Method and device for identifying dyskinesia of Parkinson's patients based on wearable devices

Abstract

The present invention relates to the technical field of movement disorder recognition, and discloses a movement disorder recognition method and device for Parkinson's disease patients based on wearable devices. The method includes the following steps: S1: Collect historical data and construct a reference database for movement disorders; S2: Collect the movement data of the user and obtain the basic movement parameters of the user; S3: Based on the basic movement parameters of the user, screen reference samples for the user from the reference database; S4: Extract features from the movement data of the user to obtain the movement feature index of the user; S5: Match the movement feature indexes of the user and the reference samples to obtain the movement disorder recognition result of the user. The present invention improves the accuracy and comprehensiveness of movement disorder recognition for Parkinson's patients.

Description

Technical Field

[0001] The present invention relates to the technical field of movement disorder recognition, and particularly to a method and device for recognizing movement disorders of Parkinson's patients based on wearable devices. Background Art

[0002] In recent years, wearable devices, such as smart watches, wristbands, and insoles, can monitor various physiological parameters of the human body in real time and transmit the data to the cloud for processing and analysis through wireless communication technology. For Parkinson's patients, wearable devices can not only be used to monitor the daily activities of patients, but also assist doctors in disease assessment and treatment efficacy monitoring, thereby improving the efficiency and accuracy of diagnosis and treatment.

[0003] Currently, most of the Parkinson's disease monitoring systems based on wearable devices on the market still have some deficiencies: existing Parkinson's disease monitoring systems often can only collect a single type of biological signal, making it difficult for the system to comprehensively and accurately identify various types of movement disorders. At the same time, the lack of effective feature extraction algorithms makes it difficult for the system to capture key features when facing complex movement patterns, affecting the reliability and accuracy of the evaluation results. There are significant individual differences among Parkinson's patients, including multiple aspects such as age, gender, weight, and muscle strength. However, current monitoring systems often ignore the influence of these individual factors in design, resulting in a one-size-fits-all evaluation standard and being unable to accurately locate according to the specific situation of each patient, reducing the practical application value of the evaluation results.

[0004] For example, the Chinese patent application with the authorization announcement number CN111544006B discloses a wearable device for quantifying and recognizing movement disorders of Parkinson's patients, aiming to solve the problem of lacking a reliable and non-invasive wearable device to accurately achieve the quantification and recognition of movement disorders of Parkinson's patients. The invention includes: collecting wrist movement signals and ankle movement signals through wrist inertial measurement units and ankle inertial measurement units worn on different sides of the body of the measured object and a healthy person, and extracting the features of the signals; sending the extracted features to an output processing terminal, and obtaining the disease degree of the measured object and outputting it through a preset method for quantifying and recognizing movement disorders of Parkinson's patients. The invention has a simple structure, is easy to operate, has a suitable size for convenient wearing, low power consumption, and combines the movement signals of the wrist and ankle, and the judgment result is accurate and reliable.

[0005] The patent application with the publication number CN115346670A discloses a Parkinson's disease rating method, an electronic device and a medium based on pose recognition. The method first obtains the movement video of a Parkinson's disease patient to be detected, processes each frame of the image in the movement video, and obtains the area where the human body bone key points are located in each frame of the image; and intercepts the photo corresponding to the area where the human body bone key points are located, and obtains the coordinates of the human body bone key points after confidence threshold test. The time series formed by the coordinates of the human body bone key points in each frame of the image; according to the Movement Disorder Society of China Unified Parkinson's Disease Rating Scale MDS-UPDRS, the time series formed by the coordinates of the human body bone key points in each frame of the image is converted into the action quality index corresponding to the movement video; collects the patient's case information and converts it into a symptom index; splices the action quality index and the symptom index into an original feature vector, and uses a support vector machine model to perform regression on the original feature vector to obtain the Parkinson's disease severity index.

[0006] All of the above technical solutions have the problems raised in this background technology: ignoring the influence of these individual factors and being unable to accurately identify according to the specific situation of each patient.

[0007] The information disclosed in this background technology section is only intended to increase the understanding of the overall background of the present invention, and should not be regarded as an admission or any form of implication that this information constitutes the prior art already known to those of ordinary skill in the art. Summary of the Invention

[0008] The technical problem to be solved by the present invention is to overcome the defects of the prior art, and provide a method and device for identifying movement disorders of Parkinson's disease patients based on wearable devices, so as to improve the accuracy and comprehensiveness of the identification of movement disorders of Parkinson's patients.

