Good gait abnormity monitoring system based on body surface electromyographic signals

Abstract

The invention discloses an old man gait abnormity monitoring system based on body surface electromyographic signals, and relates to the technical field of gait abnormity monitoring, and the system comprises a configuration module which is used for obtaining a foot bearing device set of a user, and constructing three groups of foot bearing device test scenes; the inertial measurement module is used for collecting gait data of a user and establishing gait key parameters; the myoelectricity data acquisition module is used for establishing a myoelectricity test data set; the labeling module is used for establishing a gait cycle window and establishing a myoelectricity-gait fusion fingerprint spectrum; and the early warning module is used for reading the real-time acquired data and outputting an early warning signal. The technical problems that in the prior art, monitoring of the gait abnormity of the old man is not accurate enough, abnormity is difficult to find in time and early warning is difficult, and consequently the old man faces a high falling risk and cannot be protected in time when walking are solved, the accuracy and timeliness of monitoring of the gait abnormity of the old man are improved, and the safety of the old man is improved. Therefore, the walking safety of the old people is guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of gait abnormality monitoring, and particularly to an elderly gait abnormality monitoring system based on surface electromyography signals. Background Art

[0002] With the acceleration of the global aging process, falls among the elderly have become a serious public health problem. Falls can not only cause physical injuries to the elderly, increasing the risk of fractures and other injuries, but also trigger psychological fears and reduce the quality of life. Traditional gait monitoring methods, such as manual observation, are inefficient and highly subjective, making it difficult to achieve continuous and accurate monitoring. Although some existing wearable devices can monitor simple information such as the number of steps and movement distance, they are unable to deeply analyze the subtle changes in gait, have limited ability to identify early gait abnormalities, and are unable to effectively integrate multi-source data for comprehensive and real-time gait abnormality monitoring, thus making it difficult to meet the actual needs of ensuring the walking safety of the elderly.

[0003] The prior art has the technical problems that the monitoring of elderly gait abnormalities is not accurate enough, it is difficult to detect abnormalities and give early warnings in a timely manner, resulting in a high risk of falls for the elderly when walking but they cannot be protected in time. Summary of the Invention

[0004] The present application provides an elderly gait abnormality monitoring system based on surface electromyography signals, which is used to solve the technical problems in the prior art that the monitoring of elderly gait abnormalities is not accurate enough, it is difficult to detect abnormalities and give early warnings in a timely manner, resulting in a high risk of falls for the elderly when walking but they cannot be protected in time.

[0005] In view of the above problems, the present application provides an elderly gait abnormality monitoring system based on surface electromyography signals.

[0006] The present application provides an elderly gait abnormality monitoring system based on surface electromyography signals, and the system includes:

[0007] Configuration module, which is used to obtain the set of foot-bearing devices of the user, classify and identify the influence of the foot-bearing devices on gait, and construct three groups of foot-bearing device test scenarios; inertial measurement module, which is used to collect the gait data of the user under the three groups of foot-bearing device test scenarios respectively and establish key gait parameters, where the key gait parameters include step length, step frequency, gait cycle, and gait symmetry; electromyogram data acquisition module, which is used to synchronously collect electromyogram signals and establish an electromyogram test data set, and the parts collected by the electromyogram data acquisition module include the quadriceps femoris, gastrocnemius, and gluteus maximus; annotation module, which is used to establish a gait cycle window based on the key gait parameters, and after using the gait cycle window to divide the gait segments of the key gait parameters and the electromyogram test data set, establish an electromyogram-gait fusion fingerprint map; warning module, which is used to read the real-time acquisition data of the electromyogram data acquisition module, predict the abnormal gait trend based on the real-time acquisition data using the electromyogram-gait fusion fingerprint map, and output a warning signal.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] Configuration module, which is used to obtain the set of foot-bearing devices of the user, classify and identify the influence of the foot-bearing devices on gait, and construct three groups of foot-bearing device test scenarios; inertial measurement module, which is used to collect the gait data of the user under the three groups of foot-bearing device test scenarios respectively and establish key gait parameters; electromyogram data acquisition module, which is used to synchronously collect electromyogram signals and establish an electromyogram test data set; annotation module, which is used to establish a gait cycle window based on the key gait parameters, and after using the gait cycle window to divide the gait segments of the key gait parameters and the electromyogram test data set, establish an electromyogram-gait fusion fingerprint map; warning module, which is used to read the real-time acquisition data of the electromyogram data acquisition module, predict the abnormal gait trend based on the real-time acquisition data, and output a warning signal. It achieves the technical effect of improving the accuracy and timeliness of the abnormal gait monitoring of the elderly, and thus ensuring the walking safety of the elderly. Description of the Drawings

[0010] In order to more clearly illustrate the technical solutions in 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 be obtained based on these drawings.

[0011] Figure 1 It is a schematic structural diagram of a system for monitoring abnormal gait of the elderly based on surface electromyogram signals provided by an embodiment of this application;

[0012] Figure 2This is a schematic structural diagram of the annotation module of the elderly gait abnormality monitoring system based on surface electromyogram signals provided by the embodiments of the present application.

[0013] Explanation of reference numerals: Configuration module 10, inertial measurement module 20, electromyogram data acquisition module 30, annotation module 40, warning module 50. Detailed implementation manners

[0014] The present application provides an elderly gait abnormality monitoring system based on surface electromyogram signals, which is used to solve the technical problems in the prior art that the monitoring of elderly gait abnormalities is not accurate enough, it is difficult to detect abnormalities and give warnings in time, resulting in a high risk of falling when the elderly walk but they cannot be protected in time.

[0015] Next, the technical solutions in the embodiments of the present application will be clearly and completely described with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.

