CN110946556B - Parkinson resting state tremor evaluation method based on wearable somatosensory network - Google Patents

Parkinson resting state tremor evaluation method based on wearable somatosensory network Download PDF

Info

Publication number
CN110946556B
CN110946556B CN201911375827.XA CN201911375827A CN110946556B CN 110946556 B CN110946556 B CN 110946556B CN 201911375827 A CN201911375827 A CN 201911375827A CN 110946556 B CN110946556 B CN 110946556B
Authority
CN
China
Prior art keywords
tremor
parkinson
state
data
angle
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201911375827.XA
Other languages
Chinese (zh)
Other versions
CN110946556A (en
Inventor
庄伟�
申义贤
李露
张杰锋
戴栋
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing University of Information Science and Technology
Original Assignee
Nanjing University of Information Science and Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nanjing University of Information Science and Technology filed Critical Nanjing University of Information Science and Technology
Priority to CN201911375827.XA priority Critical patent/CN110946556B/en
Publication of CN110946556A publication Critical patent/CN110946556A/en
Application granted granted Critical
Publication of CN110946556B publication Critical patent/CN110946556B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/40—Detecting, measuring or recording for evaluating the nervous system
    • A61B5/4076—Diagnosing or monitoring particular conditions of the nervous system
    • A61B5/4082—Diagnosing or monitoring movement diseases, e.g. Parkinson, Huntington or Tourette
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
    • A61B5/11—Measuring movement of the entire body or parts thereof, e.g. head or hand tremor, mobility of a limb
    • A61B5/1101—Detecting tremor
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
    • A61B5/11—Measuring movement of the entire body or parts thereof, e.g. head or hand tremor, mobility of a limb
    • A61B5/1121—Determining geometric values, e.g. centre of rotation or angular range of movement
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A61B5/389—Electromyography [EMG]
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7271—Specific aspects of physiological measurement analysis
    • A61B5/7282—Event detection, e.g. detecting unique waveforms indicative of a medical condition

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Medical Informatics (AREA)
  • Molecular Biology (AREA)
  • Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • Biophysics (AREA)
  • Public Health (AREA)
  • Veterinary Medicine (AREA)
  • Pathology (AREA)
  • Physiology (AREA)
  • Neurology (AREA)
  • Artificial Intelligence (AREA)
  • Psychiatry (AREA)
  • Signal Processing (AREA)
  • Dentistry (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Neurosurgery (AREA)
  • Geometry (AREA)
  • Developmental Disabilities (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)

Abstract

The invention discloses a Parkinson resting state tremor evaluation method based on a wearable motion sensing network, belongs to the field of wireless sensor networks and data analysis thereof, and particularly relates to a Parkinson patient arm tremor state acquisition and identification method based on the wearable motion sensing network. The invention calculates the angle variation of the elbow joint and the wrist joint by measuring the attitude angles of the upper arm, the lower arm and the wrist, extracts the characteristic of the angle variation, extracts the real-time characteristic of the electromyographic signal, trains a hidden Markov model according to the characteristic data and a UPDRS scale, and outputs the current optimal state sequence. The method can provide technical support for the evaluation of the tremor degree of the arms of the Parkinson patients, and provides theoretical basis for the population such as the Parkinson patients, the old people and the weak people who need to know the occurrence of the early Parkinson disease in time.