[0009] To solve the above technical problems, the present invention provides the following technical solutions:

[0010] On the one hand, the present invention provides a method for identifying movement disorders of Parkinson's disease patients based on wearable devices, including the following steps:

[0011] S1: Collect historical data and construct a reference database for movement disorders;

[0012] S2: Collect the movement data of the user and obtain the basic movement parameters of the user;

[0013] S3: Based on the matching index and the basic movement parameters of the user, screen reference samples for the user from the reference database;

[0014] S4: Extract features from the user's motion data to obtain the user's motion feature indicators, and determine the tremor synchronization feature and motion coordination ability of the user based on the motion feature indicators; the motion feature indicators include tremor feature indicators, activity feature indicators, and combined feature indicators;

[0015] S5: Match the motion feature indicators of the user with those of the reference samples, and obtain the motion disorder recognition result of the user by combining the tremor synchronization feature and the motion coordination ability.

[0016] As a preferred solution of the motion disorder recognition method for Parkinson's patients based on a wearable device according to the present invention, wherein: any piece of historical data in the reference database includes a motion disorder type, a first feature vector, and a second feature vector; wherein, the motion disorder type is a combination of one motion disorder or m motion disorders, and m is a positive integer greater than 1; encode the basic motion parameters to form the first feature vector; encode the motion feature indicators to form the second feature vector;

[0017] The motion data includes acceleration data at the wrist, acceleration data at the ankle, acceleration data at the chest, and angular velocity data at the head;

[0018] The user's basic motion parameters include age, gender, height, weight, joint range of motion, muscle strength, resting heart rate, and heart rate variability.

[0019] As a preferred solution of the motion disorder recognition method for Parkinson's patients based on a wearable device according to the present invention, wherein: the method for screening reference samples for the user is as follows:

[0020] Encode the user's basic motion parameters and form a basic parameter vector;

[0021] Calculate the matching index between the basic parameter vector and the first feature vector of each piece of historical data in the reference database;

[0022] Extract the M pieces of historical data corresponding to the largest M matching indexes as the M reference samples screened for the user.

[0023] As a preferred solution of the motion disorder recognition method for Parkinson's patients based on a wearable device according to the present invention, wherein: the calculation formula of the matching index is as follows: The formula is as follows:

[0024] ;

[0025] Wherein, S represents the matching index between the basic parameter vector and the first feature vector of any piece of historical data in the reference database; represents the i-th element in the basic parameter vector, where the value range of i is 1, 2, ……, n, and n is the dimension of the basic parameter vector and the first eigenvector of any piece of historical data; represents the i-th element in the first eigenvector; represents the adjustment coefficient; represents the penalty factor of

[0026] ;

[0027] where, represents the minimum threshold of the gap between and

[0028] As a preferred solution of the method for identifying movement disorders of Parkinson's patients based on wearable devices according to the present invention, wherein: the tremor characteristic indexes include wrist tremor frequency, wrist tremor amplitude, head rotation frequency, head rotation amplitude, and posture deviation index; the extraction methods of wrist tremor frequency and wrist tremor amplitude are as follows:

[0029] Extract the first tremor data segment from the acceleration data at the wrist when the user is stationary;

[0030] Perform DC component removal and filtering on the first tremor data segment, specifically as follows:

[0031] Calculate the mean value of the acceleration data in the first tremor data segment, and subtract the mean value from each acceleration data in the first tremor data segment;

[0032] Use a band-pass filter to perform filtering and noise reduction on the first tremor data segment;

[0033] Calculate the average peak value of the first tremor data segment as the wrist tremor amplitude;

[0034] Perform frequency domain transformation on the first tremor data segment and extract the wrist tremor frequency;

[0035] The extraction methods of the head rotation frequency and the head rotation amplitude are as follows: Extract the second tremor data segment from the angular velocity data of the head when the user is stationary; perform DC component removal and filtering on the second tremor data segment; calculate the average peak value of the second tremor data segment as the head rotation amplitude; perform frequency domain transformation on the second tremor data segment and extract the head rotation frequency.

[0036] The extraction method of the posture deviation index is as follows: Extract the chest acceleration data in the vertical direction when the user is stationary as the vertical acceleration; the posture deviation index is the absolute value of the difference between the vertical acceleration and the gravitational acceleration.

[0037] As a preferred solution of the method for identifying movement disorders of Parkinson's patients based on wearable devices according to the present invention, wherein: the activity characteristic indexes include gait cycle, step length, step width, and trunk stability index; among them, the gait cycle, step length, and step width are calculated based on the acceleration data at the ankle of the user during movement; the trunk stability index is the reciprocal of the variance of the chest acceleration data;

[0038] The combined characteristic indexes include starting delay and movement duration; the starting delay is the time required for the user to start moving from a stationary state when performing a specified action; the movement duration is the time required for the user to complete a specified action.