[0016] Embodiment, as Figure 1 shown, the present application provides an elderly gait abnormality monitoring system based on surface electromyogram signals, and the system includes:

[0017] A configuration module 10, configured to obtain a set of foot bearing devices of a user, classify and identify the gait influence of the set of foot bearing devices, and construct three groups of foot bearing device test scenarios.

[0018] Specifically, the configuration module 10 obtains the set of foot-bearing devices by interacting with the user or reading from a preset database. The interaction method can be to let the user select the type of foot-bearing device they use in the system interface or manually input relevant information; the preset database stores information on various common foot-bearing devices. After obtaining the device set, classification and identification are carried out based on the physical properties of the devices, such as the thickness and material of the insole, the hardness of the shoe sole, the heel height of the shoe, etc., as well as the relevant data on the impact of these properties on gait in past research. For the insole, it is classified into three categories: thin, medium, and thick according to the thickness, and the impact of different thicknesses on gait in terms of step length, step frequency, etc. is analyzed; for the shoes, they are classified into categories such as flat hard-soled shoes, flat soft-soled shoes, and high-heel hard-soled shoes according to comprehensive factors such as the hardness of the shoe sole and the heel height, and their impact on gait symmetry and gait cycle is evaluated. Based on these classifications, three groups of foot-bearing test scenarios are constructed from different perspectives such as affecting step length, step frequency, and gait symmetry. For example, the first group of scenarios is set to test the impact of insoles of different thicknesses on step length and step frequency, and different combinations of insole thicknesses are selected; the second group focuses on the impact of sole hardness and heel height on gait symmetry, and shoes with different sole hardnesses and heel heights are configured; the third group considers the impact on gait cycle under the mixture of various factors, and different types of insoles and shoes are combined, thus completing the construction of three groups of targeted foot-bearing test scenarios.

[0019] The inertial measurement module 20 is used to collect the gait data of the user under three groups of foot-bearing device test scenarios respectively and establish key gait parameters, and the key gait parameters include step length, step frequency, gait cycle, and gait symmetry.

[0020] Specifically, high-precision inertial sensors such as accelerometers and gyroscopes are used. These sensors can accurately sense the motion state of the human body during walking. In each set of test scenarios, the user is allowed to move in a normal walking manner, and the motion data of the user during walking, including information such as acceleration and angular velocity, is collected in real time. Key information related to gait is extracted from these raw data. For the calculation of step length, the displacement distance of both feet of the user within a complete walking cycle is analyzed, and combined with the changes in acceleration and angular velocity, the length of each step is accurately obtained. The step frequency is determined by counting the number of steps the user takes within a unit time. The gait cycle is the complete time interval from when one heel touches the ground to when the same heel touches the ground again. The module identifies this time period through precise analysis of the motion data. The evaluation of gait symmetry is achieved by comparing the differences in motion parameters of the left and right limbs during walking, such as the differences in step length, step frequency, and motion trajectory between the left and right feet. Through comprehensive data collection and analysis in three different test scenarios of foot-bearing devices, accurate and detailed key gait parameters are established. These parameters not only reflect the gait characteristics of the user under different foot-bearing conditions but also provide an important basis for subsequent analysis of EMG data and the establishment of EMG-gait fusion fingerprint maps, which helps improve the accuracy and reliability of the entire system for monitoring abnormal gaits of the elderly.

[0021] The electromyography (EMG) data acquisition module 30 is used to synchronously acquire EMG signals and establish an EMG test data set. The acquisition sites of the EMG data acquisition module include the quadriceps femoris, gastrocnemius, and gluteus maximus.

[0022] Specifically, the EMG data acquisition module 30 is responsible for collecting key muscle activity information. By precisely pasting surface electrodes on the quadriceps femoris, gastrocnemius, gluteus maximus and other parts of the user, the EMG signals of these muscles during movement are synchronously acquired. When the user participates in walking in three test scenarios of foot-bearing devices, the module continuously collects EMG signals at a high sampling frequency to ensure that no subtle changes in muscle electrical activity are missed. During the acquisition process, it preprocesses the weak electrical signals generated by these muscles during contraction and relaxation, such as amplification and filtering, to remove interference signals and improve the signal quality. As the test progresses, the processed EMG signals are stored in order of time, and an EMG test data set is gradually established. This data set records the electrical activity of each key muscle of the elderly under different foot-bearing conditions.

[0023] The annotation module 40 is used to establish a gait cycle window based on the key gait parameters. After using the gait cycle window to divide the gait key parameters and the EMG test data set into gait segments, an EMG-gait fusion fingerprint map is established.

[0024] Specifically, by analyzing parameters such as step length, step frequency, gait cycle, and gait symmetry, a suitable gait cycle duration is determined, and a gait cycle window is established to segment continuous gait data and electromyography data. Using the established gait cycle window, gait key parameters and electromyography test data sets are divided into gait segments. Within each divided gait segment, the gait segments under three groups of foot-bearing device test scenarios are segmented and labeled, and an order association between them is created to facilitate subsequent tracking and comparative analysis. Electromyography feature extraction is performed on the segmented gait segments to obtain electromyography feature extraction results such as time-domain features, frequency-domain features, and non-linear features. Then, muscle activation time delay constraints, electromyography-step frequency coordination constraints, left and right muscle group activation symmetry constraints, etc. are established. According to these constraints, the electromyography feature extraction results corresponding to the same segmented gait segments are spliced with the gait key parameters to form a unified feature vector. These unified feature vectors integrate information from both gait and electromyography, comprehensively reflecting the characteristics of each gait segment. Finally, based on these unified feature vectors and the order association, the three groups of foot-bearing device test scenarios are sorted first, and load thickness granularity segmentation is performed. Then, the unified feature vectors with order association are used as boundary nodes for electromyography-gait fusion prediction under the granularity segmentation results, and finally, an electromyography-gait fusion fingerprint map is established.