Description

Parkinson resting state tremor evaluation method based on wearable somatosensory network
Technical Field
The invention relates to the field of wireless sensor networks and data analysis thereof, in particular to a Parkinson resting state tremor evaluation method based on a wearable somatosensory network.
Background
Parkinson's disease is a degenerative disease of the central nervous system characterized by affecting motor abilities to varying degrees. It is more common in the elderly, most of which occur after the age of 50. Parkinson's disease affects primarily four motor functions, including bradykinesia, resting tremor, stiffness of the limbs and postural instability. Resting tremor is prominent in the upper limbs, and this tremor gradually spreads from one side to both sides, and the condition gradually worsens. Therefore, it is necessary to recognize resting tremor earlier in daily life.
The application of wearable sensors in early tremor assessment is currently a research hotspot in academia and industry. The wearable sensor can be used for high-precision tracking of a human body, long-term physiological signal monitoring and the like, and the non-implantable monitoring is used for evaluating the motion abnormality of a clinical patient. However, most of the prior art solutions use accelerometers and gyroscopes to directly acquire tremor signals, and such solutions have disadvantages that motion components in daily activities are easily mixed, and the accumulated errors of sensors and environmental noise make the measured signals unreliable. Furthermore, in the early stages of the condition, the tremor of the limbs is very slight, and its features are difficult to extract from the signals acquired by accelerometers and gyroscopes, and are easily mistaken for noise signals and discarded. Therefore, the addition of the electrical muscle signals at the onset of tremor to the assessment system is a better approach in early stage disorders. The surface electromyographic sensor directly deployed on the arm can output real-time electromyographic signals, and the early tremor state can be identified by analyzing the characteristics of the electromyographic signals.
The invention patent CN201410381634.6 discloses a wearable system for quantitatively detecting the main symptoms of a Parkinson's disease person, acceleration and angular acceleration of fingertips and wrists are obtained through wearable intelligent gloves, tremor degree and muscle stiffness state are described through characteristics, and due to the fact that inertial data are directly adopted, interference of postures and daily movement of a patient is easily caused, and therefore the recognition effect of the system is influenced.
The invention patent CN201410833652.3 discloses a Parkinson patient tremor symptom quantitative evaluation method based on approximate entropy and mutual approximate entropy, which comprises the steps of thumb tremor data acquisition, forefinger tremor data acquisition and unified Parkinson disease score scale UPDRS scoring under appointed actions, and mutual approximate entropy calculation between tremor data approximate entropy and tremor data, but myoelectric signal characteristics are not added in the method, and the recognition effect on early-stage slight tremor is weak.
Disclosure of Invention
The invention aims to: the invention provides a Parkinson patient arm resting state tremor evaluation method based on a wearable somatosensory network, which realizes accurate identification of arm tremor degree under daily behaviors and solves the following technical problems: the daily behavior of the patient affects the recognition result of the arm tremor state, and the system recognition positive detection rate is reduced; the early tremor is slight and can be easily mistaken for environmental noise and cannot be accurately identified.
The technical scheme is as follows: in order to realize the purpose of the invention, the technical scheme adopted by the invention is as follows: calculating the angle variation of the wrist joint and the elbow joint through attitude data collected by a sensor, and extracting time domain and frequency domain characteristics of the variation for representing tremor grades of different degrees; calculating the signal characteristics by extracting surface electromyographic signals deployed at the biceps brachii of the upper arm for representing tremor grades of different degrees; a multi-hidden-state-conversion HMM model is designed by combining a UPDRS rating scale and combining angle features and myoelectric features, and is used for tremor state evaluation under daily behaviors. The invention discloses a Parkinson patient arm resting state tremor evaluation method based on a wearable somatosensory network, which specifically comprises the following steps:
step 1, deploying sensors at hands and arms of a human body, initializing the sensors, eliminating zero drift of the sensors, carrying out initial calibration, and setting sampling frequency of the sensors; acquiring human body posture signals and human body surface electromyogram signals through deployed sensors;
step 2, in the daily movement of the Parkinson disease patient, low-frequency components (the frequency is less than 3Hz) are derived from conscious movement of the patient, and high-frequency components comprise tremor and noise of movement intervals; therefore, according to the resting state tremor frequency response characteristics, the posture signal and the electromyographic signal are filtered through the band-pass filter respectively to obtain a filtered posture signal and an electromyographic signal, and the filtered signals retain a posture component and an electromyographic signal component caused by tremor; calculating the variation of the attitude angle according to the attitude signal;
step 3, respectively storing the posture angle variation discrete data and the electromyographic signal discrete data output in the step 2 in a data set
Figure BDA0002340929680000021
Performing the following steps; since the early and late stages of the acquisition are easily affected by the test preparation and test stop state transition, the data sets are respectively rejected according to the time axis
Figure BDA0002340929680000023
And
Figure BDA0002340929680000022
the data of p% before and after the time axis, the data of p% to 1-p% is kept, and the data set is updated
Figure BDA0002340929680000025
And
Figure BDA0002340929680000024
preferably, the value of p% is 10%;
step 4, collecting the data set
Figure BDA0002340929680000026
And
Figure BDA0002340929680000027
respectively carrying out equal-step windowing treatment, and collecting the treated data sets
Figure BDA0002340929680000028
And
Figure BDA0002340929680000029
combining, wherein q% is selected as a training set for HMM model parameter learning, and the rest 1-q% is selected as a test set for HMM model identification;