[0039] As a preferred solution of the method for identifying movement disorders of Parkinson's patients based on wearable devices according to the present invention, wherein: the tremor synchronization characteristic is represented by a tremor synchronization index; the method for determining the tremor synchronization characteristic of the user is as follows:

[0040] Integrate the acceleration data in the first tremor data segment to obtain a third tremor data segment;

[0041] Calculate the cross-correlation coefficients between the second tremor data segment and the third tremor data segment at different lag times;

[0042] Extract the maximum cross-correlation coefficient between the second tremor data segment and the third tremor data segment and the corresponding lag time, and assign a value to the tremor synchronization index; specifically as follows:

[0043] Let the maximum cross-correlation coefficient between the second tremor data segment and the third tremor data segment be and the corresponding lag time be ; if is greater than a preset first cross-correlation threshold, and is less than a preset first lag threshold, then the tremor synchronization index is assigned a value of 1; otherwise, the tremor synchronization index is assigned a value of 0.

[0044] As a preferred solution of the method for identifying movement disorders of Parkinson's patients based on wearable devices according to the present invention, wherein: the movement coordination ability is represented by a movement coordination index; the method for determining the movement coordination ability of the user is as follows:

[0045] Extract a first movement data segment from the acceleration data at the wrist of the user when walking; extract a second movement data segment from the acceleration data at the ankle of the user when walking;

[0046] Calculate the cross-correlation coefficients between the first movement data segment and the second movement data segment at different lag times;

[0047] Extract the maximum cross - correlation coefficient and the corresponding lag time between the first motion data segment and the second motion data segment, and assign a value to the motion coordination index; specifically as follows:

[0048] Let the maximum cross - correlation coefficient between the first motion data segment and the second motion data segment be , and the corresponding lag time be ; if is greater than the preset second cross - correlation threshold, and is less than the preset second lag threshold, then the motion coordination index is assigned a value of 1; otherwise, the motion coordination index is assigned a value of 0.

[0049] As a preferred solution of the method for identifying movement disorders of Parkinson's patients based on wearable devices according to the present invention, wherein: the method of matching the movement characteristic indexes of the user with the reference samples and combining the tremor synchronization characteristics and movement coordination ability to obtain the movement disorder identification result of the user is as follows:

[0050] S501: Encode the movement characteristic indexes, tremor synchronization index, and movement coordination index of the user and form a movement characteristic vector;

[0051] S502: Calculate the tremor synchronization index and movement coordination index of each reference sample; encode the tremor synchronization index and movement coordination index of each reference sample and add them to the corresponding second feature vector;

[0052] S503: Calculate the similarity between the movement characteristic vector and the second feature vector of each reference sample;

[0053] S504: Extract N reference samples corresponding to the N second feature vectors with the maximum similarity to the movement characteristic vector as voting samples;

[0054] S505: Count the movement disorder types of the voting samples, and the movement disorder type with the most occurrences among the N voting samples is the movement disorder identification result of the user.

[0055] In a second aspect, the present invention provides a device for identifying movement disorders of Parkinson's patients based on wearable devices, including a database module, a data acquisition module, a data processing module, a feature index module, and an identification module; wherein:

[0056] The database module is used to collect historical data and construct a reference database;

[0057] The data acquisition module is used to collect the movement data of the user and obtain the basic movement parameters of the user;

[0058] The data processing module is configured with a matching index calculation formula and a penalty factor value - taking strategy for screening reference samples for the user;

[0059] The feature index module extracts the user's motion feature indexes based on the user's motion data;

[0060] The recognition module is used to match the motion feature indexes of the user and the reference sample to obtain the motion disorder recognition result of the user.

[0061] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0062] By integrating multi-source data from the wrist, ankle, chest and head, common motion disorders such as tremors, muscle rigidity, and bradykinesia can be effectively identified; by accurately calculating feature indexes such as tremor frequency, amplitude, and gait cycle, the accuracy and reliability of the recognition are improved.

[0063] By collecting basic information such as the user's age, gender, height, and weight, and combining functional test data such as joint range of motion and muscle strength, a personalized evaluation system is established. On this basis, the present invention also proposes a method based on the matching of the basic parameter vector and the historical data feature vector to screen out reference samples with similar motion abilities to the target user, thereby improving the pertinence and effectiveness of the evaluation results. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings. Among them:

[0065] Figure 1 is the flowchart of the method for identifying motion disorders of Parkinson's patients based on wearable devices provided by the present invention;

[0066] Figure 2 is the flowchart of the method for extracting the wrist tremor frequency and wrist tremor amplitude provided by the present invention;

[0067] Figure 3 is the flowchart of the method for extracting the tremor synchronization index provided by the present invention;

[0068] Figure 4 is the flowchart of the method for obtaining the motion disorder recognition result of the user provided by the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0069] The technical solution of the present invention is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations on the technical solution of the present invention. The embodiments of the present invention and the technical features in the embodiments may be combined with each other unless there is a conflict.