[0025] An early warning module 50, configured to read the real-time acquisition data of the electromyography data acquisition module, and perform abnormal prediction of gait trends based on the real-time acquisition data by using the electromyography-gait fusion fingerprint map, and output an early warning signal.

[0026] Specifically, real-time collected data is read. Once the real-time data is received, it is matched and identified according to the load thickness of the foot-bearing device currently used by the elderly, and then the myoelectric-gait fusion fingerprint map is initialized to ensure the accuracy and pertinence of subsequent comparison and analysis. After initialization, the real-time collected data is compared in detail with the initialized myoelectric-gait fusion fingerprint map. A temporal gait deviation is established between the two through the temporal gait deviation establishment sub-module, which measures the difference between the real-time gait data and the normal gait pattern in the fingerprint map in the time series. The trend score is calculated using these temporal gait deviations. This score can quantitatively reflect the development direction and speed of gait deviation, enabling the system to have a clearer judgment on the development trend of gait abnormalities. At the same time, the static comparison result between the myoelectric-gait fusion fingerprint map and the real-time collected data is obtained, and the difference between the two is analyzed from a static perspective. Finally, the gait trend abnormality prediction is carried out by comprehensively considering the static comparison result and the trend score. If the prediction result indicates that there is an abnormality in the elderly's gait trend and reaches the preset danger threshold, an early warning signal is immediately output. This early warning signal can not only timely remind the elderly of the possible walking risks, but also configure corresponding early warning response plans according to the warning level, send the early warning signal to associated devices, such as the mobile phones of family members, the monitoring terminals of community medical centers, etc., and synchronously send the location information of the elderly, so that relevant personnel can take timely measures to ensure the safety of the elderly. In addition, the actual user status mapped to the early warning signal is obtained, and the early warning trust verification of the early warning signal is carried out to lock the key early warning features, which are used to optimize the prediction management of the myoelectric-gait fusion fingerprint map, further improving the monitoring accuracy and reliability of the system.

[0027] In a possible implementation manner, as Figure 2 shown, the annotation module 40 includes:

[0028] The grouping annotation sub-module is used to segment and identify the gait segments of three groups of foot-bearing device test scenarios, and create the sequential association of the segmented gait segments.

[0029] The signal processing sub-module is used to extract the myoelectric features of the segmented gait segments respectively, and establish the myoelectric feature extraction results, which include time-domain features, frequency-domain features, and non-linear features.

[0030] The feature fusion sub-module is used to splice the myoelectric feature extraction results and gait key parameters corresponding to the same segmented gait segments to form a unified feature vector.

[0031] The fingerprint map construction sub-module is used to establish the myoelectric-gait fusion fingerprint map according to the unified feature vector and the sequential association.

[0032] Specifically, after obtaining the gait data and the collected electromyography signals acquired in three groups of foot-bearing device test scenarios, the continuous gait data in each test scenario is segmented according to preset rules. These rules may be based on gait cycles, specific action start and end points, or time intervals, etc. For example, segmented by a complete gait cycle, starting from the first heel strike of one foot until the next heel strike of the same foot, the gait data during this period is divided into a segment. For the three different foot-bearing device test scenarios, the grouping and annotation sub-module assigns unique identifiers to each segment. For example, in the first group of scenarios, the first gait segment is marked as "A1-001", the second as "A1-002", and so on; the segments in the second group of scenarios are marked as "A2-001", "A2-002", etc.; the same applies to the third group of scenarios. After completing the segmentation and identification of gait segments, an order association between the segmented gait segments is created to record the sequence of these segments during the entire test. Through this order association, subsequent modules can analyze the gait segments in the correct time sequence. For example, when analyzing gait trends or continuous changes in muscle activity, the data of different segments can be accurately traced and compared, providing a clear and orderly data basis for establishing an electromyography-gait fusion fingerprint map and ultimately for gait abnormality monitoring.

[0033] For the electromyography signals in each segmented gait segment, a variety of advanced analysis methods are used for feature extraction. In the time domain, the signal processing sub-module calculates the root mean square value to reflect the average energy level of the electromyography signal and the intensity of muscle contraction; counts the zero-crossing rate to understand the frequency of the electromyography signal crossing the zero point on the time axis and infer the change rate of muscle activity; and also analyzes the waveform length, that is, the total length of the electromyography signal within a period of time, to reflect the complexity of the signal. These time-domain features can intuitively show the activity state of the muscle at a specific moment. For frequency-domain features, the fast Fourier transform algorithm is used to convert the time-domain signal to the frequency domain to extract the power spectral density parameters. The power spectral density shows the energy distribution of the electromyography signal at different frequency components, helping to judge the force pattern and fatigue degree during muscle contraction. Different muscle activities often correspond to specific frequency ranges, and by analyzing the frequency-domain features, a deeper understanding of the muscle working mechanism can be obtained. The signal processing sub-module also extracts non-linear features such as approximate entropy and fractal dimension. Approximate entropy is used to measure the complexity and regularity of the electromyography signal. The larger the value, the more complex and irregular the signal, which is related to muscle fatigue or abnormal movement control; the fractal dimension reflects the self-similarity and space-filling ability of the signal, and can reveal the internal dynamic characteristics of muscle activity. By comprehensively extracting time-domain, frequency-domain, and non-linear features, the signal processing sub-module establishes a detailed electromyography feature extraction result.