preferably, to continuously describe the variation in tremor, the data sets are combined
Figure BDA00023409296800000210
And
Figure BDA00023409296800000211
the step length of the data transmission is set to be 100ms, the window duration is 2s, and the data overlapping rate is 95%; the value of q% is 70%;
step 5, extracting attitude angle change time domain and frequency domain characteristics of the Parkinson tremor in a resting state and electromyographic signal characteristics of the training set and the test centralized windowed data respectively;
step 6, defining an HMM model hidden state set according to a UPDRS Parkinson rating scale; respectively setting state labels for the data segments according to the features extracted from the training set;
step 7, initializing HMM model parameters, taking the characteristic parameters extracted from the training set as a training observation sequence, performing model training, and calculating the optimal parameters of the HMM model to obtain a trained HMM model; the characteristics comprise angle change characteristics and electromyographic signal characteristics;
and 8, inputting the characteristic parameters extracted from the test set into the trained HMM model as test observation sequences, solving the hidden state sequence when the probability of the observation sequences is maximum, and outputting the current optimal hidden state to obtain the tremor state evaluation result.
Further, in the step 1, an IMU sensor is deployed on the upper arm, the lower wall and the hand respectively, and an EMG sensor is deployed on the upper arm at the same time; collecting human body posture data through an IMU sensor, and collecting human body surface muscle electric signals through an EMG sensor; the IMU sensor comprises a three-axis accelerometer, a three-axis gyroscope and a three-axis magnetometer, and attitude signals acquired by the IMU sensor are output in a quaternion form. Preferably, the IMU sensor sampling frequency is set to 100Hz and the EMG sensor sampling frequency is set to 50 Hz. Preferably, a nine-axis IMU sensor model GY953 is selected, and an EMG sensor model OYmotion is selected.
Further, in the step 2, the upper and lower cut-off frequencies of the band-pass filter through which the posture signal passes are 3Hz and 6Hz, and the upper and lower cut-off frequencies of the band-pass filter through which the myoelectric signal passes are 3Hz and 12 Hz.
Further, in step 2, the current coordinate system is set to be an earth coordinate system x, y, z, and the attitude angle variation is calculated by the following method:
defining an attitude angle θ1,θ2,θ3Change amount of attitude angle Δ θ1,△θ2,△θ3Wherein theta1,θ2,θ3Respectively an elbow joint bending-stretching angle, a lower arm inward rotation-outward rotation angle, a wrist joint bending-stretching angle, delta theta1,△θ2,△θ3The elbow joint bending-stretching angle variation, the lower arm inner rotation-outer rotation angle variation and the wrist joint bending-stretching angle variation are respectively;
based on the obtained filtered attitude signal,namely, the quaternions of the attitude quaternions of the IMU sensors of the upper arm, the lower arm and the hand respectively calculate the projection alpha of the attitude angles of the IMU sensors of the upper arm, the lower arm and the hand on the x, y and z axes of a coordinate systemx,αy,αz、βx,βy,βz、γx,γy,γz；
From the arm geometry:
△θ1x＝△αx+△βx△θ3x＝△βx+△γx
△θ1y＝△αy+△βy，△θ3y＝△βy+△γy
△θ1z＝△αz+△βz△θ3z＝△βz+△γz
wherein Δ θ1x,△θ1y,△θ1z、△θ3x,△θ3y,△θ3zRespectively, elbow joint bending-stretching angle variation delta theta1Variation quantity delta theta of bending-stretching angle of wrist joint3Projection on the x, y, z axes of the coordinate system; delta alphax,△αy,△αzThe angle variation of the projection of the upper arm attitude angle on the x, y and z axes; delta betax,△βy,△βzThe angle variation of the projection of the attitude angle of the lower arm on the x, y and z axes; delta gammax,△γy,△γzThe angle variation of the projection of the hand posture angle on the x, y and z axes;
separately calculating the variation quantity Delta theta of the attitude angle1,△θ3The expression is as follows:
Figure BDA0002340929680000031
Figure BDA0002340929680000032
since the rotation of the lower arm is about the elbow and wrist axes, the rotation of the lower arm is not limited to the rotation of the lower arm,attitude angle variation quantity [ Delta ] [ theta ]2Derived directly from the quaternion output by the lower arm IMU sensor. According to the conversion relation between the quaternion and the attitude angle, the output quaternion can be converted into the angle variation.
Further, in the step 5, the posture angle change time domain and frequency domain features of the parkinson tremor in the resting state are extracted to describe the tremor degree, specifically as follows:
the method is characterized in that: the angular change of the Parkinson vibration in the rest state is obvious, so the average angular change rate of the Parkinson vibration in the angular change time domain of the rest state is adopted
Figure BDA0002340929680000041
Describing the trembling degree, and respectively calculating the attitude angle theta1,θ2,θ3Average angular rate of change of
Figure BDA0002340929680000042
Units are degree/second; the calculation formula is as follows:
Figure BDA0002340929680000043
where n denotes the data length, n ═ t × fsT is window duration, fsSampling frequency for IMU sensor;
and (2) feature: resting tremor in the Parkinson's disease usually shows involuntary movement of upper limbs of a body in a certain frequency range, so that the tremor degree is described by taking the energy of the Parkinson's tremor in an angle change frequency domain of a resting state as a characteristic;
and (3) performing discrete Fourier transform on the attitude signal data in the training set, converting the attitude signal data into frequency domain data, and calculating an energy value:
Figure BDA0002340929680000044
wherein E is an energy value, f represents a frequency point, and P (f) is a signal energy spectral density with the unit of dB/second; a and b are respectively the boundary values of the frequency range of resting tremor; [ a, b ] takes the value of the typical frequency range of resting state tremor of 3-6 Hz;
and (3) characteristic: the irregularity degree or the chaos degree of the information entropy description information can be known from the characteristics of the attitude signal frequency spectrum, and the information entropy in the frequency domain has better discrimination, so that the information entropy of the angle change of the Parkinson's tremor in the resting state is used as the characteristic description tremor degree; the information entropy T is defined as follows:
Figure BDA0002340929680000045
wherein
Figure BDA0002340929680000046
And taking 2 as a base for the log of the occurrence probability corresponding to the frequency point f.