[0070] Example 1

[0071] This embodiment introduces a method for identifying movement disorders in Parkinson's disease patients based on a wearable device. Figure 1 , the method comprises the following steps:

[0072] S1: Collect historical data and build a reference database of movement disorders;

[0073] Any historical data in the reference database includes a movement disorder type, a first feature vector, and a second feature vector; wherein the movement disorder type is a movement disorder or a combination of m movement disorders, and m is a positive integer greater than 1; the first feature vector is a feature vector composed of basic motion parameter encodings; and the second feature vector is a feature vector composed of motion feature index encodings.

[0074] The types of movement disorders include tremor, muscle rigidity, bradykinesia, gait disorder, and balance disorder; tremor that occurs when the limbs are in a static state is a static tremor; the muscle rigidity is manifested as continuous muscle tension, resulting in a limited range of motion; the bradykinesia includes delayed start and prolonged movement duration; the delayed start is manifested as difficulty in starting the movement, and the time required from static state to starting movement is too long; the prolonged movement duration is manifested as a long time required to complete the movement. The gait disorder includes gait freezing, gait instability, and gait panic; gait freezing is a sudden inability to move; gait instability is manifested as irregular or unstable gait cycle; gait panic is manifested as shorter step length, wider step width, etc. These characteristics may be to maintain the stability of the steps. The balance disorder includes posture control disorder, posture deviation, and abnormal posture reflex. Among them, posture control disorder is identified by analyzing the stability of the trunk. A large amplitude of trunk shaking indicates that there is an obstacle to posture control; posture deviation is manifested as a large offset of the center of gravity of the body when standing, which means a decrease in posture control ability. Abnormal posture reflex refers to the inability to adjust posture in time to maintain balance. Large time lags between movements of different limbs or body parts indicate poor coordination, i.e. abnormal postural reflexes.

[0075] S2: Collect the user's motion data and obtain the user's basic motion parameters;

[0076] The motion data includes acceleration data at the wrist, acceleration data at the ankle, acceleration data at the chest, and angular velocity data at the head.

[0077] Acceleration data at the wrist, acceleration data at the ankle, and acceleration data at the chest can be collected through an accelerometer; for patients with Parkinson's disease, the accelerometer can help record information such as the amplitude and frequency of tremors, changes in gait, and balance ability. Angular velocity data of the head can be collected through a gyroscope, which plays an important role in detecting changes in rotational movements and reduced postural stability, etc.

[0078] The basic motion parameters of the user include age, gender, height, weight, joint range of motion, muscle strength, resting heart rate, and heart rate variability.

[0079] There are significant differences in muscle strength, reaction time, and flexibility among people of different ages and genders; weight and height affect gait, strength, and flexibility. By collecting age, gender, height, and weight, it can be used to exclude samples with significant differences in motor ability from the user due to age differences, gender differences, and height and weight differences in the follow-up. In addition, some functional test data need to be collected, including joint range of motion, muscle strength, resting heart rate, and heart rate variability. Among them, the joint range of motion affects motor ability and postural control; muscle strength directly affects the ability of daily activities; the resting heart rate is the heart rate in a quiet state, which can reflect the basic metabolic ability and exercise level to a certain extent; the heart rate variability is represented by the low-frequency / high-frequency ratio of the heart rate and is used to exclude differences in motor ability caused by differences in the function of the autonomic nervous system. Through the comprehensive evaluation of the above basic motion parameters, samples with similar motor abilities can be screened out more comprehensively, thereby improving the accuracy of identifying movement disorders. These parameters cover multiple aspects such as physiological characteristics, functional tests, cardiopulmonary function, and neurological function, and help to exclude interference factors caused by individual differences.

[0080] S3: Based on the matching index and the basic motion parameters of the user, screen reference samples for the user from the reference database; the method is as follows:

[0081] Encode the basic motion parameters of the user and form a basic parameter vector;

[0082] Calculate the matching index between the basic parameter vector and the first feature vector of each historical data in the reference database. The formula is as follows:

[0083] ;

[0084] Among them, S represents the matching index between the basic parameter vector and the first feature vector of any historical data in the reference database; represents the i-th element in the basic parameter vector, and the value range of i is 1, 2,..., n, where n is the dimension of the basic parameter vector and the first feature vector of any historical data; represents the i-th element in the first eigenvector; represents the adjustment coefficient, which is set by those skilled in the art based on actual requirements; represents the penalty factor of

[0085] ;

[0086] wherein, represents the minimum threshold of the gap between and which is set by those skilled in the art based on actual requirements. Through the calculation of the above matching index, on the one hand, it encourages the cumulative minimum of the differences between the corresponding elements between the basic parameter vector and the first eigenvector of any historical data; on the other hand, through the adjustment of the adjustment coefficient and the penalty factor

[0087] it can avoid too large differences between any set of corresponding elements between the basic parameter vector and the first eigenvector of any historical data, thus ensuring a relatively high similarity between the first eigenvector with a larger matching index and the basic parameter vector, and all sets of corresponding elements are relatively close.