[0034] Obtain muscle activation time delay constraints, EMG-cadence synergy constraints, and left and right muscle group activation symmetry constraints from the constraint establishment submodule. Based on these constraints, for the same segmented gait segments, the EMG feature extraction results in the time domain, frequency domain, and nonlinearity obtained by the signal processing submodule are sequentially spliced ​​with key gait parameters such as step length, cadence, gait cycle, and gait symmetry collected by the inertial measurement module and processed by the annotation module. For example, according to the order of muscle activation time, the EMG features at a specific moment are combined with the corresponding gait parameters to make the EMG information and gait information correlated with each other. After such splicing operations, the originally scattered EMG and gait data form a unified feature vector containing multi-dimensional information.

[0035] Receive the unified feature vector and the sequential association information created by the grouping and annotation submodule. Using the sequential association, the three groups of foot-bearing device test scenes are sorted, and on this basis, the bearing thickness granularity segmentation is performed, and the test scenes with different bearing thicknesses are subdivided into multiple granularity levels to establish the granularity segmentation results. Such segmentation helps to analyze the gait characteristics under different bearing conditions in more detail. Subsequently, the unified feature vector of sequential association is used as the boundary node, and the electromyography-gait fusion prediction is performed under the framework of the granularity segmentation result. According to the electromyography and gait information at the boundary node, combined with the characteristics of different granularity levels, the electromyography-gait relationship in the entire test scene is predicted and inferred. Through this fusion prediction, the information from all aspects is integrated, and finally the electromyography-gait fusion fingerprint map is constructed, which contains the comprehensive characteristics of the gait and electromyography of the elderly under different foot-bearing devices, and improves the accuracy of monitoring the abnormal gait of the elderly.

[0036] In a possible implementation, the feature fusion submodule includes:

[0037] The constraint establishment submodule is used to establish muscle activation time delay constraints, electromyography-cadence synergy constraints, and left and right muscle group activation symmetry constraints.

[0038] The splicing submodule is used to perform splicing of the electromyographic feature extraction results and the key gait parameters according to the muscle activation time delay constraint, the electromyographic-step frequency synergy constraint, and the left and right muscle group activation symmetry constraint.

[0039] Specifically, the constraint establishment sub-module constructs muscle activation time delay constraints, electromyogram-step frequency coordination constraints, and left-right muscle group activation symmetry constraints through data collection, feature extraction, data analysis, and model establishment. In the data collection stage, electromyogram sensors with high precision are used to collect electromyogram signals of key muscles such as the quadriceps femoris, gastrocnemius, and gluteus maximus. At the same time, an inertial measurement unit is used to obtain information such as the acceleration and angular velocity of the gait, and multiple groups of data of a large number of different individuals in the natural walking state are recorded. In terms of feature extraction, the collected electromyogram signals are preprocessed, such as filtering and denoising, and then time-domain and frequency-domain features such as root mean square value and average power frequency are extracted, and key gait parameters such as step length, step frequency, and gait cycle are extracted. For the muscle activation time delay constraint, the starting moment of the electromyogram signal is analyzed, and by comparing the starting time points of the electromyogram signals of different muscles, the time difference between them is calculated. Using statistical methods, such as mean and standard deviation analysis, the range of muscle activation time delay under normal conditions is determined, so as to establish the muscle activation time delay constraint. When constructing the electromyogram-step frequency coordination constraint, the extracted electromyogram features are correlated with the step frequency parameters, and the correlation analysis method is used to explore the quantitative relationship between the electromyogram features and the step frequency, and a mathematical model between the two is constructed to determine the electromyogram-step frequency coordination constraint. When establishing the left-right muscle group activation symmetry constraint, the electromyogram features of the same-name muscles on the left and right sides are extracted respectively, and the difference between the feature values on the left and right sides is calculated. Using the hypothesis testing method, it is judged whether the activation of the muscles on both sides is symmetric, and then the threshold range of the left-right muscle group activation symmetry is determined to form the left-right muscle group activation symmetry constraint.

[0040] Working according to the constraint conditions determined by the constraint establishment sub-module, when splicing the electromyogram feature extraction results and the key gait parameters, according to the muscle activation time delay constraint, the electromyogram features at a specific moment are matched and combined with the corresponding key gait parameters. For example, at the moment when the gluteus maximus is activated, the time-domain, frequency-domain, and non-linear features of the gluteus maximus at this time are integrated with the gait parameters such as step length and step frequency at the same moment. According to the electromyogram-step frequency coordination constraint, when the step frequency changes, it is ensured that the changed electromyogram features can be accurately spliced with the key gait parameters under the new step frequency, reflecting the coordination relationship between the two. According to the left-right muscle group activation symmetry constraint, the electromyogram features and the corresponding key gait parameters of the same-name muscles on the left and right sides are symmetrically processed. If they meet the symmetry requirements, they are spliced together. In this way, the splicing sub-module combines the scattered electromyogram features and key gait parameters to form a unified feature vector.

[0041] In a possible implementation manner, the fingerprint map construction sub-module includes:

[0042] The granularity segmentation sub-module is used to perform bearing thickness granularity segmentation and establish granularity segmentation results after sorting the three groups of foot-bearing device test scenarios according to the sequential association.

[0043] The calibration prediction sub-module is used to take the sequentially associated unified feature vectors as boundary nodes, perform the myoelectric-gait fusion prediction under the granularity segmentation result, and establish the myoelectric-gait fusion fingerprint map according to the fusion prediction result.

[0044] Specifically, according to the sequential association created by the grouping annotation sub-module, the three groups of foot bearing device test scenarios constructed by the configuration module are sorted. This sorting is based on a certain logical relationship of the influence of different test scenarios on gait. For example, it is sorted according to the bearing thickness from thin to thick. After the sorting is completed, the granularity segmentation operation is performed for the key factor of bearing thickness. It divides the bearing thickness into multiple different granularity levels. For example, the bearing thickness range is equally divided into several intervals, and each interval is a granularity level. Through such an operation, the granularity segmentation result is established.

