprising the following steps:

[0008] First layer decomposition: using a modal decomposition method to make a first layer decomposition on the EMG to obtain a first layer modal component, which contains a plurality of first layer modal functions;

[0009] Second layer decomposition: calculating the envelope of the first layer modal function to obtain an envelope line; using a modal decomposition method to make a second layer decomposition on the envelope line to obtain a second layer modal component, which contains a plurality of second layer modal functions;

[0010] ACC decomposition: using a modal decomposition method to make decomposition on the ACC to obtain an ACC modal component, which contains a plurality of ACC modal functions; and

[0011] Constructing the holographic phase Hilbert spectrum for tremor analysis; specifically:

[0012] When the reference phase is contained in the multi-dimensional, the instantaneous phase of the ACC modal function is calculated as the reference phase.

[0013] when the modulation frequency is included in the multi-dimensions, the instantaneous frequency of the first layer modal function is calculated as the modulation frequency;

[0014] when the frequency modulation phase is included in the multi-dimensions, the instantaneous phase of the first layer modal function is calculated as the frequency modulation phase;

[0015] when the amplitude modulation frequency is included in the multi-dimensions, the instantaneous frequency of the second layer modal function is calculated as the amplitude modulation frequency;

[0016] when the amplitude modulation phase is included in the multi-dimensions, the instantaneous phase of the second layer modal function is calculated as the amplitude modulation phase; and,

[0017] based on at least two of the six dimensions of the reference phase, the modulation frequency, the frequency modulation phase, the amplitude modulation frequency, the amplitude modulation phase and the time, a multi-dimensional space energy distribution about the at least two dimensions is obtained, the energy being the amplitude of the envelope of the second layer modal component; thus obtaining the holographic phase Hilbert spectrum,

[0018] the relationship between the EMG and the ACC can be reflected.

[0019] the multi-dimensional space energy distribution can be a multi-dimensional space energy distribution about at least two of the reference phase, the modulation frequency, the frequency modulation phase, the amplitude modulation frequency, the amplitude modulation phase and the time, for example, it can be a two-dimensional space energy distribution about the frequency modulation phase and the reference phase, a two-dimensional space energy distribution about the reference phase and the time, a two-dimensional space energy distribution about the frequency modulation phase, the reference phase and the time, and other four-dimensional, five-dimensional or all six-dimensional and so on. When only the space energy distribution of certain dimensions is obtained, in the step of constructing the holographic phase Hilbert spectrum, the other dimensions can not be calculated; for example, when the obtained holographic phase Hilbert spectrum is a three-dimensional space energy distribution about the reference phase, the amplitude modulation phase and the time, the modulation frequency, the frequency modulation phase and the amplitude modulation frequency do not need to be calculated, that is, the instantaneous frequency of the first layer modal function, the instantaneous phase of the first layer modal function and the instantaneous frequency of the second layer modal function in this step do not need to be calculated, which is easy to understand.

[0020] in an embodiment, in the S200 step:

[0021] in the second layer decomposition, the envelope of at least one first layer modal function in the first layer modal component is calculated to obtain an envelope line; the envelope line is subjected to a second layer decomposition by using a modal decomposition method to obtain a second layer modal component, which contains a plurality of second layer modal functions; the at least one first layer modal function is the most important component of the EMG;

[0022] and,

[0023] constructing the holographic phase Hilbert spectrum, specifically comprising the following steps:

[0024] When the multi-dimension contains a reference phase, calculating the instantaneous phases of the ACC modal components, and selecting the instantaneous phase of one of the ACC modal functions as the reference phase; wherein the selected ACC modal function is the most dominant component of the ACC;

[0025] When the multi-dimension contains a modulation frequency, calculating the instantaneous frequencies of the at least one first-layer modal function as the modulation frequency;

[0026] When the multi-dimension contains a frequency modulation phase, calculating the instantaneous phases of the at least one first-layer modal function as the frequency modulation phase;

[0027] When the multi-dimension contains an amplitude modulation frequency, calculating the instantaneous frequencies of the second-layer modal components, and selecting the instantaneous frequencies of several second-layer modal functions as the amplitude modulation frequency;

[0028] When the multi-dimension contains an amplitude modulation phase, calculating the instantaneous phases of the second-layer modal components, and selecting the instantaneous phase of one of the second-layer modal functions as the amplitude modulation phase; the period of the selected second-layer modal function is closest to the period of an ACC modal function, which is the most dominant component of the ACC; and,

[0029] Based on at least two of the six dimensions of the reference phase, the modulation frequency, the frequency modulation phase, the amplitude modulation frequency, the amplitude modulation phase and the time, obtaining a multi-dimensional space energy distribution about the at least two dimensions, the energy being the amplitude of the envelope of the second-layer modal components; thereby obtaining the holographic phase Hilbert spectrum.

[0030] In an embodiment, the selected at least one first-layer modal function is at least one of a first first-layer modal function and a second first-layer modal function in the first-layer modal components; preferably, both the first first-layer modal function and the second first-layer modal function are selected.

[0031] In an embodiment, the monitoring method further comprises S300 tremor type identification: using a machine learning classification method to establish an identification model for holographic phase Hilbert spectrums of known tremor types, inputting holographic phase Hilbert spectrums of unknown tremor types into the identification model, thereby predicting the corresponding tremor type.

[0032] In an embodiment, the machine learning classification method selects a LightGBM algorithm or an XGBoost algorithm; and the tremor types include essential tremor (ET) and Parkinson tremor (PD).

[0033] In an embodiment, the holographic phase Hilbert spectrum is a multi-dimensional spatial energy distribution with respect to at least one of the reference phase, the amplitude modulation phase and the frequency modulation phase. Alternatively, the holographic phase Hilbert spectrum is at least a multi-dimensional spatial energy distribution with respect to the reference phase. Alternatively, the holographic phase Hilbert spectrum is at least a multi-dimensional spatial energy distribution with respect to the reference phase and the amplitude modulation frequency. Alternatively, the holographic phase Hilbert spectrum is at least a multi-dimensional spatial energy distribution with respect to the reference phase and the amplitude modulation phase. Alternatively, the holographic phase Hilbert spectrum is at least a multi-dimensional spatial energy distribution with respect to the reference phase and the modulation frequency.

[0034] In an embodiment, in the S100 step, the EMG and the ACC can be pre-processed, the pre-processing including at least one of filtering, wave-trapping, outlier removal and electrical noise removal; and in the S200 step, the pre-processed EMG and ACC are subjected to holographic phase Hilbert spectrum analysis.

[0035] In an embodiment, in the first layer decomposition, the EMG is subjected to first layer decomposition by EMD or EEMD, the obtained first layer modal components including a plurality of intrinsic mode functions or a plurality of complete intrinsic mode functions, respectively as the plurality of first layer modal functions; in the second layer decomposition, the envelope is subjected to second layer decomposition by EMD or EEMD, the obtained second layer modal components including a plurality of intrinsic mode functions or a plurality of complete intrinsic mode functions, respectively as the plurality of second layer modal functions; and in the ACC decomposition, the ACC is subjected to decomposition by EMD or EEMD, the obtained ACC modal components including a plurality of intrinsic mode functions or a plurality of complete intrinsic mode functions, respectively as the plurality of ACC modal functions.

[0036] In an embodiment, in the S100 step, the EMG signal is from a pair of antagonistic muscles, the collected EMG signals including EMG1 and EMG2; and in the S200 step, the holographic phase Hilbert spectrum analysis is performed on EMG1 and the ACC respectively, and on EMG2 and the ACC.