[0088] S4: Extract the motion features of the user's motion data to obtain the user's motion feature indicators, and determine the tremor synchronization feature and motion coordination ability of the user based on the motion feature indicators;

[0089] The motion feature indicators include tremor feature indicators, activity feature indicators, and combined feature indicators;

[0090] The tremor feature indicators include wrist tremor frequency, wrist tremor amplitude, head rotation frequency, head rotation amplitude, and posture deviation index; Refer to Figure 2 , the extraction methods of wrist tremor frequency and wrist tremor amplitude are as follows:

[0091] Extract the first tremor data segment from the acceleration data at the user's wrist when at rest;

[0092] Perform DC component removal and filtering on the first tremor data segment, specifically as follows:

[0093] Calculate the mean value of the acceleration data in the first tremor data segment, and subtract the mean value from each acceleration data in the first tremor data segment;

[0094] A band - pass filter is used to filter and denoise the first tremor data segment. For example, the upper cut - off frequency of the band - pass filter is 8 Hz, and the lower cut - off frequency is 4 Hz.

[0095] Calculate the average peak value of the first tremor data segment as the wrist tremor amplitude. When calculating the average peak value, the absolute value of the peak value of the acceleration data in the first tremor data segment is used.

[0096] Perform a frequency - domain transformation on the first tremor data segment and extract the wrist tremor frequency. For example, convert the first tremor data segment to the frequency domain through FFT, plot the spectrogram, calculate the spectral energy, and the frequency corresponding to the maximum energy is the wrist tremor frequency.

[0097] One of the typical symptoms of Parkinson's disease patients is tremor, especially resting tremor. By measuring the peak value in the accelerometer data, the amplitude of the tremor can be estimated; and through frequency - domain analysis, the main frequency components of the tremor can be determined. These information helps to monitor the severity of the tremor.

[0098] The methods for extracting the head rotation frequency and head rotation amplitude are as follows:

[0099] Extract a second tremor data segment from the angular velocity data of the user's head at rest. The second tremor data segment corresponds to the first tremor data segment in time; that is, the start time and end time of the first tremor data segment and the second tremor data segment are the same.

[0100] Perform DC component removal and filtering on the second tremor data segment.

[0101] Calculate the average peak value of the second tremor data segment as the head rotation amplitude. When calculating the average peak value, the absolute value of the peak value of the angular velocity data in the second tremor data segment is used.

[0102] Perform a frequency - domain transformation on the second tremor data segment and extract the head rotation frequency. For example, convert the second tremor data segment to the frequency domain through FFT, plot the spectrogram, calculate the spectral energy, and the frequency corresponding to the maximum energy is the head rotation frequency.

[0103] Some movement disorders of Parkinson's disease patients involve the rotation of the body or limbs, such as the involuntary rotation of the head. Understanding the intensity of these rotational movements helps to identify whether there is abnormal rotational behavior.

[0104] The method for extracting the posture deviation index is as follows: Extract the chest acceleration data in the vertical direction when the user is stationary as the vertical acceleration; the posture deviation index is the absolute value of the difference between the vertical acceleration and the gravitational acceleration. The gravitational acceleration is 9.8 m / s². When the user stands, ideally, the center of gravity of the body should be on the vertical line. However, due to reasons such as muscle weakness and loss of balance, the center of gravity may shift during actual standing. When the user is stationary, the vertical acceleration measured by the accelerometer should be close to the gravitational acceleration, that is, 9.8 m / s². If the vertical acceleration measured by the accelerometer deviates from the gravitational acceleration, it indicates that the body's center of gravity has shifted. For example, if the vertical acceleration is less than 9.8 m / s², it may be because the body is leaning forward; if it is greater than 9.8 m / s², it may be that the body is leaning backward. Parkinson's disease patients often exhibit balance problems, such as unsteadiness when standing. Evaluate the patient's postural stability and detect whether there is a risk of falling.

[0105] The activity characteristic indicators include gait cycle, step length, step width, and trunk stability index; among them, the gait cycle, step length, and step width are calculated based on the acceleration data at the ankle of the user during movement; by detecting the periodic pattern in the acceleration data at the ankle, the gait cycle can be calculated. The acceleration data at the ankle is three-axis acceleration data, and the displacement in different directions can be calculated through the three-axis acceleration data, so as to obtain the step length and step width. Gait disorders are one of the common symptoms of Parkinson's disease patients, including gait freezing, festinating gait, etc. By analyzing the periodic pattern in the accelerometer data, the step length and step width can be calculated, and these parameters can reflect whether the patient's gait pattern is normal. In addition, the change in gait periodicity can also help identify whether there is a situation of gait instability.

[0106] The trunk stability index is the reciprocal of the variance of the chest acceleration data; intercept a section of the chest acceleration data during the user's movement and calculate the variance. If the trunk is stable, the variance of the chest acceleration is small and the trunk stability index is high.