[0045] Using the support vector machine (SVM) algorithm, first, guided by the sequential association created by the grouping annotation sub-module, the unified feature vectors generated by the feature fusion sub-module are obtained and used as boundary nodes. These boundary nodes contain rich myoelectric features and gait key parameter information. For the granularity segmentation result, each granularity level is regarded as an independent data set. For each data set, the myoelectric-gait fusion prediction is performed using the support vector machine algorithm. Taking a certain granularity level as an example, the unified feature vectors within this level are used as input data to train the support vector machine. During the training process, the SVM finds an optimal classification hyperplane to separate different classes of data (corresponding to different myoelectric-gait states) as much as possible. In the prediction stage, for a new combination of myoelectric and gait data, the SVM determines its belonging class according to the classification hyperplane obtained from training, so as to predict the association relationship between the myoelectric signal and the gait parameters at this bearing thickness granularity. After completing the prediction of all granularity levels, the myoelectric-gait fusion fingerprint map is constructed according to these fusion prediction results. It integrates the prediction results of each granularity level and arranges the results of different levels in order according to the sequential association. For example, for each granularity level of the bearing thickness from thin to thick, the predicted myoelectric-gait relationship information is sequentially spliced to form a complete map structure. In the map, each part represents the myoelectric-gait fusion characteristics under a specific bearing thickness, and finally, a myoelectric-gait fusion fingerprint map that can comprehensively reflect the relationship between the gait and myoelectric signal of the elderly under different bearing conditions is constructed.

[0046] In a possible implementation manner, the early warning module 50 includes:

[0047] The matching initialization sub-module is used to initialize the myoelectric-gait fusion fingerprint map after receiving the real-time acquisition data and calling the calibration prediction sub-module to perform the matching recognition of the current bearing thickness.

[0048] A matching prediction sub-module, which is used to predict the abnormal trend of gait of real-time collected data according to the initialized myoelectric-gait fusion fingerprint map and output a warning signal.

[0049] Specifically, the real-time collected data is preprocessed, and the sliding average filtering algorithm is used to remove the high-frequency noise in the data and improve the data quality. After that, key features are extracted from the preprocessed data, such as the root mean square value and average power frequency of the myoelectric signal, and the step length and step frequency of the gait data. Then, the calibration prediction sub-module is called. In this module, the constructed myoelectric-gait fusion fingerprint maps under different bearing thicknesses are stored for matching and recognition of the current bearing thickness. The cosine similarity algorithm is used to calculate the similarity between the feature vector of the real-time collected data and the feature vectors of each bearing thickness granularity level in the fingerprint map. The cosine similarity algorithm measures the directional similarity of two vectors by calculating the cosine value of the included angle between them. The closer the value is to 1, the more similar they are. The bearing thickness granularity level with the highest similarity is found and determined as the bearing thickness corresponding to the current real-time data. Finally, based on the recognized current bearing thickness, the feature data and pattern information under this bearing thickness are extracted from the myoelectric-gait fusion fingerprint map to complete the initialization of the myoelectric-gait fusion fingerprint map. These initialized map information are stored in a temporary cache for the subsequent matching prediction sub-module to access and use quickly.

[0050] After the matching initialization sub-module completes the initialization of the myoelectric-gait fusion fingerprint map, it continuously obtains the real-time collected myoelectric signals and gait data, processes these data in the same feature extraction method as when constructing the fingerprint map, and obtains the myoelectric features and key gait parameters of the real-time data. These real-time feature parameters are compared and analyzed with the normal patterns in the initialized myoelectric-gait fusion fingerprint map. The dynamic time warping (DTW) algorithm is used to calculate the similarity of the real-time gait data and the standard gait pattern in the fingerprint map in the time series to measure the overall morphological difference of the gait. At the same time, through statistical analysis methods, the distribution of the myoelectric features of the real-time myoelectric features and the corresponding muscles in the fingerprint map in the normal state is compared, such as comparing the root mean square value, frequency components, etc. If the difference between the real-time data and the normal pattern in the fingerprint map exceeds the preset threshold, the matching prediction sub-module will determine that there is an abnormal gait trend. For example, when the real-time step frequency continuously deviates from the normal step frequency range in the fingerprint map, and the amplitude and change law of the myoelectric signal also do not conform to the normal pattern, the abnormal judgment mechanism is triggered. Once an abnormality is detected, a warning signal is immediately output, and this signal can be transmitted in various ways, such as sending a text message to notify the guardian or popping up a warning window on a smart device, so that relevant personnel can take measures in time to ensure the walking safety of the elderly.

[0051] In a possible implementation manner, the matching prediction sub-module further includes:

[0052] A temporal gait deviation establishment sub-module, configured to establish the temporal gait deviation between the electromyogram-gait fusion fingerprint map and the real-time collected data.

[0053] A trend score calculation sub-module, configured to calculate a trend score by using the temporal gait deviation, where the trend score is used to measure the development direction and speed of the gait deviation.

[0054] A static comparison result acquisition sub-module, configured to acquire the static comparison result between the electromyogram-gait fusion fingerprint map and the real-time collected data.

[0055] An early warning signal output sub-module, configured to complete the abnormal gait trend prediction according to the static comparison result and the trend score, and output an early warning signal.