Further, in the step 5, extracting the electromyographic signal characteristics to describe parkinsonism tremor specifically as follows:
and (4) feature: the electromyographic signals of the Parkinson tremor show that the peaks of partial areas are steep when the tremor occurs, so that the electromyographic signal kurtosis coefficient is selected as a characteristic to describe the Parkinson tremor; the kurtosis coefficient k is defined as follows:
Figure BDA0002340929680000047
wherein mu is a data mean value, lambda is a mathematical expectation, sigma is a standard deviation, and x is electromyographic signal discrete data;
and (5) feature: the electromyographic signals of the Parkinson tremor are also shown in the fact that partial areas of the Parkinson tremor are subjected to intensive oscillation when the tremor occurs, namely the signal has high fluctuation density, and therefore the zero crossing rate of the electromyographic signals is selected as a characteristic to describe the Parkinson tremor; normalizing the electromyographic signal standard, and measuring the number of zero electromyographic data after normalization
Figure BDA0002340929680000048
I.e. the zero crossing rate. The normalization method comprises the following steps: mapping electromyographic data to [ -1,1]Interval, new data obtained are: x ═ x-xmean)/(xmax-xmin) Wherein x ismeanIs the mean value of the data, xmaxIs a maximum value, xminIs the minimum value.
Further, in step 6, according to the UPDRS parkinson rating scale, an hidden state set of the HMM model is defined, where the tremor amplitude is set to h, and the hidden state set includes:
state 0: no tremble;
state 1: slight tremor, namely the tremor occurs and 0< h is less than or equal to 1 cm;
state 2: moderate tremor, namely the tremor occurs, and 1< h is less than or equal to 3 cm;
and a state 3: severe tremor, namely the tremor occurs, and the h is more than 3 and less than or equal to 10 cm;
and 4: severe tremor, i.e. tremor occurs and h >10 cm;
according to the features extracted from the training set, state labels 0, 1, 2, 3 and 4 are respectively set for each data segment.
Further, in step 7, the HMM model parameters include an HMM hidden state number M, an initial probability pi, and a state observation probability B; initializing an HMM hidden state number M to be 5, initializing an initial probability pi to be [1,0,0,0,0], and expressing a state observation probability B by a set of Gaussian mixture densities; let parameter λ ═ (a, B, pi), a denote the implicit state transition probability;
and taking the characteristic parameters extracted from the training set as a training observation sequence, performing model training by using a Baum-Welch algorithm, and iteratively calculating the optimal parameter lambda of the HMM model.
Further, in step 8, the characteristic parameter Λ 'extracted in the test set is used as a test observation sequence, and is input into the trained HMM model, the Viterbi algorithm is used to solve the hidden state sequence when P (Λ' | λ) is maximum, and the current optimal hidden state (i.e. 0, 1, 2, 3, 4) is output; the characteristic parameter Lambda' is an angle characteristic and myoelectricity characteristic set; p (Λ '| λ) represents the probability that the Viterbi algorithm will solve for the occurrence of the observation sequence Λ' given the HMM model parameters λ.
Has the advantages that: compared with the prior art, the technical scheme of the invention has the following beneficial technical effects:
the method can provide technical support for the evaluation of the tremor degree of the arms of the Parkinson patients, and provides theoretical basis for the population such as the Parkinson patients, the old, the weak and the like who need to know the occurrence of the early Parkinson disease in time. Because the method adopts the characteristic of the relative angle variation of the wrist joint and the elbow joint of the human body, the defect that motion data in the conventional accelerometer or gyroscope scheme is influenced by the posture and the behavior of the human body can be overcome, the tremor state can be better represented under the daily behavior, and the method has better robustness and reliability. In addition, the real-time electromyogram signal characteristics of the EMG sensor deployed at the biceps brachii are adopted, so that the tremor degree under early Parkinson's disease can be extracted, and the defect that the early tremor is difficult to obtain by an IMU sensor is avoided. Experiments show that: on the premise that the tremor grade is set to be 5 according to the rating scale, the recognition rate of the HMM model can reach more than 98%, and the HMM model has good popularization capability and engineering application value.
Drawings
FIG. 1 is a kinetic model of the human upper limb kinetic chain;
FIG. 2 is a flow chart of a tremor assessment method based on angular variation characteristics and electromyographic characteristics;
FIG. 3 is a schematic diagram of tremor level hidden state transitions based on an HMM model;
FIG. 4 is a graph showing the results of an experiment of the method of the present invention;
description of reference numerals: 1-upper arm IMU sensor deployment position, 2-EMG sensor measurement electrode deployment position, 3-lower arm IMU sensor deployment position, 4-hand IMU sensor deployment position, 5-elbow joint flexion-extension angle, 6-lower arm internal rotation-external rotation angle, 7-hand flexion-extension angle.
Detailed Description
The technical solution of the present invention is further described below with reference to the accompanying drawings and examples. In order to better describe the tremor degree of the parkinson patients in daily activities, in the specific implementation process, the invention defines the following 8 situations:
s1: lying on a bed;
s2: sitting on the chair statically, the arms are relaxed naturally;
s3: sitting still on the chair, and laying the arms on the table top;
s4: standing still and naturally relaxing arms;
s5: after walking for 10 seconds at a slow speed, sitting on the chair statically, and naturally relaxing the arms;
s6: after the teacup is grabbed and drunk, the arm is flatly placed on the desktop;
s7: after the teacup is grabbed and the teacup is drunk, the elbow joint is supported on the tabletop;
s8: standing, and naturally relaxing arms after the upper arms are bent and stretched for 5 times;
and respectively carrying out the step operation of the method of the invention on the 8 situations, and establishing tremor evaluation models under different situations for testing the reliability and effectiveness of the method. Taking the scenario of S3 as an example, the implementation process is as follows:
step 1, three nine-axis IMU sensors with the model GY953 are respectively deployed on the upper arm, the lower wall and the hand of a human body, an EMG sensor with the model OYmotion is deployed on the biceps brachii of the arm, and the deployment positions are shown in fig. 1; initializing the sensor, eliminating the sensor zero drift, carrying out initial calibration, and setting the IMU sensor sampling frequency to be 100Hz and the EMG sensor sampling frequency to be 50 Hz.
Acquiring human body posture signals through three deployed IMU sensors, and acquiring human body surface electromyographic signals through an EMG sensor; and the attitude signals acquired by the IMU sensor are output in a quaternion form.
Step 2, filtering the attitude signal through band-pass filters with upper and lower cut-off frequencies of 3Hz and 6Hz, and filtering the electromyographic signal through band-pass filters with upper and lower cut-off frequencies of 3Hz and 12Hz to obtain a filtered attitude signal and an electromyographic signal; setting a current coordinate system as an earth coordinate system x, y and z, and calculating the variation of the attitude angle according to the attitude signal, wherein the specific method comprises the following steps:
defining an attitude angle theta1,θ2,θ3Change amount of attitude angle Δ θ1,△θ2,△θ3Wherein θ1,θ2,θ3Respectively as elbow joint bending-stretching angle, lower arm inward turning-outward turning angle, wrist joint bending-stretching angle, delta theta1,△θ2,△θ3The elbow joint bending-stretching angle variation, the lower arm inward rotation-outward rotation angle variation and the wrist joint bending-stretching angle variation are respectively;
respectively calculating the projection alpha of the attitude angles of the IMU sensors of the upper arm, the lower arm and the hand in the coordinate system x, y and z according to the obtained filtered attitude signals, namely the attitude quaternions of the IMU sensors of the upper arm, the lower arm and the handx,αy,αz、βx,βy,βz、γx,γy,γz；
From the arm geometry:
△θ1x＝△αx+△βx△θ3x＝△βx+△γx
△θ1y＝△αy+△βy，△θ3y＝△βy+△γy
△θ1z＝△αz+△βz△θ3z＝△βz+△γz
wherein Δ θ1x,△θ1y,△θ1z、△θ3x,△θ3y,△θ3zRespectively as elbow joint bending-stretching angle variation quantity delta theta1Variation quantity Delta theta of wrist joint bending-stretching angle3Projection on the x, y, z axes of the coordinate system; delta alphax,△αy,△αzThe angle variation of the projection of the upper arm attitude angle on the x, y and z axes; delta betax,△βy,△βzThe angle variation of the projection of the lower arm attitude angle on the x, y and z axes; delta gammax,△γy,△γzThe angle variation of the projection of the hand posture angle on the x, y and z axes;
separately calculating the variation quantity Delta theta of the attitude angle1,△θ3The expression is as follows:
Figure BDA0002340929680000071
Figure BDA0002340929680000072
since the lower arm rotates inward or outward along the axis of the elbow joint and the wrist joint, the posture angle change amount Δ θ2Derived directly from the quaternion output by the lower arm IMU sensor. According to the conversion relation between the quaternion and the attitude angle, the output quaternion can be converted into the angle variation.
Step 3, respectively storing the posture angle variation discrete data and the electromyographic signal discrete data output in the step 2 in a data set
Figure BDA0002340929680000078
The preparation method comprises the following steps of (1) performing; culling datasets separately on a time axis
Figure BDA0002340929680000079
And
Figure BDA00023409296800000710
before and after 10% of data, update data set
Figure BDA00023409296800000711
And
Figure BDA00023409296800000712
。
step 4, collecting the data set
Figure BDA00023409296800000713
And
Figure BDA00023409296800000714
respectively performing equal-step windowing, and continuously describing the change of tremor by using a data set
Figure BDA00023409296800000716
And
Figure BDA00023409296800000715
is set to 100ms, the window duration is 2s, a numberAccording to the overlapping rate of 95 percent; the processed data set
Figure BDA00023409296800000717
And
Figure BDA00023409296800000718
and combining, selecting 70% of the training sets as the training sets for HMM model parameter learning, and selecting the rest 30% of the training sets as the test sets for HMM model identification.
Step 5, extracting attitude angle change time domain and frequency domain characteristics of the Parkinson tremor in a resting state and electromyographic signal characteristics of the training set and the test centralized windowed data respectively; the method comprises the following steps:
extracting the attitude angle change time domain and frequency domain characteristics of the Parkinson tremor in the resting state, specifically comprising the following steps:
the method is characterized in that: average angular change rate of Parkinson vibration in angular change time domain of rest state
Figure BDA0002340929680000073
Describing the trembling degree, and respectively calculating the attitude angle theta1,θ2,θ3Average angular rate of change of
Figure BDA0002340929680000074
The unit is degree/second; the calculation formula is as follows:
Figure BDA0002340929680000075
where n denotes the data length, n ═ t × fsT is window duration, fsSampling frequency for IMU sensor;
and (2) characteristic: the energy of the Parkinson tremor in the angle change frequency domain of the resting state is used as a characteristic to describe the tremor degree;
and (3) performing discrete Fourier transform on the attitude signal data in the training set, converting the attitude signal data into frequency domain data, and calculating an energy value:
Figure BDA0002340929680000076
wherein E is an energy value, f represents a frequency point, and P (f) is a signal energy spectral density with the unit of dB/second; a and b are respectively the boundary values of the frequency range of the resting state tremor; [ a, b ] takes the value of the typical frequency range of resting state tremor of 3-6 Hz;
and (3) feature: information entropy of angle change of the Parkinson tremor in a resting state is used as a characteristic to describe the tremor degree; the information entropy T is defined as follows:
Figure BDA0002340929680000077
wherein
Figure BDA0002340929680000081
And taking 2 as the base of the log as the occurrence probability corresponding to the frequency point f.
Extracting electromyographic signal characteristics to describe Parkinson tremor, which comprises the following steps:
and (4) feature: selecting an electromyographic signal kurtosis coefficient as a characteristic to describe the Parkinson tremor; the kurtosis coefficient k is defined as follows:
Figure BDA0002340929680000082
wherein mu is a data mean value, lambda is a mathematical expectation, sigma is a standard deviation, and x is electromyographic signal discrete data;
and (5) characteristic: selecting the electromyographic signal zero crossing rate as a characteristic to describe Parkinson tremor; normalizing the electromyographic signal standard, and measuring the number of zero electromyographic data after normalization