[0037] In an embodiment, the S200 step obtains an amplitude-modulation frequency-modulation amplitude-phase-frequency-reference phase-time six-dimensional space energy distribution, the energy being an envelope amplitude of the second layer modal component, thereby obtaining the holographic phase Hilbert spectrum; and integrating the holographic phase Hilbert spectrum in N dimensions to obtain a holographic phase Hilbert spectrum about the remaining 6-N dimensions; N being a positive integer less than 6. The difference between this embodiment and the foregoing is that the foregoing scheme can directly obtain a two-dimensional space energy distribution, a three-dimensional space energy distribution, a four-dimensional space energy distribution, a five-dimensional space energy distribution, or a six-dimensional space energy distribution; but this embodiment first directly obtains a six-dimensional space energy distribution, and then obtains a space energy distribution of other dimensions by integration; although the results are the same, the processes are different.

[0038] In an embodiment, N = 4.

[0039] In an embodiment, the 6-N dimensions include at least one of the amplitude modulation phase, the frequency modulation phase, and the reference phase. Alternatively, the 6-N dimensions include at least the reference phase.

[0040] The second aspect of the present application provides an intelligent tremor monitoring device based on holographic phase Hilbert spectrum analysis, which can implement the monitoring method of any one of the foregoing embodiments, and the monitoring device comprises:

[0041] A data acquisition module configured to synchronously acquire EMG and ACC;

[0042] A data processing module configured for holographic phase Hilbert spectrum analysis, specifically,

[0043] A modal decomposition method is used to perform first layer decomposition on the EMG to obtain first layer modal components, which include a plurality of first layer modal functions;

[0044] An envelope of the first layer modal functions is calculated to obtain an envelope line; a modal decomposition method is used to perform second layer decomposition on the envelope line to obtain second layer modal components, which include a plurality of second layer modal functions;

[0045] A modal decomposition method is used to perform decomposition on the ACC to obtain ACC modal components, which include a plurality of ACC modal functions;

[0046] Instantaneous phases of the first layer modal functions, the second layer modal functions, and the ACC modal functions are respectively calculated, and instantaneous frequencies of the first layer modal functions and the second layer modal functions are respectively calculated; and

[0047] A holographic phase Hilbert spectrum is constructed for tremor analysis; specifically,

[0048] An instantaneous phase of the ACC modal functions is selected as the reference phase;

[0049] selecting the instantaneous frequency of the first layer modal function as the modulation frequency and its instantaneous phase as the frequency modulation phase;

[0050] selecting the instantaneous frequency of the second layer modal function as the amplitude modulation frequency and its instantaneous phase as the amplitude modulation phase; and,

[0051] obtaining a multi-dimensional space energy distribution of at least two dimensions of the six dimensions of the reference phase, the modulation frequency, the frequency modulation phase, the amplitude modulation frequency, the amplitude modulation phase and the time, the energy being the amplitude of the envelope of the second layer modal component; thereby obtaining the holographic phase Hilbert spectrum; and,

[0052] a result output module configured to output the holographic phase Hilbert spectrum to display the monitoring result of the tremor.

[0053] In an embodiment, the monitoring device further comprises a tremor type identification module: wherein the result output module outputs the holographic phase Hilbert spectrum of the known tremor type, trains it using a machine learning classification method to obtain a trained identification model; the result output module outputs the holographic phase Hilbert spectrum of the unknown tremor type and inputs it into the identification model to predict the tremor type.

[0054] In addition, a large number of embodiments in the monitoring method can also be reasonably combined in the monitoring device, which will not be described here.

[0055] The third aspect of the present application provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the monitoring method according to any one of the preceding embodiments when executing the computer program.

[0056] The fourth aspect of the present application provides a computer readable storage medium having a computer program stored thereon, wherein the program is executed by a processor to implement the monitoring method according to any one of the preceding embodiments.

[0057] The monitoring method or monitoring device provided by at least one embodiment of the present application introduces the concept of phase, which can solve the problem of correlation between the existing holographic Hilbert spectrum and the electromyography signal and the accelerometer signal.

[0058] The monitoring method or monitoring device provided by at least one embodiment of the present application introduces the phase information of the ACC, the phase information of the EMG, and the phase information of the EMG envelope into the tremor analysis, and combines the features of the holographic phase Hilbert spectrum with the machine learning method to construct a tremor identification model for diagnosing and identifying the tremor type. BRIEF DESCRIPTION OF DRAWINGS

[0059] Figure 1 is a flowchart of a monitoring method according to an embodiment of the present application;

[0060] Figure 2 is a schematic diagram of a monitoring device according to an embodiment of the present application;

[0061] Figure 3 are the pre-processed data of ACC, EMG1 and EMG2, respectively;

[0062] Figure 4a is the result of the first layer decomposition of EMG1;

[0063] Figure 4b is the result of the second layer decomposition of F-IMF1 of EMG1 (i.e. F1-IMF1);

[0064] Figure 5a are shown are F1-IMF1, S1-IMF3, and RP-ACC-IMF, respectively;

[0065] Figure 5b are shown are F2-IMF1, S2-IMF3, and RP-ACC-IMF, respectively;

[0066] Figure 5c are shown are the instantaneous phase of F1-IMF1, S1-IMF3, and the instantaneous phase of RP-ACC-IMF, respectively;

[0067] Figure 5d are shown are the instantaneous phase of F2-IMF1, S2-IMF3, and the instantaneous phase of RP-ACC-IMF, respectively;

[0068] Figure 6a is the instantaneous frequency of F1-IMF1 and its S1-IMFs;

[0069] Figure 6b is the instantaneous phase of RP-ACC-IMF, and the instantaneous phase of F1-IMF1 and its S1-IMFs;

[0070] Figure 6c is the instantaneous amplitude (amplitude of the envelope) of the S1-IMFs of F1-IMF1;

[0071] Figure 7a is the FM-AM-holographic phase Hilbert spectrum of EMG1;

[0072] Figure 7b is the FM-RP-holographic phase Hilbert spectrum of EMG1;

[0073] Figure 7c is the AM-RP-holographic phase Hilbert spectrum of EMG1;

[0074] Figure 7d is the FM-AM-holographic phase Hilbert spectrum of EMG2;

[0075] Figure 7e is the FM-RP-holographic phase Hilbert spectrum of EMG2;

[0076] Figure 7f is the AM-RP-holographic phase Hilbert spectrum of EMG2;

[0077] Figure 8a is the AMP-RP-holographic phase Hilbert spectrum of EMG1;

[0078] Figure 8b is the diagonalized-phase-holographic phase Hilbert spectrum of EMG1;

[0079] Figure 8c is the centralization-phase-holographic phase Hilbert spectrum of EMG1;

[0080] Figure 8d is the AMP-RP-holographic phase Hilbert spectrum of EMG2;

[0081] Figure 8e is the diagonalized-phase-holographic phase Hilbert spectrum of EMG2;

[0082] Figure 8f is the centralization-phase-holographic phase Hilbert spectrum of EMG2;

[0083] Figure 9 This is the comparison result of evaluation parameters of tremor recognition models based on PSD, Holo and PhaseHolo;

[0084] Figure 10 It is a connection diagram of computer equipment. DETAILED DESCRIPTION

[0085] In order to make the purpose, technical solutions and advantages of this application more clear, the following description and explanation of this application are made in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application. Based on the embodiments provided in this application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0086] It is apparent that the drawings in the following description merely show some examples or embodiments of the present application, and for those skilled in the art, the present application can be applied to other similar situations without creative labor on the basis of these drawings. In addition, it can be understood that although the efforts made in this development process can be complex and lengthy, some changes in design, manufacture or production and the like made on the basis of the technical content disclosed in the present application by those skilled in the art related to the content disclosed in the present application are only routine technical means and should not be understood as insufficient disclosure of the content disclosed in the present application.