[0056] Specifically, taking the constructed electromyogram-gait fusion fingerprint map as a standard reference, this map covers the comprehensive characteristics and time series relationship of the electromyogram signal and gait parameters under normal gait. For the real-time collected electromyogram signal and gait data, the dynamic time warping (DTW) algorithm is used for processing. The DTW algorithm can find the optimal matching path between time series with different time lengths. By comparing the time sequence and interval of the characteristic points of the corresponding electromyogram signals in the real-time data and the map (such as the start and peak moments of muscle activation) and the key gait parameters (such as step length change, start and end points of the gait cycle), the time difference between the two can be accurately calculated. For example, if a certain muscle in the map is activated in a specific time sequence, and the activation sequence or time interval of this muscle in the real-time collected data changes, the DTW algorithm can convert this change into a quantified value, thereby establishing the temporal gait deviation between the electromyogram-gait fusion fingerprint map and the real-time collected data, providing a key basis for judging whether the gait is abnormal in terms of the time dimension difference.

[0057] Preprocess the temporal gait deviation data to remove outliers and noise interference, ensuring the accuracy and stability of the data. Then, segment the data using the sliding window method. For example, set a sliding window with a fixed duration, move the window one time step each time, and calculate the mean and standard deviation of the data within the window. By observing the change in the mean of adjacent window data, judge the development direction of gait deviation. If the mean of the latter window is greater than that of the previous window, it indicates that the gait deviation is developing in a more abnormal direction; otherwise, it is improving in the normal direction. For calculating the speed of gait deviation, calculate the difference between the means of adjacent windows and divide it by the time step of window movement to obtain the deviation change per unit time, thereby quantifying the speed of gait deviation. Finally, comprehensively quantify the development direction and speed, set weights to perform weighted summation of the quantified values of the development direction and the speed, and obtain a numerical value as the trend score. This score can intuitively reflect the development trend of gait deviation.

[0058] Extract the static feature data representing the normal state from the electromyogram-gait fusion fingerprint spectrum, including the amplitude distribution of the electromyogram signal at specific moments, the proportion of frequency components, and key gait parameters such as the mean of step length, the stable interval of step frequency, and the quantification index of left-right gait symmetry. At the same time, extract the corresponding static features from the real-time collected data. Then, for the amplitude of the electromyogram signal, use the mean square error algorithm to calculate the overall size of the amplitude difference between the spectrum and the real-time data; for the frequency components, use the cosine similarity to measure the similarity of their spectral distributions. In terms of key gait parameters, calculate the absolute difference between the real-time step length and the mean step length in the spectrum to quantify the deviation degree of the step length; judge whether the step frequency is abnormal by comparing the relationship between the real-time step frequency and the stable interval of step frequency in the spectrum; use the symmetry measurement algorithm to evaluate the difference between the real-time left-right gait symmetry and the spectrum standard. Finally, integrate and process the comparison results of each item, assign corresponding weights to different comparison indexes according to their importance in judging gait abnormality, and then sum the weighted difference values of each item to obtain a comprehensive static comparison result value. The larger this value is, the greater the static difference between the real-time collected data and the electromyogram-gait fusion fingerprint spectrum, indicating a higher possibility of gait abnormality.

[0059] Receive the static comparison result provided by the static comparison result acquisition submodule and the trend score generated by the trend score calculation submodule. When predicting gait trend abnormality, the static comparison result and trend score will be compared with the preset threshold first. For the static comparison result, if its value exceeds the static abnormality threshold, it means that the static difference between the real-time collected data and the electromyography-gait fusion fingerprint map is large, and there may be a risk of gait abnormality; and if the trend score is greater than the positive abnormal speed threshold, it means that the gait is rapidly developing in a more abnormal direction, or the trend score is less than the negative abnormal speed threshold, indicating that although there is an improvement trend, the degree of abnormality cannot be ignored. When these two conditions are met at the same time or when a single condition that is more serious is met, the early warning signal output submodule will determine that there is a gait trend abnormality. Once an abnormality judgment is made, an early warning signal will be output immediately. The early warning signal can be transmitted in a variety of ways, such as sending a text message to notify the guardian that the elderly may have a risk of gait abnormality, and attaching a brief abnormal information; popping up a striking warning window on the smart device to remind nearby caregivers to pay attention to the elderly's condition, and triggering a corresponding voice alarm to attract the attention of people around in time, so that measures can be taken to ensure the safety of the elderly's walking and avoid accidents.

[0060] In a possible implementation, the system further includes:

[0061] The auxiliary wearing module is used to activate the auxiliary wearing module and execute the airbag inflation response of the auxiliary wearing module when the warning signal is higher than a preset threshold.

[0062] Specifically, the auxiliary wear module plays a role in safety protection. When the warning signal output submodule determines that the gait trend is abnormal based on the static comparison results and trend scores, and the intensity of the warning signal is higher than the preset threshold, it means that the elderly are very likely to fall or have accidents. At this time, the auxiliary wear module will be activated immediately. This module is mainly composed of an anti-fall vest airbag vest device. Once activated, its built-in rapid inflation system will respond quickly and inflate the airbag. Its head airbag will expand rapidly, which can not only protect the elderly's head, but also reduce the weight through optimized design, without adding too much burden while protecting; the face airbag will also be inflated synchronously, increasing the protection of the face and reducing the risk of facial injuries; the waist airbag adopts a hidden design, which can provide waist support after inflation to reduce waist injuries; the hip airbag will achieve three-dimensional protection and provide cushioning for the elderly's hips. The entire inflation process is fast and efficient, and can be completed in a very short time, providing all-round protection for the elderly at the moment of possible fall, minimizing the degree of injury to the elderly due to falls, and ensuring the safety of the elderly.

[0063] In a possible implementation, the system further includes:

[0064] The synchronous early warning reporting module is used to configure an early warning response plan according to the early warning level of the early warning signal, send the early warning signal to associated devices by using the early warning response plan, and synchronously send the location information of the user.