[0107] The combined characteristic indicators include start delay and movement duration; the start delay is the time required for the user to start moving from the stationary state when performing a specified action; the movement duration is the time required for the user to complete the specified action. For example, when the user performs the action of raising the arm, the start delay and movement duration can be extracted based on the acceleration data at the wrist. Bradykinesia refers to the difficulty of the patient in starting an action and the slow movement process. By measuring the time required to start moving from rest (start delay) and the time required to complete a specific task (movement duration), the degree of bradykinesia can be quantified.

[0108] The tremor synchronization characteristic is represented by the tremor synchronization index; refer to Figure 3, the method for determining the tremor synchronization characteristics of the user is as follows:

[0109] Integrate the acceleration data in the first tremor data segment to obtain a third tremor data segment; the third tremor data segment is obtained by integrating the first tremor data segment in the time domain, and it contains the linear velocity data corresponding to the acceleration data at the wrist.

[0110] Calculate the cross-correlation coefficients between the second tremor data segment and the third tremor data segment at different lag times;

[0111] Extract the maximum cross-correlation coefficient between the second tremor data segment and the third tremor data segment and the corresponding lag time, and assign a value to the tremor synchronization index; specifically as follows:

[0112] Let the maximum cross-correlation coefficient between the second tremor data segment and the third tremor data segment be , and the corresponding lag time be ; if is greater than a preset first cross-correlation threshold, and is less than a preset first lag threshold, then the tremor synchronization index is assigned a value of 1; otherwise, the tremor synchronization index is assigned a value of 0.

[0113] The above method for assigning the tremor synchronization index evaluates whether the tremors are synchronized by analyzing the phase relationship between the accelerometer data at the wrist and the gyroscope data at the head, that is, using the cross-correlation function to detect the consistency of tremors between different body parts. When the cross-correlation coefficient is high (greater than the preset first cross-correlation threshold) and the corresponding lag time is close to 0 (less than the preset first lag threshold), it indicates that the tremor synchronization between the wrist and the head is high. By analyzing the phase relationship between the accelerometer data of different parts, it can be found whether the tremors occur synchronously in different parts. This helps to further distinguish the types of tremors or evaluate the effect of tremor control treatment.

[0114] The motor coordination ability is represented by a motor coordination index; the method for determining the motor coordination ability of the user is as follows:

[0115] The method for extracting the motor coordination index is as follows:

[0116] Extract a first motion data segment from the acceleration data at the wrist when the user is walking; extract a second motion data segment from the acceleration data at the ankle when the user is walking;

[0117] Calculate the cross-correlation coefficients between the first motion data segment and the second motion data segment at different lag times;

[0118] Extract the maximum cross-correlation coefficient between the first motion data segment and the second motion data segment and the corresponding lag time, and assign a value to the motor coordination index; specifically as follows:

[0119] Let the maximum cross - correlation coefficient between the first motion data segment and the second motion data segment be , and the corresponding lag time be ; If is greater than a preset second cross - correlation threshold, and is less than a preset second lag threshold, then the motion coordination index is assigned a value of 1; otherwise, the motion coordination index is assigned a value of 0.

[0120] The above - mentioned method for assigning the motion coordination index detects the coordination between the hands and feet when the user is walking by analyzing the cross - correlation coefficient between the acceleration data at the wrist and the acceleration data at the ankle. When the cross - correlation coefficient is high (greater than the preset second cross - correlation threshold) and the corresponding lag time is close to 0 (less than the preset second lag threshold), it indicates that the coordination between the hands and feet when the user is walking is good. Patients with Parkinson's disease may experience a decline in motor coordination, that is, the coordinated movements between different parts of the body are impaired. By analyzing the action time lag and synchronization between different limbs in the acceleration data, the motor coordination ability of the patient can be evaluated.

[0121] S5: Match the motion feature indicators of the user with the reference samples, and combine the tremor synchronization feature and the motor coordination ability to obtain the motion disorder recognition result of the user. Refer to Figure 4 , the steps are as follows:

[0122] S501: Encode the motion feature indicators, tremor synchronization index, and motion coordination index of the user and form a motion feature vector;

[0123] S502: Calculate the tremor synchronization index and motion coordination index of each reference sample; after encoding the tremor synchronization index and motion coordination index of each reference sample, add them to the corresponding second feature vector;

[0124] S503: Calculate the similarity between the motion feature vector and the second feature vector of each reference sample; methods such as Euclidean distance and cosine similarity can be used to calculate the similarity;

[0125] S504: Extract the N reference samples corresponding to the N second feature vectors with the largest similarity to the motion feature vector as the voting samples; N is a positive integer, and the optimal value of N can be selected through cross - validation.

[0126] S505: Count the types of motion disorders of the voting samples, and the type of motion disorder that appears most frequently among the N voting samples is the motion disorder recognition result of the user.

[0127] Embodiment 2

[0128] This embodiment is the second embodiment of the present invention; based on the same inventive concept as Embodiment 1, this embodiment introduces a movement disorder recognition device for Parkinson's patients based on a wearable device, including a database module, a data acquisition module, a data processing module, a feature index module, and an identification module; wherein:

[0129] The database module is used to collect historical data and construct a reference database; any piece of historical data in the reference database includes a movement disorder type, a first feature vector, and a second feature vector.

[0130] The data acquisition module is used to collect the movement data of the user and obtain the basic movement parameters of the user; this module is configured with sensors such as accelerometers and gyroscopes and other test devices, and can measure the movement data and basic movement parameters of different parts of the user;

[0131] The data processing module is configured with a matching index calculation formula and a penalty factor value strategy, and is used to screen reference samples for the user;

[0132] The feature index module extracts the movement feature indexes of the user based on the movement data of the user, including tremor feature indexes, activity feature indexes, and combined feature indexes;

[0133] The identification module is used to match the movement feature indexes of the user and the reference samples to obtain the movement disorder recognition result of the user. Voting samples are screened from the reference samples by calculating the vector similarity, and based on the voting samples, the movement disorder recognition result of the user is obtained by voting.

[0134] For the specific function implementation of the above-mentioned modules, refer to the relevant content in the movement disorder recognition method for Parkinson's patients based on a wearable device described in Embodiment 1, and will not be elaborated here.

[0135] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0136] The embodiments of the present invention have been described above in conjunction with the accompanying drawings, but the present invention is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not restrictive. Under the inspiration of the present invention, those of ordinary skill in the art can also make many forms without departing from the purpose and scope of the present invention, and these all belong to the protection scope of the present invention.

Claims

1. A method for identifying movement disorders in Parkinson's disease patients based on wearable devices, characterized in that: The following steps are involved: S1: Collect historical data and build a reference database of movement disorders; S2: Collect the user's motion data and obtain the user's basic motion parameters; S3: based on the matching index and the basic motion parameters of the user, selecting reference samples for the user from a reference database; S4: extracting features from the user's motion data to obtain a motion feature index of the user, and determining the user's tremor synchronization features and motion coordination ability based on the motion feature index; the motion feature index includes a tremor feature index, an activity feature index, and a combined feature index; The tremor synchronization feature is represented by a tremor synchronization index; the method for determining the tremor synchronization feature of a user is as follows: Integrating the acceleration data in the first tremor data segment to obtain a third tremor data segment; wherein the first tremor data segment includes acceleration data at the wrist of the user when the user is stationary; Calculating the correlation coefficient between the second tremor data segment and the third tremor data segment at different lag times; wherein the second tremor data segment includes angular velocity data of the user's head when the user is stationary; Extracting the maximum mutual correlation coefficient and the corresponding lag time between the second tremor data segment and the third tremor data segment, and assigning a value to the tremor synchronization index; specifically as follows: Let the maximum correlation coefficient between the second tremor data segment and the third tremor data segment be , the corresponding lag time is ;like is greater than a preset first cross-correlation threshold, and If the tremor synchronization index is smaller than a preset first hysteresis threshold, the tremor synchronization index is assigned a value of 1; otherwise, the tremor synchronization index is assigned a value of 0; The motor coordination ability is represented by a motor coordination index; the method for determining the motor coordination ability of a user is as follows: Extracting a first motion data segment from the acceleration data at the wrist when the user is walking; extracting a second motion data segment from the acceleration data at the ankle when the user is walking; Calculating the correlation coefficient between the first motion data segment and the second motion data segment at different lag times; Extract the maximum correlation coefficient and the corresponding lag time between the first motion data segment and the second motion data segment, and assign a value to the motion coordination index; specifically as follows: Let the maximum correlation coefficient between the first motion data segment and the second motion data segment be , the corresponding lag time is ;like is greater than a preset second cross-correlation threshold, and If the value of the motion coordination index is less than a preset second hysteresis threshold, the motion coordination index is assigned a value of 1; otherwise, the motion coordination index is assigned a value of 0; S5: Matching the motion feature indicators of the user and the reference sample, and combining the tremor synchronization characteristics with the motion coordination ability to obtain the user's movement disorder identification result.

2. The method for identifying movement disorders in Parkinson's disease patients based on a wearable device as claimed in claim 1, characterized in that: Any piece of historical data in the reference database includes a movement disorder type, a first feature vector, and a second feature vector; wherein the movement disorder type is a movement disorder or a combination of m movement disorders, and m is a positive integer greater than 1; the basic movement parameters are encoded to form a first feature vector; and the movement feature index is encoded to form a second feature vector; The motion data includes acceleration data at the wrist, acceleration data at the ankle, acceleration data at the chest, and angular velocity data at the head; The user's basic sports parameters include age, gender, height, weight, range of joint motion, muscle strength, resting heart rate, and heart rate variability.