Figure BDA0002340929680000083
I.e. the zero-crossing rate. The normalization method comprises the following steps: mapping electromyographic data to [ -1,1]Interval, new data obtained are: x ═ xmean)/(xmax-xmin) Wherein x ismeanIs the mean value of the data, xmaxIs a maximum value, xminIs the minimum value.
Step 6, defining an HMM model hidden state set according to a UPDRS Parkinson rating scale; let the tremor amplitude be h, the set of implicit states includes:
state 0: no tremble;
state 1: slight tremor, namely the tremor occurs and 0< h is less than or equal to 1 cm;
state 2: moderate tremor, namely the tremor occurs, and 1< h is less than or equal to 3 cm;
state 3: severe tremor, namely the tremor occurs, and the h is more than 3 and less than or equal to 10 cm;
and 4: severe tremor, i.e. tremor occurs and h >10 cm;
wherein h represents tremor amplitude;
according to the features extracted from the training set, state labels 0, 1, 2, 3 and 4 are respectively set for each data segment.
Step 7, initializing HMM model parameters, where the HMM hidden state number M is 5, the initial probability pi is [1,0,0,0,0], and the state observation probability B is represented by a set of gaussian mixture densities; let parameter λ ═ (a, B, pi), a denote the implicit state transition probability; taking the characteristic parameters extracted from the training set as a training observation sequence, performing model training by using a Baum-Welch algorithm, and iteratively calculating the optimal parameter lambda of the HMM model to obtain a trained HMM model; the characteristics comprise angle change characteristics and electromyographic signal characteristics.
Step 8, inputting the characteristic parameter Λ 'extracted in the test set as a test observation sequence into a trained HMM model, solving a hidden state sequence when P (Λ' | lambda) is maximum by using a Viterbi algorithm, and outputting current optimal hidden states (namely 0, 1, 2, 3 and 4); the characteristic parameter Lambda' is an angle characteristic and myoelectricity characteristic set; p (Λ '| λ) represents the probability that the Viterbi algorithm will solve for the occurrence of the observation sequence Λ' given the HMM model parameters λ.
According to the research, most resting tremor is reflected in the upper limb of the human body. Therefore, the present invention primarily selects the upper limb for analysis. The human arm can be modeled as a kinematic chain, as shown in fig. 1, consisting of three arm segments (upper arm, lower arm and hand) and three joints (shoulder joint, elbow joint and wrist joint), three IMU sensorsDeployed on the proximal elbow of the upper arm (above the distal humerus, position 1 in the figure), proximal wrist of the lower arm (above the distal radius and ulna, position 3 in the figure) and dorsal surface of the hand (position 4 in the figure), respectively. The soft tissue of the arm has little influence on the positions, and the signal is not easily interfered by the micro-motion of the soft tissue. Meanwhile, EMG sensor measurement electrodes are deployed on the surface of the upper arm biceps brachii muscle (position 2 in the figure), since elbow flexion-extension and medial-lateral rotation have a direct relationship with the biceps brachii muscle, and therefore surface electromyographic signals at this position need to be measured for later EMG feature extraction. Because resting state tremor is usually represented by combined motion of arm wrist joint and elbow joint rotation, the invention selects three angles for describing rotation characteristics as input quantities for later-stage characteristic extraction, wherein the three angles are respectively theta1(elbow flexion-extension angle, angle 5 in the figure), theta2(lower arm inner rotation-outer rotation angle, angle 6 in the figure) and theta3(hand flexion-extension angle, angle 7 in the figure).
Fig. 2 shows a general flow chart of the tremor evaluation method of the present invention, which is mainly divided into three parts, namely, data acquisition and preprocessing, feature extraction, and HMM model training and recognition. The calculation process of the model comprises an off-line stage and an on-line stage, wherein the off-line stage is used for tremor data segmentation and tremor characteristic data set construction; the online phase is used for tremor feature identification and state assessment. Compared with the conventional threshold evaluation method, the HMM method has the following advantages: the Markov chain structure can keep the structural information of the tremor characteristic time sequence; the HMM model parameters may represent the statistical properties of tremor; HMM, as a probabilistic model, no longer needs to set a threshold. Therefore, the conversion process of the tremor state can be directly reflected on the time sequence by adopting the HMM model, and the time sequence characteristics of the tremor of the Parkinson patients in daily life can be better described.
Hidden Markov Models (HMMs) can better characterize time-dependent signals and can be well characterized as stochastic processes of parameters relative to other types of time-series signal analysis models. It retains the structural information of the signature signal without the need for the thresholds used in the heuristic rules. In order to better describe tremor state conversion and improve generalization capability of the model, the angle characteristic information and myoelectric characteristic information are selected as model input, the characteristic matrix calculation amount is small, and the method is suitable for occasions with low calculation capability and high energy efficiency requirement of the wearable somatosensory network.
According to the actual process of the tremor state transition, the invention only selects the continuous gradual state transition process, as shown in fig. 3, namely 0-0, 0-1,1-1, 1-2,2-2, 2-1, 2-3,3-3, 3-2, 3-4,4-4, 4-3, and abandons the tremor state jump process 0-2,2-0,0-4,1-3, 4-2 and the like; furthermore, it is defined that the initial state and the end state of the data set are both 0 states, i.e. starting first from the non-tremor state until the end of the non-tremor state.
Fig. 4 shows the wrist joint angle variation, the average angle variation rate, the angle frequency domain energy value, the angle frequency domain information entropy, and the hidden state sequence of HMM model recognition from top to bottom. According to the actual measurement result, the optimal sequence output by the HMM model can be obtained as follows: 0-1-2-1-0. It can be seen that the selected angle features have better descriptive and distinguishing degrees, and have better identification effect on the Parkinson tremor evaluation. The method can realize the state recognition rate of more than 98 percent.