[0065] Specifically, after the early warning signal output sub-module outputs an early warning signal, the synchronous early warning reporting module will first obtain the early warning level of this early warning signal. The early warning level is generally divided according to the comprehensive situation of the static comparison result and the trend score. For example, it can be divided into three levels: mild, moderate, and severe. For different early warning levels, this module will configure corresponding early warning response plans. For mild early warnings, only send a vibration reminder to the associated smart bracelet or mobile phone application, and at the same time display a simple prompt message on the screen, informing the guardian that the elderly may have a slight abnormal gait trend, but there is no need to be overly nervous for the time being, and at the same time attach the real-time location information of the elderly to facilitate the guardian to keep track of the elderly's dynamics at any time. When a moderate early warning occurs, in addition to sending vibration and prompt messages to the associated devices, it will also automatically call the guardian's phone and inform the elderly of the abnormal gait situation and the current location in the form of voice broadcast, reminding the guardian to pay close attention to the elderly's condition and contact the elderly to confirm their safety if necessary. In the case of a severe early warning, immediately send a strong sound and light alarm to all associated devices, automatically call the emergency contact phone and the emergency phone at the same time, and send the accurate location information of the elderly to the emergency personnel and the guardian in real time to ensure that when the elderly have a serious abnormal gait and may face dangerous situations such as falling, the rescue personnel can quickly arrive at the scene for rescue. By configuring different early warning response plans according to the early warning level in this way, the synchronous early warning reporting module can ensure that in different danger levels, the early warning signal and location information can be sent to relevant personnel in a timely manner in a suitable way, maximizing the safety of the elderly.

[0066] In a possible implementation manner, the system further includes:

[0067] The feedback correction module is used to obtain the actual user state mapped to the early warning signal, verify the early warning trust of the early warning signal according to the actual user state, lock the key early warning features, and perform prediction optimization management of the myoelectric-gait fusion fingerprint map according to the early warning trust verification result and the key early warning features.

[0068] Specifically, after the warning signal output sub-module issues a warning signal, the feedback correction module will first obtain the actual user status corresponding to the warning signal, which is achieved through various means, such as data interaction with other sensor devices carried by the user (such as wearable cameras, motion sensors, etc.), or obtaining it through the user's active feedback (such as pressing the confirmation button on the smart device). The actual user status includes, but is not limited to, information such as whether the user has actually fallen, whether the gait is truly abnormal, and whether the body is uncomfortable. Then, based on the obtained actual user status, the warning trust verification of the warning signal is performed. If the actual user status shows that the user is indeed in an abnormal state, such as having fallen, then the credibility of the warning signal will increase; conversely, if the actual user status shows that the user is normal while the system issues a warning signal, then the credibility of the warning signal will decrease. At the same time, the module will lock the key warning features, which may be specific abnormal patterns shown in the electromyogram-gait fusion fingerprint map when the warning signal is issued, such as abnormal fluctuations in the electromyogram signal within a certain period of time, sudden changes in gait parameters, etc. Finally, according to the warning trust verification result and the locked key warning features, predictive optimization management of the electromyogram-gait fusion fingerprint map is performed. If the credibility of the warning signal is high, further analyze the key warning features and incorporate them into the update of the electromyogram-gait fusion fingerprint map to improve the prediction accuracy of subsequent similar abnormal situations; if the credibility of the warning signal is low, check the reasons for the false alarm, adjust and optimize the relevant features in the map, and remove the interference factors that may cause false alarms, thereby continuously enhancing the reliability of the entire system for monitoring and warning of abnormal elderly gaits and reducing the occurrence of false alarms and missed alarms.

[0069] In a possible implementation manner, the system further includes:

[0070] A learning module, configured to perform incremental learning of the user's gait after the user enables the learning module, establish an incremental database, and compensate the electromyogram-gait fusion fingerprint map according to the incremental database.

[0071] Specifically, the learning module collects the user's gait and electromyography data by means of the collaborative work of multiple sensors. These sensors include electromyography sensors distributed at key parts of the user's body (such as the legs and waist) for accurately capturing the electrical signals generated by muscle activities, and pressure sensors and accelerometers installed in the shoes to obtain gait parameters such as step length, step frequency, speed, and acceleration. Then, the collected raw data is preprocessed. The high-frequency noise and low-frequency drift in the electromyography signals are removed through a band-pass filtering algorithm, and outliers in the data of the pressure sensors and accelerometers are eliminated using the threshold method. Subsequently, feature extraction techniques are employed to extract time-domain features such as root mean square value and average absolute value, and frequency-domain features such as power spectral density from the processed electromyography signals; key parameters such as step length, step frequency, and support phase time are extracted from the gait data. The extracted features are compared and analyzed with the existing electromyography-gait fusion fingerprint maps. Similarity matching algorithms in machine learning, such as Euclidean distance and cosine similarity, are used to determine the similarity degree between the new data and the existing patterns in the maps. If the similarity is lower than the set threshold, it is considered that a new gait or electromyography pattern has emerged. The data of the newly emerged pattern is stored in the incremental database, and at the same time, the incremental database is classified and managed, for example, divided according to different activity scenarios (such as walking, running, going up and down stairs) and the degree of abnormality. When compensating the electromyography-gait fusion fingerprint maps, a weighted update method is adopted. Higher weights are assigned to the new features that appear frequently and stably in the incremental database; lower weights are assigned to the features that appear occasionally. Through weighted summation, the new features are incorporated into the original maps to complete the update and compensation of the electromyography-gait fusion fingerprint maps, thereby improving the accuracy and adaptability of the system for monitoring the user's gait abnormalities.