[0087] Reference to "an embodiment" in this application means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the application. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment, nor are they necessarily mutually exclusive or alternative embodiments. It is explicitly contemplated that embodiments described in this application can be combined with other embodiments in a non- conflicting manner.

[0088] Unless otherwise defined, technical terms or scientific terms used in the present application shall have the ordinary meaning understood by a person of ordinary skill in the art to which the present application pertains. The terms "a", "an", "one", "this", and similar terms in the present application do not denote a quantity of one, but denote the presence of at least one of something or a single instance of something. The terms "including", "containing", "having" and any variations thereof in the present application are intended to cover a non-exclusive inclusion; for example, a process, method, system, product or device that includes a list of steps or modules (units) is not limited to the listed steps or units, but can also include other steps or units not listed or can also include other steps or units inherent to such a process, method, product or device. The terms "connected", "connected to", "coupled" and similar terms in the present application are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. The term "plurality" in the present application means two or more. The term "and / or" describes the association relationship of the associated objects, which means that there can be three relationships.

[0089] The first embodiment of the present application provides an intelligent tremor monitoring method based on holographic phase Hilbert spectrum analysis (which can be simply referred to as monitoring method); Figure 1is a flow chart of the monitoring method according to the present embodiment. It is to be noted that the steps shown in the flow of the monitoring method or the flow chart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions. The present embodiment is mainly explained with respect to a six-dimensional space energy distribution of amplitude modulation frequency (AM) - modulation frequency (FM) - amplitude modulation phase (AMP) - frequency modulation phase (FMP) - reference phase (RP) - time (T), and other dimensions can be subtracted accordingly, which is easy to understand. The monitoring method comprises the following steps:

[0090] S100 acquisition of a signal to be analyzed:

[0091] An electrode is placed at a muscle tremor site of a subject to be monitored to acquire an electromyography signal (EMG). A tremor accelerator is arranged to synchronously acquire an accelerator signal (ACC); both are data with respect to time T.

[0092] In an embodiment, EMG of a pair of antagonistic muscles can be acquired respectively, for example, EMG1 and EMG2; for example, one is a flexor muscle and the other is an extensor muscle. In a specific embodiment, the pair of antagonistic muscles are located near the elbow of the hand, and the tremor accelerator is placed near the web to acquire the ACC, as shown in Figure 2 In other embodiments, for example, when testing the standing tremor, the pair of antagonistic muscles can also be located in the thigh, and the tremor accelerator can be placed near the knee to acquire the ACC. The acquisition of the electromyography signal EMG and the accelerator signal ACC can be performed by using conventional electrode and accelerator arrangement in medical detection.

[0093] Acquiring EMG of a pair of antagonistic muscles is a conventional operation in tremor testing, and in some diseases, the two muscles contract simultaneously, and in some diseases, they contract alternately. The monitoring method of the present application can quantify whether the two muscles contract synchronously or alternately, and by taking the ACC as a reference, it can be known how much the two muscles are phase-shifted from the ACC, so as to obtain the phase difference of the two muscles and determine whether they contract alternately or synchronously. The tremor electromyography can be alternating or synchronous, and different diseases will have differences.

[0094] In addition, further, the obtained EMG and ACC can be respectively pre-processed to reduce external interference. For example, the data can be pre-processed by filtering, wave trapping, outlier removal and electric noise removal.

[0095] S200 holographic phase Hilbert spectrum analysis:

[0096] S201: a modal decomposition method is used to perform first layer decomposition on the pre-processed EMG to obtain first layer modal components, which contain a plurality of first layer modal functions (F-IMFs).

[0097] Specifically, the EMD or EEMD can be used to decompose the preprocessed EMG to obtain first layer modal components, which contain a plurality of intrinsic mode functions or a plurality of ensemble intrinsic mode functions as F-IMFs.

[0098] S202: Envelopes of at least one of the plurality of first layer modal functions (F-IMFs) are calculated to obtain an envelope curve; and a second layer of modal decomposition is performed on the envelope curve to obtain second layer modal components, which contain a plurality of second layer modal functions (S-IMFs).

[0099] Wherein, the steps of calculating the envelope of any function and obtaining the envelope curve are as follows:

[0100] (1) Take the absolute value of the function;

[0101] (2) Determine all the maximum values of the absolute value of the function;

[0102] (3) Construct the envelope by the natural spline function of all the maximum values to obtain the envelope curve (Envelope).

[0103] Therefore, for any one of the first layer modal functions in the embodiment, (1) take the absolute value of the first layer modal function; (2) determine all the maximum values of the absolute value of the first layer modal function; (3) construct the envelope by the natural spline function of all the maximum values to obtain the envelope curve of the first layer modal function (F-IMFs-Envelope).

[0104] In an embodiment, after obtaining the envelope curve of the first layer modal function, the EMD or EEMD can be used to perform a second layer of decomposition. The EMD or EEMD for data decomposition can also refer to CN119969961A, so as to obtain second layer modal components, which contain a plurality of intrinsic mode functions or a plurality of ensemble intrinsic mode functions as S-IMFs.

[0105] S203: The preprocessed ACC is decomposed by using the modal decomposition method to obtain ACC modal components, which contain a plurality of ACC modal functions (ACC-IMFs).

[0106] Specifically, the ACC can also be decomposed by using an empirical mode decomposition method (EMD) or an ensemble empirical mode decomposition method (EEMD). The decomposition of data by using the EMD or the EEMD can refer to CN119969961A, so as to obtain ACC modal components containing a plurality of intrinsic mode functions or a plurality of ensemble intrinsic mode functions as ACC-IMFs.

[0107] S204: Instantaneous phases of the first layer modal functions (F-IMFs), the second layer modal functions (S-IMFs), and the ACC modal functions (ACC-IMFs) are respectively calculated, and instantaneous frequencies of the first layer modal functions (F-IMFs) and the second layer modal functions (S-IMFs) are respectively calculated.

[0108] In the calculation of the instantaneous phase, an instantaneous phase calculation method such as Hilbert transform or direct quadrature method can be used. Similarly, in the calculation of the instantaneous frequency, an instantaneous frequency calculation method such as Hilbert transform or direct quadrature method can be used.

[0109] S205: A holographic phase Hilbert spectrum is constructed for tremor analysis. Specifically:

[0110] (1) An instantaneous phase of an ACC modal function is selected as a reference phase (RP);

[0111] In an embodiment, the selected ACC modal function corresponding to the reference phase (RP-ACC-IMF) is a modal function with the largest amplitude in the ACC modal components and is the main component of the ACC. Such a selection can make the noise small and the result more reliable.

[0112] (2) An instantaneous frequency of at least one first layer modal function in the first layer modal components is selected as a modulation frequency (FM), and an instantaneous phase thereof is selected as a frequency modulation phase (FMP);

[0113] In an embodiment, the selected at least one first layer modal function is the main component of the EMG. Since the obtained plurality of first layer modal functions are generally arranged from high frequency to low frequency, the main component of the EMG is at high frequency. Therefore, when the main component of the EMG is selected as the at least one first layer modal function, if one is selected, the first first layer modal function or the second first layer modal function can be selected; if two are selected, the first first layer modal function and the second first layer modal function can be selected at the same time, and so on.

[0114] (3) selecting the instantaneous frequency of all second layer intrinsic mode functions (S-IMFs) corresponding to the at least one first layer intrinsic mode function as the amplitude modulation frequency (AM);

[0115] selecting the instantaneous phase of one of the second layer intrinsic mode functions in the second layer modal component as the amplitude modulation phase (AMP); wherein the period of the selected second layer intrinsic mode function is closest to the period of the RP-ACC-IMF;

[0116] (4) obtaining a six-dimensional space energy distribution of the amplitude modulation frequency (AM)-modulation frequency (FM)-amplitude modulation phase (AMP)-frequency modulation phase (FMP)-reference phase (RP)-time (T), and the energy is the amplitude of the envelope of the second layer modal component of the EMG; thereby obtaining the holographic phase Hilbert spectrum of the monitored object, which contains phase information and can reflect the relationship between the EMG and the ACC, achieving the purpose of multi-dimensional tremor monitoring. It is worth noting that since the ACC and the EMG are both time (T) distribution data, the obtained various frequency and phase parameters are also time (T) distribution data, which is well known or easily understood by those skilled in the art.

[0117] Further, the holographic phase Hilbert spectrum can also be integrated in one or more dimensions to obtain an integrated holographic phase Hilbert spectrum. For example, after integration in the AMP, FMP, RP, and T four dimensions, AM and FM are left, and the integrated holographic phase Hilbert spectrum is referred to as the "AM-FM-holographic phase Hilbert spectrum".

[0118] In an embodiment, the holographic phase Hilbert spectrum is integrated in N dimensions to obtain a holographic phase Hilbert spectrum with respect to the remaining 6-N dimensions; N is a positive integer less than 6.

[0119] In an embodiment, the N=4.

[0120] In an embodiment, the 6-N dimensions include at least one of the amplitude modulation phase (AMP), the frequency modulation phase (FMP), and the reference phase (RP). Preferably, at least the reference phase (RP) is included.

[0121] In an embodiment, the holographic phase Hilbert spectrum can be a holographic phase Hilbert spectrum corresponding to a plurality of first layer modal functions of the first layer modal components, and a plurality of second layer modal functions of the second layer modal components; or one or more of them. For example, the first layer modal components have three first layer modal functions, i.e. F-IMF1, F-IMF2 and F-IMF3; envelope calculation is performed on F-IMF1 to obtain an envelope line, and second layer decomposition is performed thereon to obtain a second layer modal component having two second layer modal functions, i.e. S-IMF1.1 and S-IMF1.2; envelope calculation is performed on F-IMF2 to obtain an envelope line, and second layer decomposition is performed thereon to obtain a second layer modal component having four second layer modal functions, i.e. S-IMF2.1, S-IMF2.2, S-IMF2.3 and S-IMF2.4; envelope calculation is performed on F-IMF3 to obtain an envelope line, and second layer decomposition is performed thereon to obtain a second layer modal component having three second layer modal functions, i.e. S-IMF3.1, S-IMF3.2 and S-IMF3.3. When the holographic phase Hilbert spectrum is constructed in S205, it can be constructed based on all the above data (i.e. the holographic phase Hilbert spectrum contains all the data), or it can be constructed only by using part of them, for example, only F-IMF1 and its S-IMF1.3 can be selected for construction; or only F-IMF2 and its S-IMF1.3 can be selected for construction, mainly considering whether their components can reflect the discharge frequency (8-450 Hz) of the electromyogram and whether their envelope periods are closest to the RP-ACC-IMF.

[0122] It is worth understanding that the above described steps can be adjusted according to actual conditions, and not necessarily according to the described steps to perform the related process. For example, in S201-S203, the EMG can be first decomposed into the first layer and the second layer, and then the ACC can be decomposed; or the ACC can be decomposed first, and then the EMG can be decomposed. In the S204 step, the instantaneous frequency can be calculated first, and then the instantaneous phase can be calculated, or vice versa. In the S205 step, the selected steps can also be adjusted, and so on.

[0123] The monitoring method further comprises:

[0124] S300 tremor type identification:

[0125] A classification method of machine learning is used to establish an identification model for the holographic phase Hilbert spectrum of known tremor types, and the holographic phase Hilbert spectrum of unknown tremor types is input into the identification model, so as to identify the corresponding tremor type. The classification method of machine learning can use the classification algorithm in the prior art, such as the LightGBM algorithm and the XGBoost algorithm.

[0126] Specifically: a plurality of different known tremor types of signals to be analyzed are obtained by S100 step, and the holographic phase Hilbert spectrum analysis is performed by S200 step respectively to obtain the holographic phase Hilbert spectrum based on different tremor types; the holographic phase Hilbert spectrum is trained by using the classification method of machine learning, so as to obtain the trained recognition model; then, the unknown tremor type of signal to be analyzed is obtained by S100 step, and the holographic phase Hilbert spectrum analysis is performed by S200 step to obtain the holographic phase Hilbert spectrum thereof; the holographic phase Hilbert spectrum of the unknown tremor type is input into the trained recognition model, so as to predict the tremor type of the unknown tremor type.

[0127] The present application will be described in detail below in combination with more specific embodiments. It should be understood that these embodiments are only some preferred embodiments of the present application and cannot be understood as limiting the protection scope of the present application.

[0128] Example: Intelligent tremor monitoring method based on holographic phase Hilbert spectrum analysis

[0129] S100 acquisition of signal to be analyzed:

[0130] The electromyography (EMG) and the accelerator signal (ACC) of the tremor muscle are obtained; in this embodiment, two electromyography signals EMG1 and EMG2 of a pair of antagonistic muscles are taken as examples for analysis.

[0131] As shown in Figure 2 , two electrodes are respectively attached to a pair of antagonistic muscles at the elbow part (corresponding to flexor and extensor muscles respectively) to collect EMG1 and EMG2 respectively, and a tremor accelerator is placed at the web space of the palm to collect ACC. The first end of the amplifier is connected to the two electrodes and the accelerator to collect amplified EMG and ACC, and the second end is connected to the data interface of the computer end (PC) to transmit to the computer; the computer end (PC) includes a data interface, a data processing module, a PC display (showing real-time data and data analysis report) and the like, which are some conventional arrangements when collecting data.

[0132] The data sampling frequency is 1000 Hz, the data analysis window is set to 10 seconds, and the time T is 10000 points; according to the actual situation, different frequencies and times can also be set.

[0133] Further, in order to reduce the influence of data drift, 1Hz high-pass filtering is performed on ACC, EMG1 and EMG2 respectively; since the main energy density of EMG and ACC is below 450Hz, 450Hz low-pass filtering is performed on ACC, EMG1 and EMG2; in order to remove the electrical noise interference in the environment, 50Hz notch filtering and 100Hz notch filtering are performed on ACC, EMG1 and EMG2; thereby obtaining the preprocessed ACC, EMG1 and EMG2 respectively. As shown inFigure 3 Fig. 1 shows the preprocessed data of the collected 10-second ACC, EMG1 and EMG2.

[0134] S200 holographic phase Hilbert spectrum analysis:

[0135] S201 and S203:

[0136] The preprocessed EMG1 is decomposed by EEMD to obtain first layer modal components, which contain multiple first layer modal functions (F1-IMFs), all of which are ensemble intrinsic modal functions.

[0137] The preprocessed EMG2 is decomposed by EEMD to obtain first layer modal components, which contain multiple first layer modal functions (F2-IMFs), all of which are ensemble intrinsic modal functions.

[0138] The preprocessed ACC is decomposed by EEMD to obtain ACC modal components, which contain multiple ACC modal functions (ACC-IMFs), all of which are ensemble intrinsic modal functions.

[0139] Figure 4a is the first layer modal component obtained after the first layer decomposition of EMG1, which contains nine ensemble intrinsic modal functions, namely F1-IMF1 to F1-IMF9. EMG2 and ACC are understood in the same way.

[0140] S202: Calculate the envelope of each F1-IMF to obtain the corresponding envelope line; decompose the envelope line of each F1-IMF by EEMD to obtain second layer modal components; each second layer modal component contains multiple second layer modal functions (S1-IMFs), all of which are ensemble intrinsic modal functions.

[0141] Calculate the envelope of each F2-IMF to obtain the corresponding envelope line; decompose the envelope line of each F2-IMF by EEMD to obtain second layer modal components; each second layer modal component contains multiple second layer modal functions (S2-IMFs), all of which are ensemble intrinsic modal functions.

[0142] Figure 4b is the second layer modal component obtained after the second layer decomposition of F1-IMF1 in Figure 4a , which contains seven ensemble intrinsic modal functions, namely S1-IMF1.1 to S1-IMF1.7; other F1-IMFs and F2-IMFs can also be decomposed accordingly to obtain corresponding second layer modal components. For example, Figure 4aAs shown, F1-IMF1 is the most important component of EMG1 (the high frequency component of the EMG is the most important), so it is selected as the first layer modal function of EMG1 in this embodiment; in addition, the following is also mainly illustrated by taking F1-IMF1 as an example for convenience. The same is true for EMG2.

[0143] In the above steps, the calculation of the envelope of each F-IMF includes the following steps: taking the absolute value of each F-IMF, determining all the maximum values of the F-IMF absolute value, constructing the envelope through the natural spline function of all the maximum values, and obtaining the envelope curve.

[0144] S204: Calculate the instantaneous frequency and instantaneous phase of F-IMFs (including F1-IMFs and F2-IMFs) and S-IMFs (including S1-IMFs and S2-IMFs) respectively using Hilbert transform or direct quadrature method; calculate the instantaneous phase of ACC-IMFs using Hilbert transform or direct quadrature method.

[0145] Calculate the instantaneous frequency and instantaneous phase of F1-IMF1 and its S1-IMFs respectively, and calculate the instantaneous phase of ACC-IMFs, the results of which are shown in Figure 6a , Figure 6b and Figure 6c ; Figure 6a -c only shows 2000 time series data (T=2000) of 10000 time series data, but is sufficient as an example; wherein,

[0146] The instantaneous frequency of F1-IMF1 and the instantaneous frequency of S1-IMFs (S1-IMF1.1 to S-IMF1.6) of F1-IMF1 are shown in Figure 6a

[0147] The instantaneous phase of the amplitude maximum ACC-F-IMF, the instantaneous phase of F1-IMF1, and the instantaneous frequency of S1-IMFs (S1-IMF1.1 to S1-IMF1.6) are shown in Figure 6b

[0148] The instantaneous amplitude of S1-IMFs (S1-IMF1.1 to S-IMF1.6) of F1-IMF1, i.e. the amplitude of the envelope curve of each S1-IMF, is shown in Figure 6c

[0149] ​​​S205: Constructing the Hilbert spectrum of the holographic phase for tremor analysis. Specifically, take F1-IMF1 of EMG1 as an example for illustration, and the same understanding for EMG2.

[0150] Select the instantaneous phase of one ACC modal function (RP-ACC-IMF) as the reference phase (RP);

[0151] Select the instantaneous frequency and instantaneous phase of F1-IMF1 as the modulation frequency (FM) and the frequency modulation phase (FMP), respectively;

[0152] Select the instantaneous frequency and instantaneous phase of the S1-IMFs of F1-IMF1 as the amplitude modulation frequency (AM) and the amplitude modulation phase (AMP), respectively;

[0153] Wherein, the selected RP-ACC-IMF is the modal function with the largest amplitude in the ACC modal component, that is, it is the most important component.

[0154] In this embodiment, the instantaneous frequency and instantaneous phase of which first layer modal function in the first layer modal component are selected as FM and FMP, mainly considering that the selected first layer modal function is the most important component in the first layer modal component.

[0155] In addition, the instantaneous frequency of all S-IMFs (except the remainder) corresponding to the first layer modal function (F1-IMF1) is selected as AM; the instantaneous phase of the S-IMF in these S-IMFs, whose period is closest to the period of RP-ACC-IMF, is selected as AMP.

[0156] For example, in this embodiment, because F1-IMF1 is the most important component (the high-frequency component of the electromyogram is the most important) in F1-IMFs, the instantaneous frequency and instantaneous phase of F1-IMF1 are selected as FM and FMP, respectively, the instantaneous frequencies of the S1-IMFs (S1-IMF1.1 to S1-IMF1.6) of F1-IMF1 are selected as AM, and the period of S1-IMF1.3 is closest to the period of RP-ACC-IMF, so the instantaneous phase of S1-IMF1.3 is selected as AMP. Of course, multiple F1-IMFs can also be selected, for example, F1-IMF1 and F1-IMF2 are selected at the same time. For the first layer modal function with a small component ratio, it can also not be selected, because its influence in the subsequent is also very small and can be basically ignored; and in the case of large data, the computer calculation space can also be saved, and the calculation speed can be improved. The same understanding for EMG2.

[0157] After the parameters of EMG1 are obtained, the six-dimensional energy distribution of AM-FM-AMP-FMP-RP-T is constructed, where the time T is the whole data length (i.e. 10000), and the corresponding energy is the instantaneous amplitude of the envelope of S1-IMFs, which can be referred to Figure 6c ; thus the holographic phase Hilbert spectrum of EMG1 is obtained.

[0158] The holographic phase Hilbert spectrum of EMG2 can be obtained by replacing EMG1 with EMG2, which will not be described here.

[0159] Furthermore, the holographic phase Hilbert spectrum can be further integrated in one or more dimensions.

[0160] The holographic phase Hilbert spectrum of F-IMFs and S-IMFs will be described below by taking the example of the holographic phase Hilbert spectrum of the remaining two dimensions after the holographic phase Hilbert spectrum of F-IMFs and S-IMFs is integrated in four dimensions; in FIG. 7, Figure 7a -c is the holographic phase Hilbert spectrum of EMG1, Figure 7d -f is the holographic phase Hilbert spectrum of EMG2; in FIG. 8, Figure 8a -c is the holographic phase Hilbert spectrum of EMG1, Figure 8d -f is the holographic phase Hilbert spectrum of EMG2.

[0161] (1) The holographic phase Hilbert spectrum of modulation frequency (FM)-amplitude modulation frequency (AM) is obtained by integrating in the four dimensions of amplitude modulation phase (AMP)-frequency modulation phase (FMP)-reference phase (RP)-time (T), which is called FM-AM-holographic phase Hilbert spectrum. Figure 7a is the FM-AM-holographic phase Hilbert spectrum of EMG1, Figure 7d is the FM-AM-holographic phase Hilbert spectrum of EMG2, which are represented in the form of contour lines. Figure 7a In FIG. 6, the horizontal coordinate is the FM of EMG1, and the vertical coordinate is the AM of EMG1, and it can be seen that the energy of EMG1 is the highest when the FM (64-512 Hz) is in the AM (4 Hz or so), i.e. the discharge frequency of the electromyogram corresponding to EMG1 is 64-512 Hz, and the amplitude modulation frequency is 4 Hz or so, which corresponds to FIG. 6. Figure 7d In FIG. 8, the energy of EMG2 is the highest when the FM (64-512 Hz) is in the AM (4 Hz), i.e. the discharge frequency of the electromyogram corresponding to EMG2 is 64-512 Hz, and the amplitude modulation frequency is 4 Hz or so. FIG. 7 and FIG. 8 correspond to FIG. 6, except that their presentation forms are different.

[0162] (2) Integrating in the four dimensions of amplitude modulation frequency (AM)-amplitude modulation phase (AMP)-frequency modulation phase (FMP)-time (T), the two-dimensional energy spectrum about modulation frequency (FM)-reference phase (RP) is obtained, which is called FM-RP-holographic phase Hilbert spectrum. Figure 7b is the FM-RP-holographic phase Hilbert spectrum of EMG1, Figure 7e is the FM-RP-holographic phase Hilbert spectrum of EMG2. Figure 7b In the figure, the horizontal coordinate is RP, and the vertical coordinate is the FM of EMG1. It can be seen that the energy of the FM (64-512 Hz) of EMG1 is the highest when RP is (π / 4-π / 2). Figure 7e It can be seen from the figure that the energy of the FM (64-512 Hz) of EMG2 is the highest when RP is (π / 4-π / 2). For the convenience of understanding, please refer to Figure 5c , the red solid line with asterisk in the figure is the instantaneous phase of RP-ACC-IMF as the reference phase RP, and the gray solid line is the amplitude variation of F1-IMF1. It can be seen that the maximum amplitude of F1-IMF1 always appears at about π / 4-π / 2 of the instantaneous phase of RP-ACC-IMF.

[0163] (3) Integrating in the four dimensions of modulation frequency (FM)-amplitude modulation phase (AMP)-frequency modulation phase (FMP)-time (T), the two-dimensional energy spectrum about amplitude modulation frequency (AM)-reference phase (RP) is obtained, which is called AM-RP-holographic phase Hilbert spectrum. Figure 7c is the AM-RP-holographic phase Hilbert spectrum of EMG1, Figure 7f is the AM-RP-holographic phase Hilbert spectrum of EMG2. Figure 7c In the figure, the horizontal coordinate is RP, and the vertical coordinate is the AM of EMG1. It can be seen that the energy of the AM (4-64 Hz) of EMG1 is the highest when RP is (π / 4-π / 2). Figure 7f It can be seen from the figure that the energy of the AM (4-64 Hz) of EMG2 is the highest when RP is (π / 4-π / 2).

[0164] (4) Integrating in the four dimensions of amplitude modulation frequency (AM)-modulation frequency (FM)-frequency modulation phase (FMP)-time (T), the two-dimensional energy spectrum about amplitude modulation phase (AMP)-reference phase (RP) is obtained, which is called AMP-RP-holographic phase Hilbert spectrum. Figure 8a is the AMP-RP-holographic phase Hilbert spectrum of EMG1, Figure 8d is the AMP-RP-holographic phase Hilbert spectrum of EMG2, which is represented in the form of p-color map. Figure 8a In the figure, the horizontal coordinate is RP, and the vertical coordinate is AMP. The period of the second layer modal function corresponding to the AMP is closest to the period of RP-ACC-IMF; for example,Figure 5a As shown, the red solid line is RP-ACC-IMF (main component), and the blue solid line is S1-IMF1.3 of F1-IMF1, which has a period closest to RP-ACC-IMF, but the phases are different, that is, the wave crests and troughs do not overlap; as shown Figure 5c As shown, the amplitude of F1-IMF1 is always maximum at the phase of RP-ACC-IMF π / 4~π / 2, and is maximum at the phase of 0π or 2π of the envelope line (amplitude modulation phase), and this feature can also be seen from Figure 8a , but it is not intuitive enough, so the data of the AMP-RP-holographic phase Hilbert spectrum is adjusted, the part above the amplitude modulation phase (AMP) π is rotated 45 degrees to the right, and the part below the amplitude modulation phase (AMP) π is rotated 45 degrees to the left (i.e. diagonalization processing), which is called diagonalization-phase-holographic phase Hilbert spectrum, as shown Figure 8b As shown, there is a phase shift of π / 4~π / 2 between RP and the amplitude modulation phase AMP, and the corresponding objective phenomenon can be seen from Figure 5c . In order to further quantify how RP and AMP wrap the first layer modal function (F1-IMF1) of EMG1, the energy in the two-dimensional space of RP and AMP is arranged from the center to the outside in order from large to small (i.e. centering processing), and the centering-phase-holographic phase Hilbert spectrum is obtained, as shown Figure 8c As shown, the more concentrated the energy is at the center position, the tighter the envelope line of RP-ACC-IMF and F1-IMF1 wraps F1-IMF1, indicating a strong causal relationship between EMG1 and ACC; as shown Figure 5a As shown, the wrapping of F1-IMF1 is achieved by moving the envelope line of RP-ACC-IMF and F1-IMF1. Similarly, Figure 8d -f respectively correspond to the results of EMG2, and the details can be referred to Figure 5b and 5d . Figure 5a -d data comes from Figure 3 .

[0165] Through holographic phase Hilbert spectrum analysis, the amplitude modulation frequency and the modulation frequency in tremor electromyography, the phase shift between electromyography and the accelerator, the strength of the causal relationship between electromyography and the accelerator, and the phase shift information of the two muscles are fully displayed, realizing the relationship analysis between the accelerator and the electromyography signal, and filling the gap in the holographic Hilbert spectrum analysis method in the prior art.

[0166] Through the above steps, the holographic phase Hilbert spectrum of different tremor types can be obtained respectively; wherein the strength and concentration of different tremor types (such as Parkinson's tremor, essential tremor) are different. Therefore, further, the monitoring method further comprises: S300 tremor type identification:

[0167] Identify common essential tremor (ET) and Parkinson tremor (PD). Specifically, select M ET patients and N PD patients from the database; use S100 to obtain the signals to be analyzed (i.e., obtain EMG and ACC) of the M ET patients respectively, and use S200 to analyze and obtain the holographic phase Hilbert spectrum of each patient about ET; similarly, use S100 to obtain the signals to be analyzed of the N PD patients respectively, and use S200 to analyze and obtain the holographic phase Hilbert spectrum of each patient about PD; for the holographic phase Hilbert spectrum of these known ET patients and PD patients, use the LightGBM algorithm to train these holographic phase Hilbert spectrum, thereby obtaining the trained identification model. When the holographic phase Hilbert spectrum of an unknown tremor type is input into the identification model, it can predict the tremor type of the patient, ET or PD, which can be used to assist medical treatment, etc.

[0168] The classification method of machine learning is used to establish an identification model based on the power spectral density method (PSD) in the prior art, the holographic Hilbert spectrum method (Holo), and the holographic phase Hilbert spectrum method (PhaseHolo) in the present application, respectively. For the data in the database, the leave-one-out method is used for training and testing, that is, one data is left as the test set each time, and the others are used as the training set, until the prediction of all data is completed. Based on the prediction results, the ROC-AUC is calculated, and based on the prediction results, the confusion matrix is constructed to calculate the Accuracy, Precision Sensitivity (Recall), Specificity, F1-Score, as the evaluation parameters of the model.

[0169] As shown in Figure 9 , the accuracy (Accuracy) and ROC-AUC of the model constructed based on the holographic phase Hilbert spectrum of the present application are 98.5% and 1, respectively, which are 1.5% and 0.01 higher than Holo, and 6% and 0.04 higher than PSD. It can be seen that the holographic phase Hilbert spectrum provided by the present application has outstanding advantages in tremor monitoring.

[0170] The AM-RP-holographic phase Hilbert spectrum and the centralized-phase-holographic phase Hilbert spectrum in the above results are used for training of a tremor classification model. The tremor classification model is constructed based on the holographic phase Hilbert spectrum of different tremor types, and the specific features are the AM-RP-holographic phase Hilbert spectrum and the centralized-phase-holographic phase Hilbert spectrum. Based on the existing data set statistics, these two features are the best in distinguishing different tremor types; and a machine learning method is used to learn and identify the features of different tremor types. When constructing the classification model for two tremor types (ET, PD), the evaluation parameters of the model based on different input features are different, and the evaluation parameter of the model based on the holographic phase Hilbert spectrum provided in the embodiment is the best.

[0171] Further, the above results can be further sorted, the data and results can be archived, the results can be displayed on the screen, and a corresponding data analysis report can be generated.

[0172] The second embodiment of the present application provides an intelligent tremor monitoring device (which can be referred to as a monitoring device) based on holographic phase Hilbert spectrum analysis. The device is used to implement the monitoring method described in any of the preceding embodiments. The monitoring device comprises:

[0173] A data acquisition module configured to acquire an electromyography signal EMG using an electrode, and simultaneously acquire an accelerometer signal ACC using a tremor accelerometer;

[0174] A data processing module configured to:

[0175] Perform first layer decomposition on the EMG using a modal decomposition method to obtain first layer modal components, which include a plurality of first layer modal functions;

[0176] Calculate the envelope of at least one of the plurality of first layer modal functions to obtain an envelope line, and perform second layer decomposition on the envelope line using a modal decomposition method to obtain second layer modal components, which include a plurality of second layer modal functions;

[0177] Perform decomposition on the ACC using a modal decomposition method to obtain ACC modal components, which include a plurality of ACC modal functions;

[0178] Calculate the instantaneous phase of the first layer modal function, the second layer modal function, and the ACC modal function, respectively, and calculate the instantaneous frequency of the first layer modal function and the second layer modal function, respectively;

[0179] Construct a holographic phase Hilbert spectrum for tremor analysis. Specifically:

[0180] (1) Select an instantaneous phase of an ACC modal function as a reference phase;

[0181] (2) selecting the instantaneous frequency of at least one first-layer modal function in the first-layer modal component as the modulation frequency, and the instantaneous phase as the frequency modulation phase;

[0182] (3) selecting the instantaneous frequencies of all second-layer modal functions (the remainder can be removed) corresponding to the at least one first-layer modal function as the amplitude modulation frequencies, and selecting the instantaneous phase of one of the second-layer modal functions in the second-layer modal component as the amplitude modulation phase, wherein the period of the selected second-layer modal function is closest to the period of the ACC modal function as the reference phase;

[0183] (4) obtaining the six-dimensional space energy distribution of the amplitude modulation frequency-modulation frequency-amplitude modulation phase-frequency modulation phase-reference phase-time, and the energy is the amplitude of the envelope line of the second-layer modal component, thereby obtaining the holographic phase Hilbert spectrum.

[0184] The result output module is configured to output the holographic phase Hilbert spectrum to display the monitoring result.

[0185] In some embodiments, the monitoring device further comprises a tremor type identification module, wherein the result output module outputs the holographic phase Hilbert spectrum of the known tremor type, trains it by using a machine learning classification method to obtain a trained identification model; the result output module outputs the holographic phase Hilbert spectrum of the unknown tremor type, and inputs it into the identification model to predict the tremor type.

[0186] In some embodiments, the holographic phase Hilbert spectrum can also be integrated in one or more dimensions to obtain an integrated holographic phase Hilbert spectrum.

[0187] In some embodiments, the monitoring device can further comprise:

[0188] The data archiving module is configured to store data for future reference.

[0189] The data display module is configured to display the monitoring result and the comparison result on a screen. Therefore, the data display module can comprise a PC display.

[0190] It is worth understanding that the technical features described in the monitoring method can also be reasonably applied to the monitoring device of the present embodiment, and those that have been described will not be repeated. It should be noted that the above-mentioned modules can be functional modules or program modules, and can be implemented by software or hardware. For modules implemented by hardware, the above-mentioned modules can be located in the same processor, or the above-mentioned modules can also be located in different processors in any combination.

[0191] The third implementation of the present application provides a computer device, which can include a processor 101 and a memory 102 having computer program instructions stored therein, as shown in Figure 10

[0192] Specifically, the processor 101 can include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement one or more embodiments of the present application. The memory 102 can include a mass storage for data or instructions. By way of example, and not limitation, the memory 102 can include a hard disk drive (HDD), a floppy disk drive, a solid-state drive (SSD), a flash drive, a compact disc (CD) or other optical disk, a tape drive, a universal serial bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory 102 can include removable or non-removable (or fixed) media. Where appropriate, the memory 102 can be internal or external to the data processing device.

[0193] The memory 102 can be used to store or cache various data files needed for processing and / or communication, and possible computer program instructions executed by the processor 101.

[0194] The processor 101 reads and executes the computer program instructions stored in the memory 102 to implement the monitoring method described in any of the preceding embodiments.

[0195] In some embodiments, the computer device can further include a communication interface 103 and a bus 104. As shown in Figure 10 , the processor 101, the memory 102, and the communication interface 103 are connected through the bus 104 and complete communication with each other. The communication interface 103 is used to implement communication between various modules, devices, units and / or equipment in the embodiments of the present application. The communication interface 103 can also be used to communicate data with other components.

[0196] The fourth implementation of the present application provides a computer readable storage medium having computer program instructions stored thereon; the computer program instructions are executed by a processor to implement the monitoring method described in any of the preceding embodiments.

[0197] ​The technical features of the above-described embodiments can be combined in any manner. For the sake of brevity, not all possible combinations of the technical features in the above-described embodiments are described, however, as long as the combination of the technical features does not result in a contradiction, it should be considered to be within the scope of the present disclosure.

[0198] The above-described embodiments only express several embodiments of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the application. It should be noted that for ordinary skilled persons in the art, under the premise of not departing from the concept of the present application, a number of modifications and improvements can be made, which are all within the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. An intelligent tremor monitoring method based on holographic phase Hilbert spectrum analysis, characterized in that: The following steps are involved: S100: Acquisition of a signal to be analyzed; Synchronously collect electromyographic signals (EMG) and accelerometer signals (ACC); S200 holographic phase Hilbert spectrum analysis, which includes the following steps: First layer decomposition: Use the modal decomposition method to decompose the EMG into the first layer to obtain the first layer modal components. It contains multiple first-layer modal functions; Second-level decomposition: Calculate the envelope of the first-level modal function to obtain the envelope line; use the modal decomposition method to perform the second-level decomposition on the envelope line to obtain the second-level modal components, which contain multiple second-level modal functions; ACC decomposition: Decomposing the ACC using modal decomposition to obtain ACC modal components, which contain multiple ACC modal functions; and The holographic phase Hilbert spectrum is constructed for vibration analysis.

2. The monitoring method according to claim 1, characterized in that: In step S200, the step of constructing the holographic phase Hilbert spectrum specifically includes: When the multi-dimensionality includes a reference phase, calculating the instantaneous phase of the ACC mode function as the reference phase; When the multi-dimensional matrix includes a modulation frequency, the instantaneous frequency of the first-layer modal function is calculated as the modulation frequency; When the multi-dimensionality includes a frequency modulation phase, calculating the instantaneous phase of the first layer mode function as the frequency modulation phase; When the multi-dimensional data includes an amplitude modulation frequency, the instantaneous frequency of the second-layer modal function is calculated as the amplitude modulation frequency; When the multi-dimensionality includes an amplitude modulation phase, calculating the instantaneous phase of the second-layer modal function as the amplitude modulation phase; and Based on at least two of the six dimensions of the reference phase, modulation frequency, frequency modulation phase, amplitude modulation frequency, amplitude modulation phase and time, a multidimensional spatial energy distribution about the at least two dimensions is obtained, This energy is the amplitude of the envelope of the second-layer modal component; thus, the holographic phase Hilbert spectrum is obtained, which can reflect the relationship between EMG and ACC.

3. The monitoring method according to claim 1, characterized in that In the step S200: In the second-level decomposition, the envelope of at least one first-level modal function in the first-level modal components is calculated to obtain an envelope line; the envelope line is subjected to a second-level decomposition using a modal decomposition method to obtain the second-level modal components, which include a plurality of second-level modal functions; the at least one first-level modal function is the most important component of the EMG; and Constructing the holographic phase Hilbert spectrum specifically includes the following steps: When the multi-dimensional model includes a reference phase, the instantaneous phase of the ACC modal component is calculated, and the instantaneous phase of one of the ACC modal functions is selected as the reference phase; wherein the selected ACC modal function is the most important component of the ACC; When the multi-dimensionality includes a modulation frequency, calculating the instantaneous frequency of the at least one first-layer modal function as the modulation frequency; When the multi-dimensionality includes a frequency modulation phase, calculating the instantaneous phase of the at least one first-layer modal function as the frequency modulation phase; When the multi-dimensional data includes amplitude modulation frequency, the instantaneous frequency of the second-layer modal component is calculated, and the instantaneous frequencies of several second-layer modal functions are selected as the amplitude modulation frequency; When the multidimensional data includes an amplitude modulation phase, the instantaneous phase of the second-layer modal component is calculated, and the instantaneous phase of one of the second-layer modal functions is selected as the amplitude modulation phase; the period of the selected second-layer modal function is closest to the period of an ACC modal function, and the ACC modal function is the most important component of the ACC; and Based on at least two of the six dimensions of the reference phase, modulation frequency, frequency modulation phase, amplitude modulation frequency, amplitude modulation phase and time, a multidimensional spatial energy distribution about the at least two dimensions is obtained, and the energy is the amplitude of the envelope of the second-layer modal component; thereby obtaining the holographic phase Hilbert spectrum.

4. The monitoring method according to claim 3, characterized in that: The selected at least one first-layer modal function is at least one of a first first-layer modal function and a second first-layer modal function in the first-layer modal component.

5. The monitoring method according to any one of claims 1 to 4, characterized in that: It also includes the identification of S300 tremor types: a machine learning classification method is used to establish an identification model for the holographic phase Hilbert spectrum of known tremor types, and the holographic phase Hilbert spectrum of unknown tremor types is input into the identification model to predict the corresponding tremor type.

6. The monitoring method according to claim 5, characterized in that: The machine learning classification method selects the LightGBM algorithm or the XGBoost algorithm; the tremor types include essential tremor and Parkinson's tremor.

7. The monitoring method according to any one of claims 1 to 4 and 6, characterized in that: The holographic phase Hilbert spectrum is a multi-dimensional spatial energy distribution with respect to at least one of a reference phase, an amplitude modulation phase, and a frequency modulation phase.

8. The monitoring method according to any one of claims 1 to 4 and 6, characterized in that: In step S100, the EMG and ACC are preprocessed, and the preprocessing includes at least one of filtering, notching, outlier removal, and electrical noise removal; in step S200, the preprocessed EMG and ACC are subjected to holographic phase Hilbert spectrum analysis; In the first-layer decomposition, EMD or EEMD is used to perform a first-layer decomposition on the EMG, and the obtained first-layer modal components include multiple intrinsic mode functions or multiple collective intrinsic mode functions, which are respectively used as the multiple first-layer modal functions; in the second-layer decomposition, EMD or EEMD is used to perform a second-layer decomposition on the envelope, and the obtained second-layer modal components include multiple intrinsic mode functions or multiple collective intrinsic mode functions, which are respectively used as the multiple second-layer modal functions; in the ACC decomposition, EMD or EEMD is used to decompose the ACC, and the obtained ACC modal components include multiple intrinsic mode functions or multiple collective intrinsic mode functions, which are respectively used as the multiple ACC modal functions; In step S100, the electromyographic signals come from a pair of antagonistic muscles, and the collected electromyographic signals include EMG1 and EMG2; in step S200, holographic phase Hilbert spectrum analysis is performed on EMG1 and ACC, and holographic phase Hilbert spectrum analysis is performed on EMG2 and ACC.

9. The monitoring method according to any one of claims 2 to 4, characterized in that: In step S200, a six-dimensional spatial energy distribution of amplitude modulation frequency-modulation frequency-amplitude modulation phase-frequency modulation phase-reference phase-time is obtained, where the energy is the amplitude of the envelope of the second-layer modal component, thereby obtaining the holographic phase Hilbert spectrum; the holographic phase Hilbert spectrum is integrated over N dimensions to obtain holographic phase Hilbert spectra for the remaining 6-N dimensions; N is a positive integer less than 6; and the 6-N dimensions include at least the reference phase.

10. An intelligent tremor monitoring device based on holographic phase Hilbert spectrum analysis, capable of implementing the monitoring method according to any one of claims 1 to 9, characterized in that: The monitoring device comprises: A data acquisition module is configured to synchronously acquire EMG and ACC; The data processing module is configured for holographic phase Hilbert spectrum analysis, specifically, The modal decomposition method is used to decompose the EMG into the first layer to obtain the first layer modal components, which contain multiple first layer modal functions; Calculate the envelope of the first-layer modal function to obtain an envelope line; perform a second-layer decomposition of the envelope line using a modal decomposition method to obtain a second-layer modal component, which contains multiple second-layer modal functions; The modal decomposition method is used to decompose the ACC to obtain the ACC modal components, which contain multiple ACC modal functions; Calculate the instantaneous phase of the first layer mode function, the second layer mode function and the ACC mode function respectively, and Calculate the instantaneous frequencies of the first layer modal function and the second layer modal function respectively; and, Construct a holographic phase Hilbert spectrum for tremor analysis; specifically, The instantaneous phase of the ACC mode function is selected as the reference phase; Select the instantaneous frequency of the first layer modal function as the modulation frequency, and its instantaneous phase as the FM phase; Selecting the instantaneous frequency of the second layer mode function as the amplitude modulation frequency, and its instantaneous phase as the amplitude modulation phase; and, Obtaining a multidimensional spatial energy distribution in at least two dimensions of the six dimensions of reference phase, modulation frequency, frequency modulation phase, amplitude modulation frequency, amplitude modulation phase, and time, wherein the energy is the amplitude of the envelope of the second-layer modal component; thereby obtaining a holographic phase Hilbert spectrum; and The result output module is configured to output the holographic phase Hilbert spectrum to display the monitoring result of the tremor.

11. The monitoring device according to claim 10, characterized in that: It also includes a tremor type identification module: wherein the result output module outputs the holographic phase Hilbert spectrum of the known tremor type, and trains it using a machine learning classification method to obtain a trained identification model; the result output module outputs the holographic phase Hilbert spectrum of the unknown tremor type, and inputs it into the identification model to predict the tremor type.

12. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the intelligent tremor monitoring method based on holographic phase Hilbert spectrum analysis according to any one of claims 1 to 9 is implemented.

13. A computer-readable storage medium having a computer program stored thereon, wherein when the program is executed by a processor, the intelligent tremor monitoring method based on holographic phase Hilbert spectrum analysis according to any one of claims 1 to 9 is implemented.

Images