Claims (10)

1. A Parkinson resting state tremor evaluation method based on a wearable somatosensory network is characterized by comprising the following steps: the method comprises the following steps:
step 1, deploying sensors at hands and arms of a human body, initializing the sensors, eliminating zero drift of the sensors, carrying out initial calibration, and setting sampling frequency of the sensors; acquiring human body posture signals and human body surface electromyographic signals through deployed sensors;
step 2, filtering the attitude signal and the electromyographic signal through a band-pass filter respectively according to the resting state tremor frequency response characteristic to obtain the filtered attitude signal and electromyographic signal, and calculating the variation of the attitude angle according to the attitude signal;
step 3, respectively storing the posture angle variation discrete data and the electromyographic signal discrete data output in the step 2 in a data set N, C; respectively eliminating p% of data before and after the data sets N and C according to the time axis, and updating the data sets N and C;
step 4, respectively carrying out equal-step-length windowing on the data sets N and C, combining the processed data sets N and C, selecting q% of the data sets N and C as a training set for HMM model parameter learning, and using the rest 1-q% of the data sets N and C as a test set for HMM model identification;
step 5, extracting attitude angle change time domain and frequency domain characteristics and electromyographic signal characteristics of the Parkinson's tremor in a resting state from the training set and the test centralized windowed data respectively, wherein the attitude angle change time domain and frequency domain characteristics and the electromyographic signal characteristics comprise average angle change rate in an angle change time domain, energy in an angle change frequency domain, information entropy of angle change, electromyographic signal kurtosis coefficient and electromyographic signal zero crossing rate;
step 6, defining an HMM model hidden state set according to a UPDRS Parkinson rating scale; respectively setting state labels for the data segments according to the features extracted from the training set;
step 7, initializing HMM model parameters, taking the characteristic parameters extracted from the training set as a training observation sequence, performing model training, and calculating the optimal parameters of the HMM model to obtain a trained HMM model;
and 8, inputting the characteristic parameters extracted in the test set into the trained HMM model by taking the characteristic parameters as test observation sequences, solving the hidden state sequence when the probability of the observation sequences is maximum, and outputting the current optimal hidden state to obtain the tremor state evaluation result.
2. The method for assessing Parkinson resting tremor based on wearable somatosensory mesh according to claim 1, wherein the method comprises the following steps: step 1, deploying IMU sensors on an upper arm, a lower arm and a hand respectively, and deploying EMG sensors on the upper arm at the same time; acquiring human body posture data through an IMU sensor, and acquiring human body surface muscle electric signals through an EMG sensor; the IMU sensor comprises a three-axis accelerometer, a three-axis gyroscope and a three-axis magnetometer, and attitude signals acquired by the IMU sensor are output in a quaternion form.
3. The method for assessing Parkinson resting-state tremor based on wearable somatosensory network according to claim 2, wherein the method comprises the following steps: the IMU sensor sampling frequency was set to 100Hz and the EMG sensor sampling frequency was set to 50 Hz.
4. The method for assessing Parkinson resting-state tremor based on wearable somatosensory network according to claim 2, wherein the method comprises the following steps: a nine-axis IMU sensor with the model of GY953 is selected, and an EMG sensor with the model of OYmotion is selected.
5. The method for assessing Parkinson resting tremor based on wearable somatosensory mesh according to claim 2, wherein the method comprises the following steps: in the step 2, the current coordinate system is set as an earth coordinate system x, y, z, and the attitude angle variation is calculated, wherein the method comprises the following steps:
defining an attitude angle theta1,θ2,θ3Change amount of attitude angle Δ θ1,Δθ2,Δθ3Wherein θ1,θ2,θ3Respectively an elbow joint bending-stretching angle, a lower arm inward rotation-outward rotation angle, a wrist joint bending-stretching angle delta theta1,Δθ2,Δθ3The elbow joint bending-stretching angle variation, the lower arm inner rotation-outer rotation angle variation and the wrist joint bending-stretching angle variation are respectively;
respectively calculating the projection alpha of the attitude angles of the IMU sensors of the upper arm, the lower arm and the hand on the x, y and z axes of a coordinate system according to the attitude quaternion of the IMU sensors of the upper arm, the lower arm and the handx,αy,αz、βx,βy,βz、γx,γy,γz；
From the arm geometry we obtain:
Δθ1x＝Δαx+ΔβxΔθ3x＝Δβx+Δγx
Δθ1y＝Δαy+Δβy，Δθ3y＝Δβy+Δγy
Δθ1z＝Δαz+ΔβzΔθ3z＝Δβz+Δγz
where Δ θ1x,Δθ1y,Δθ1z、Δθ3x,Δθ3y,Δθ3zRespectively as elbow joint bending-stretching angle variation delta theta1Wrist joint flexion-extension angle variation delta theta3Projection on the x, y, z axes of the coordinate system; delta alphax,Δαy,ΔαzThe angle variation of the projection of the upper arm attitude angle on the x, y and z axes; delta betax,Δβy,ΔβzThe angle variation of the projection of the attitude angle of the lower arm on the x, y and z axes; delta gammax,Δγy,ΔγzThe angular variation of the projection of the hand attitude angle on the x, y and z axes;
respectively calculating the variation delta theta of the attitude angle1,Δθ3The expression is as follows:
Figure FDA0003681809370000021
Figure FDA0003681809370000022
attitude angle variation Δ θ2Derived directly from the quaternion output by the lower arm IMU sensor.
6. The method for assessing Parkinson resting-state tremor based on wearable somatosensory network according to claim 5, wherein the method comprises the following steps: the step 5 of extracting the attitude angle change time domain and frequency domain characteristics of the Parkinson's tremor in the resting state specifically comprises the following steps:
the method is characterized in that: average angular change rate of Parkinson vibration in angle change time domain of rest state
Figure FDA0003681809370000023
Describing the trembling degree, and respectively calculating the attitude angle theta1,θ2,θ3Average angular rate of change of
Figure FDA0003681809370000024
Calculating outThe formula is as follows:
Figure FDA0003681809370000025
where n denotes the data length, n ═ t × fsT is window duration, fsSampling frequency for IMU sensor;
and (2) characteristic: the energy of the Parkinson tremor in the angle change frequency domain of the resting state is used as a characteristic to describe the tremor degree;
performing discrete Fourier transform on attitude signal data in a training set, converting the attitude signal data into frequency domain data, and calculating an energy value:
Figure FDA0003681809370000026
wherein E is an energy value, f represents a frequency point, and P (f) is a signal energy spectral density; a and b are respectively the boundary values of the frequency range of resting tremor;
and (3) feature: information entropy of angle change of the Parkinson tremor in a resting state is used as a characteristic to describe the tremor degree; the information entropy T is defined as follows:
Figure FDA0003681809370000031
wherein
Figure FDA0003681809370000032
And taking 2 as the base of the log as the occurrence probability corresponding to the frequency point f.
7. The method for assessing Parkinson resting tremor based on wearable somatosensory mesh according to claim 1, wherein the method comprises the following steps: in the step 5, extracting the electromyographic signal characteristics to describe the Parkinson tremor, which comprises the following steps:
and (4) feature: selecting the crest factor of the electromyographic signal as a characteristic to describe the Parkinson's tremor; the kurtosis coefficient k is defined as follows:
Figure FDA0003681809370000033
wherein mu is a data mean value, Λ is a mathematical expectation, σ is a standard deviation, and x is electromyographic signal discrete data;
and (5) feature: selecting the electromyographic signal zero crossing rate as a characteristic to describe Parkinson tremor; normalizing the electromyographic signal standard, and measuring the number of zero electromyographic data after normalization
Figure FDA0003681809370000034
Namely the zero-crossing rate; the normalization method comprises the following steps: mapping electromyographic data to [ -1,1]And (3) obtaining new data according to the calculation formula: x ═ xmean)/(xmax-xmin) Wherein x ismeanIs the mean value of the data, xmaxIs a maximum value, xminIs the minimum value.
8. The method for assessing Parkinson resting tremor based on wearable somatosensory mesh according to claim 1, wherein the method comprises the following steps: step 6, defining an hidden state set of the HMM model according to the UPDRS Parkinson rating scale, setting the tremor amplitude as h, wherein the hidden state set comprises:
state 0: no tremble;
state 1: slight tremor, namely tremor occurs, and h is more than 0 and less than or equal to 1 cm;
state 2: moderate tremor, namely tremor occurs, and 1< h is less than or equal to 3 cm;
and a state 3: severe tremor, namely the tremor occurs, and the h is more than 3 and less than or equal to 10 cm;
and 4, state 4: severe tremor, i.e. tremor occurs and h >10 cm;
according to the features extracted from the training set, state labels 0, 1, 2, 3 and 4 are respectively set for each data segment.
9. The method for assessing Parkinson resting tremor based on wearable somatosensory mesh according to claim 1, wherein the method comprises the following steps: the step 7, the HMM model parameters include HMM hidden state number M, initial probability pi, and state observation probability B; initializing an HMM hidden state number M to be 5, initializing a probability pi to be [1,0,0,0,0], and representing a state observation probability B by a group of Gaussian mixture densities; let parameter λ ═ (a, B, pi), a denote the implicit state transition probability;
and taking the characteristic parameters extracted from the training set as a training observation sequence, performing model training by using a Baum-Welch algorithm, and iteratively calculating the optimal parameter lambda of the HMM model.
10. The method for assessing Parkinson resting tremor based on wearable somatosensory mesh according to any of claims 1-9, wherein: step 8, inputting the characteristic parameter Λ 'extracted from the test set as a test observation sequence into a trained HMM model, solving the hidden state sequence when P (Λ' | λ) is maximum using a Viterbi algorithm, and outputting the current optimal hidden state; the characteristic parameter Lambda' is an angle characteristic and myoelectricity characteristic set; p (Λ '| λ) represents the probability of the occurrence of the observation sequence Λ' when the Viterbi algorithm solves for a given HMM model parameter λ.
CN201911375827.XA 2019-12-27 2019-12-27 Parkinson resting state tremor evaluation method based on wearable somatosensory network Active CN110946556B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911375827.XA CN110946556B (en) 2019-12-27 2019-12-27 Parkinson resting state tremor evaluation method based on wearable somatosensory network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911375827.XA CN110946556B (en) 2019-12-27 2019-12-27 Parkinson resting state tremor evaluation method based on wearable somatosensory network