3. The method for identifying movement disorders in Parkinson's disease patients based on a wearable device as claimed in claim 2, characterized in that: The method for screening reference samples for users is as follows: Encode the user's basic motion parameters and form a basic parameter vector; Calculating a matching index between the basic parameter vector and a first feature vector of each piece of historical data in a reference database; The M pieces of historical data corresponding to the largest M matching indexes are extracted as M reference samples for user screening.

4. The method for identifying movement disorders in Parkinson's disease patients based on a wearable device as claimed in claim 3, characterized in that: The calculation formula of the matching index is as follows: ； Where S represents the matching index between the basic parameter vector and the first feature vector of any historical data in the reference database; Represents the i-th element in the basic parameter vector, where the value range of i is 1, 2, ..., n, and n is the dimension of the first eigenvector of the basic parameter vector and any historical data; represents the i-th element in the first eigenvector; represents the adjustment coefficient; express The penalty factor is as follows: ； in, express and The minimum threshold for the difference between .

5. The method for identifying movement disorders in Parkinson's disease patients based on a wearable device as claimed in claim 4, characterized in that: The tremor characteristic indexes include wrist tremor frequency, wrist tremor amplitude, head rotation frequency, head rotation amplitude, and posture deviation index; the extraction method of wrist tremor frequency and wrist tremor amplitude is as follows: Extracting a first tremor data segment from the acceleration data at the wrist of the user when the user is stationary; The first tremor data segment is subjected to a DC component removal and filtering process, specifically as follows: Calculating a mean value of the acceleration data in the first tremor data segment, and subtracting the mean value from each acceleration data in the first tremor data segment; Using a bandpass filter to perform filtering and noise reduction processing on the first tremor data segment; calculating an average peak value of the first tremor data segment as the wrist tremor amplitude; Performing frequency domain transformation on the first tremor data segment and extracting the wrist tremor frequency; The method for extracting the head rotation frequency and the head rotation amplitude is as follows: extracting a second tremor data segment from the angular velocity data of the head when the user is stationary; performing DC component removal and filtering processing on the second tremor data segment; calculating the average peak value of the second tremor data segment as the head rotation amplitude; performing frequency domain transformation on the second tremor data segment, and extracting the head rotation frequency; The posture deviation index is extracted as follows: extract chest acceleration data of the user in the vertical direction when the user is stationary as the vertical acceleration; the posture deviation index is the absolute value of the difference between the vertical acceleration and the gravity acceleration.

6. The method for identifying movement disorders in Parkinson's disease patients based on a wearable device as claimed in claim 5, characterized in that: The activity characteristic index includes gait cycle, step length, step width, and trunk stability index; wherein the gait cycle, step length, and step width are calculated based on the acceleration data at the ankle when the user is exercising; the trunk stability index is the inverse of the variance of the chest acceleration data; The combined characteristic indicators include starting delay and motion duration; the starting delay is the time required for the user to go from a stationary state to start motion when performing a specified action; the motion duration is the time required for the user to complete the specified action.

7. The method for identifying movement disorders in Parkinson's disease patients based on a wearable device as claimed in claim 6, characterized in that: The method of matching the motion feature indicators of the user with the reference sample and combining the tremor synchronization feature with the motion coordination ability to obtain the user's movement disorder identification result is as follows: S501: Encode the user's motion characteristic index, tremor synchronization index, and motion coordination index to form a motion characteristic vector; S502: Calculate the tremor synchronization index and the movement coordination index of each reference sample; encode the tremor synchronization index and the movement coordination index of each reference sample and add them to the corresponding second feature vector; S503: Calculate the similarity between the motion feature vector and the second feature vector of each reference sample; S504: extracting N reference samples corresponding to the N second feature vectors having the greatest similarity to the motion feature vectors as voting samples; S505: Count the movement disorder types of the voting samples, and the movement disorder type that appears most frequently in the N voting samples is the movement disorder recognition result of the user.

8. A device for identifying movement disorders in patients with Parkinson's disease based on a wearable device, used to implement the method for identifying movement disorders in patients with Parkinson's disease based on a wearable device according to any one of claims 1 to 7, characterized in that: It includes database module, data acquisition module, data processing module, characteristic index module and identification module; among which: The database module is used to collect historical data and build a reference database; The data collection module is used to collect the user's motion data and obtain the user's basic motion parameters; The data processing module is equipped with a matching index calculation formula and a penalty factor value selection strategy to screen reference samples for users; The characteristic index module extracts the user's motion characteristic index based on the user's motion data; The recognition module is used to match the motion feature indicators of the user with those of the reference sample to obtain the user's movement disorder recognition result.

Images