[0072] It should be noted that the above order of the embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. And the above specific embodiments of this specification have been described. Additionally, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0073] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application shall be included within the protection scope of the present application.

[0074] This specification and the drawings are only exemplary descriptions of the present application and are considered to have covered any and all modifications, variations, combinations, or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications.

Claims

1. An abnormal gait monitoring system for the elderly based on surface electromyographic signals, characterized in that: The system comprises: A configuration module is used to obtain a set of foot bearing devices of a user, classify and identify gait influences on the set of foot bearing devices, and construct three sets of foot bearing device test scenarios; An inertial measurement module is used to collect gait data of the user in three groups of foot-bearing device test scenarios and establish key gait parameters, including step length, step frequency, gait cycle, and gait symmetry; The electromyographic data acquisition module is used to synchronously acquire electromyographic signals and establish an electromyographic test data set. The electromyographic data acquisition module acquires signals from the quadriceps, gastrocnemius, and gluteus maximus muscles. A labeling module is used to establish a gait cycle window based on the key gait parameters, and to establish an electromyography-gait fusion fingerprint map after dividing the key gait parameters and the electromyography test data set into gait segments using the gait cycle window; The early warning module is used to read the real-time collected data of the electromyography data collection module, use the electromyography-gait fusion fingerprint map to predict the abnormal gait trend based on the real-time collected data, and output an early warning signal.

2. The elderly gait abnormality monitoring system based on body surface electromyographic signals as claimed in claim 1, characterized in that: The marking module comprises: A grouping and labeling submodule is used to segment and identify the gait segments of the three groups of foot-bearing device test scenarios and to create a sequential association of the segmented gait segments; The signal processing submodule is used to extract electromyographic features of the segmented gait segments and establish electromyographic feature extraction results, wherein the electromyographic feature extraction results include time domain features, frequency domain features, and nonlinear features; The feature fusion submodule is used to perform feature splicing on the electromyographic feature extraction results and gait key parameters corresponding to the same segmented gait segments to form a unified feature vector; The fingerprint map construction submodule is used to establish the electromyography-gait fusion fingerprint map according to the unified feature vector and the sequential association.

3. The elderly gait abnormality monitoring system based on body surface electromyographic signals as claimed in claim 2, characterized in that: The feature fusion submodule includes: The constraint establishment submodule is used to establish muscle activation time delay constraints, EMG-cadence synergy constraints, and left and right muscle group activation symmetry constraints; The splicing submodule is used to perform splicing of the electromyographic feature extraction results and the key gait parameters according to the muscle activation time delay constraint, the electromyographic-step frequency synergy constraint, and the left and right muscle group activation symmetry constraint.

4. The elderly gait abnormality monitoring system based on body surface electromyographic signals as claimed in claim 2, characterized in that: The fingerprint map construction submodule includes: A granularity segmentation submodule is used to sort the three groups of foot bearing device test scenes according to the sequential association, perform bearing thickness granularity segmentation, and establish a granularity segmentation result; The correction prediction submodule is used to use the sequentially associated unified feature vector as the boundary node, perform the EMG-gait fusion prediction under the granular segmentation result, and establish the EMG-gait fusion fingerprint map according to the fusion prediction result.

5. The elderly gait abnormality monitoring system based on body surface electromyographic signals as claimed in claim 4, characterized in that: The early warning module comprises: The matching initialization submodule is used to call the correction prediction submodule to perform matching identification of the current bearing thickness after receiving the real-time acquisition data, and then initialize the electromyography-gait fusion fingerprint map; The matching prediction submodule is used to predict the abnormal gait trend of real-time collected data based on the initialized electromyography-gait fusion fingerprint map and output a warning signal.

6. The elderly gait abnormality monitoring system based on body surface electromyographic signals as claimed in claim 5, characterized in that: The matching prediction submodule also includes: A time-series gait deviation establishment submodule is used to establish the time-series gait deviation between the electromyography-gait fusion fingerprint map and the real-time collected data; A trend score calculation submodule, used to calculate a trend score using the time-series gait deviation, wherein the trend score is used to measure the development direction and speed of the gait deviation; The static comparison result acquisition submodule is used to obtain the static comparison result between the electromyography-gait fusion fingerprint map and the real-time collected data; The warning signal output submodule is used to complete the gait trend abnormality prediction according to the static comparison result and the trend score, and output a warning signal.

7. The elderly gait abnormality monitoring system based on body surface electromyographic signals as claimed in claim 6, characterized in that: The system further comprises: The auxiliary wearing module is used to activate the auxiliary wearing module and execute the airbag inflation response of the auxiliary wearing module when the warning signal is higher than a preset threshold.

8. The elderly gait abnormality monitoring system based on body surface electromyographic signals as claimed in claim 1, characterized in that: The system further comprises: The synchronous early warning output module is used to configure an early warning response plan according to the early warning level of the early warning signal, use the early warning response plan to send the early warning signal to the associated device, and synchronously send the user's location information.

9. The elderly gait abnormality monitoring system based on body surface electromyographic signals as claimed in claim 1, characterized in that: The system further comprises: A feedback correction module is used to obtain the actual user state mapped to the warning signal, verify the warning trust of the warning signal according to the actual user state, lock the key warning features, and perform predictive optimization management of the electromyography-gait fusion fingerprint map according to the warning trust verification results and the key warning features.

10. The elderly gait abnormality monitoring system based on body surface electromyographic signals according to claim 1, the system further comprising: The learning module is used to perform incremental gait learning of the user after the user activates the learning module, establish an incremental database, and compensate the electromyography-gait fusion fingerprint map according to the incremental database.

Images