Publications (2)

Publication Number Publication Date
CN110946556A CN110946556A (en) 2020-04-03
CN110946556B true CN110946556B (en) 2022-07-15

Family

ID=69984542

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911375827.XA Active CN110946556B (en) 2019-12-27 2019-12-27 Parkinson resting state tremor evaluation method based on wearable somatosensory network

Country Status (1)

Country Link
CN (1) CN110946556B (en)

Families Citing this family (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111528842B (en) * 2020-05-26 2023-01-03 复嶂环洲生物科技(上海)有限公司 Quantitative assessment method for Parkinson disease symptoms based on physiological and behavioral indexes
CN111513724A (en) * 2020-05-28 2020-08-11 长春理工大学 Wearable device suitable for rehabilitation monitoring and evaluation of hand function of Parkinson patient
CN111643092A (en) * 2020-06-02 2020-09-11 四川大学华西医院 Epilepsia alarm device and epilepsia detection method
CN112766041B (en) * 2020-12-25 2022-04-22 北京理工大学 Method for identifying hand washing action of senile dementia patient based on inertial sensing signal
CN112826504B (en) * 2021-01-07 2024-03-26 中新国际联合研究院 Game parkinsonism grade assessment method and device
CN113100756A (en) * 2021-04-15 2021-07-13 重庆邮电大学 Stacking-based Parkinson tremor detection method
CN113609975A (en) * 2021-08-04 2021-11-05 苏州小蓝医疗科技有限公司 Modeling method for tremor detection, hand tremor detection device and method
CN115336979B (en) * 2021-09-03 2024-09-06 中国人民解放军总医院 Automatic detection method and detection device for multitasking tremor based on wearable device
CN114224296B (en) * 2022-01-13 2023-07-21 福州大学 Parkinson's motion symptom quantitative evaluation system based on wearable sensing device
CN114788687B (en) * 2022-06-23 2022-09-27 中国科学院自动化研究所 Quantitative assessment method and device for Parkinson myotonia symptoms
CN117298449B (en) * 2023-10-31 2024-04-09 首都医科大学宣武医院 Closed-loop DBS regulation and control method and system based on wearable equipment

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104398263A (en) * 2014-12-25 2015-03-11 中国科学院合肥物质科学研究院 Method for quantitatively evaluating symptoms of tremor of patient with Parkinson's disease according to approximate entropy and cross approximate entropy
CN105930663A (en) * 2016-04-26 2016-09-07 北京科技大学 Parkinson's disease early diagnosis method
CN109452942A (en) * 2017-09-06 2019-03-12 扬州工业职业技术学院 A kind of Parkinson's hand trembles monitoring system

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6734834B1 (en) * 2000-02-11 2004-05-11 Yoram Baram Closed-loop augmented reality apparatus
CN104127187B (en) * 2014-08-05 2017-04-05 戴厚德 For the wearable system of patient's Parkinson cardinal symptom quantitative determination
US20190365286A1 (en) * 2018-06-01 2019-12-05 Apple Inc. Passive tracking of dyskinesia/tremor symptoms
CN109344195B (en) * 2018-10-25 2021-09-21 电子科技大学 HMM model-based pipeline security event recognition and knowledge mining method

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104398263A (en) * 2014-12-25 2015-03-11 中国科学院合肥物质科学研究院 Method for quantitatively evaluating symptoms of tremor of patient with Parkinson's disease according to approximate entropy and cross approximate entropy
CN105930663A (en) * 2016-04-26 2016-09-07 北京科技大学 Parkinson's disease early diagnosis method
CN109452942A (en) * 2017-09-06 2019-03-12 扬州工业职业技术学院 A kind of Parkinson's hand trembles monitoring system

Also Published As

Publication number Publication date
CN110946556A (en) 2020-04-03

Similar Documents

Publication Publication Date Title
CN110946556B (en) Parkinson resting state tremor evaluation method based on wearable somatosensory network
CN109243572B (en) Accurate motion evaluation and rehabilitation training system
Palermo et al. Repeatability of grasp recognition for robotic hand prosthesis control based on sEMG data
Hargrove et al. The effect of electrode displacements on pattern recognition based myoelectric control
Fougner et al. Resolving the limb position effect in myoelectric pattern recognition
Yoon et al. Improvement of dynamic respiration monitoring through sensor fusion of accelerometer and gyro-sensor
CN108960155B (en) Adult gait extraction and anomaly analysis method based on Kinect
Jose et al. Classification of forearm movements from sEMG time domain features using machine learning algorithms
Mangukiya et al. Electromyography (EMG) sensor controlled assistive orthotic robotic arm for forearm movement
Ismail et al. Hand motion pattern recognition analysis of forearm muscle using MMG signals
CN113609975A (en) Modeling method for tremor detection, hand tremor detection device and method
Shouldice et al. Real time breathing rate estimation from a non contact biosensor
Dorofeev et al. The assessment of gait features according to the data of a portable acceleration sensor in an intelligent monitoring system
CN113520413B (en) Lower limb multi-joint angle estimation method based on surface electromyogram signals
Nougarou et al. Muscle activity distribution features extracted from HD sEMG to perform forearm pattern recognition
Kang et al. A Precise Muscle activity onset/offset detection via EMG signal
Rigas et al. Real-time quantification of resting tremor in the Parkinson's disease
Al-Quraishi et al. Impact of feature extraction techniques on classification accuracy for EMG based ankle joint movements
WO2008152549A2 (en) Device for functional assessment of a shoulder
KR101071214B1 (en) Apparatus for measurement of angular velocity in disease patients and analysis system for the same
Mamikoglu et al. Elbow joint angle estimation by using integrated surface electromyography
CN116269413A (en) Continuous electrocardiographic waveform reconstruction system and method using smart wristband motion sensor
Guo et al. Method of gait disorders in Parkinson's disease classification based on machine learning algorithms
CN115024716A (en) Ballistocardiogram signal reconstruction method based on heart rate label generation
Lee et al. Towards the ambulatory assessment of movement quality in stroke survivors using a wrist-worn inertial sensor

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant