Intelligent auxiliary method and system for nursing rehabilitation

Abstract

The invention discloses an intelligent assistance method and system for nursing rehabilitation, and relates to the technical field of medical intelligent assistance. The method comprises the following steps: acquiring residual neural signals of an amputation part, and performing regionalized decoding on different channel signals in combination with a distributed processing unit to generate an action intention signal matrix; on the basis of the action intention signal matrix, synchronously triggering the activation of the corresponding nerve region and the residual limb muscle to form a nerve-muscle synchronous action data stream; performing comparative analysis on the monitoring data and the nerve-muscle synchronous action data flow, and extracting a dynamic error signal; nerve stimulation parameters and muscle stimulation parameters are optimized through the multi-modal data fusion model; a complete nerve-muscle cooperative control scheme is generated, and the residual limb or the artificial limb of the patient is driven to complete the target action in real time. According to the invention, the problems of poor neural signal decoding accuracy, mismatching of nerve-muscle responses and the like in the prior art are effectively solved.

Description

Technical Field

[0001] The present invention relates to the field of medical intelligent assistance technology, and in particular to an intelligent assistance method and system for nursing rehabilitation. Background Art

[0002] With the continuous development of rehabilitation medicine and bioengineering technology, the application of neuromuscular interface technology in the field of medical treatment and nursing rehabilitation has made significant progress. By connecting the human nervous system with external devices, this technology can realize the acquisition and decoding of neural signals, providing a way to restore neural function for patients with nerve damage. In the rehabilitation of amputees, neural interface technology is used to obtain neural signals from the residual limbs, and use signal decoding technology to reconstruct the intention of movement, so as to achieve movement recovery through muscle stimulation equipment. In addition, the advent of multi-channel neural interface devices in recent years has significantly improved the spatial resolution of neural signal acquisition, and combined with distributed computing technology, complex neural signals can be efficiently processed. These technologies have brought new opportunities to the field of neurorehabilitation. However, existing neural interface technologies still face multiple challenges in practical applications, including insufficient signal acquisition quality, limited decoding accuracy, and instability of neuromuscular responses, which seriously restrict the functional realization of rehabilitation equipment and the actual benefits of patients.

[0003] Current technical methods mainly focus on the collection and analysis of single nerve or muscle signals, but due to the complexity of the human nervous system and muscles, this one-dimensional signal processing method cannot fully reflect the dynamic synergistic relationship of the nerve-muscle system. For example, traditional nerve stimulation devices usually rely on fixed stimulation parameters and fail to adjust the stimulation scheme according to the patient's real-time action needs, resulting in a large deviation between nerve activation and muscle response. On the other hand, existing neural signal decoding methods mostly use static models, ignoring the dynamic characteristics of neural signals in time and space, and the decoding accuracy and action intention reconstruction effect are poor. In addition, the muscle stimulation device lacks synchronous regulation with the neural response signal during the execution of the action, which further aggravates the deviation of the target action. These technical deficiencies have led to limited rehabilitation effects and it is difficult to meet the patient's refined action needs, especially in high-precision motion control and real-time feedback regulation. There are obvious defects. Summary of the invention

[0004] In view of the problems existing in the above-mentioned background technology, the present invention proposes an intelligent auxiliary method and system for nursing rehabilitation.

[0005] Therefore, the problem to be solved by the present invention is how to solve the problems of poor accuracy in neural signal decoding and mismatch between nerve and muscle response in the prior art.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: In the first aspect, the present invention provides an intelligent auxiliary method for nursing rehabilitation, which includes, S1: obtaining residual nerve signals at the amputation site through a multi-channel implantable or non-implantable neural interface device, combining a distributed processing unit to perform regional decoding on signals of different channels, and generating an action intention signal matrix; S2: based on the action intention signal matrix, through the coordinated work of a neural stimulation device and a muscle stimulation device, synchronously triggering the activation of corresponding nerve areas and residual limb muscles, recording neural response data and muscle dynamic output data, and forming a neural-muscle synchronous action data stream; S3: real-time monitoring of muscle mechanical parameters and dynamic bioelectric signals, comparing and analyzing the monitoring data with the neural-muscle synchronous action data stream, and extracting dynamic error signals; S4: using the dynamic error signal, optimizing neural stimulation parameters and muscle stimulation parameters through a multimodal data fusion model; S5: generating a complete neural-muscle collaborative control scheme, driving the patient's residual limb or prosthesis to complete the target action in real time, and monitoring the matching of the action execution effect with the target requirements through sensors.

[0007] As a preferred embodiment of the intelligent auxiliary method for nursing rehabilitation described in the present invention, S1 includes the following operation steps: according to the anatomical characteristics of the amputation site and the residual nerve distribution, different signal acquisition channels are selected through a multi-channel implantable or non-implantable neural interface device, and regional division is performed according to the spatial relationship of the signal acquisition channels in the nerve distribution to generate regional identification data; frequency domain and time domain feature extraction is performed, and the regional feature vector matrix is ​​formed in combination with the regional identification data ; Based on the regional feature vector matrix , calculate the interaction pattern between regions; combine the regional feature vector matrix and the regional interaction pattern data, calculate the action intention information of the multi-channel neural signal through the distributed decoding algorithm, and generate the action intention signal matrix.

[0008] As a preferred solution of the intelligent auxiliary method for nursing and rehabilitation described in the present invention, the interaction mode between the regions includes correlation, coherence and directionality analysis.

[0009] As a preferred embodiment of the intelligent auxiliary method for nursing rehabilitation described in the present invention, the generation of the action intention signal matrix includes: using variational inference to decode regional interaction pattern data and regional feature vector matrix The decoded action intention signal is represented by a matrix to obtain the action intention signal matrix .

[0010] As a preferred solution of the intelligent auxiliary method for nursing rehabilitation described in the present invention, wherein: S2 includes the following operating steps: based on the action intention signal matrix, combined with the patient's personalized nerve conduction characteristics and muscle response characteristics, nerve stimulation parameters and muscle stimulation parameters are generated through parameter mapping to form a stimulation parameter data set; the stimulation parameter data set is distributed to the multi-channel output end of the nerve stimulation device, the regional identification of the stimulation point is determined through the spatial positioning algorithm, the nerve area corresponding to the action intention signal matrix is ​​activated, and nerve activation data is generated; based on the nerve activation data, an electrical stimulation signal synchronized with nerve stimulation is applied to the target muscle group through the muscle stimulation device to induce muscle action response and generate muscle dynamic output data; the response signal of the nerve area and the dynamic output data of the target muscle are collected through the embedded sensor, and the dynamic output data are fused to form a real-time response data stream of nerve-muscle. ; The real-time response data stream and action intention signal matrix Synchronous correlation analysis is performed, and signal delays are eliminated through a time correction algorithm to generate a neuromuscular synchronous action data stream containing neural activation data and muscle dynamic output data.

[0011] As a preferred solution of the intelligent auxiliary method for nursing rehabilitation described in the present invention, the time correction algorithm adopts dynamic time warping and adaptive window function, including the following steps: responding to the real-time data stream and action intention signal matrix After normalization, we get and ; Extract the characteristic points of the signal respectively and generate characteristic time series and ;structure and The Euclidean distance matrix ; Calculate the cumulative cost matrix C recursively; Find the optimal time alignment path P; by estimating and The global trend of constructing an adaptive window function , constrain the search area of ​​the alignment path P; Align according to the optimal path P and , and low-pass filter the synchronized data stream to obtain the time-corrected neuromuscular synchronized action data stream .

[0012] As a preferred solution of the intelligent auxiliary method for nursing rehabilitation described in the present invention, wherein: the comparison and analysis of the monitoring data with the neuromuscular synchronous action data stream to extract the dynamic error signal includes: Synchronous neuromuscular movement data flow Conduct comparative analysis and extract dynamic error signals; For error signal Perform frequency domain and time domain analysis to extract dynamic error characteristics; calculate error signals The spatiotemporal correlation between the components is used to extract the synergistic characteristics, and the analysis results are fused to generate the final dynamic error signal. .

[0013] In a second aspect, the present invention provides an intelligent assistance system for nursing rehabilitation, comprising: A neural signal acquisition and decoding module is used to obtain residual neural signals at the amputation site through a multi-channel implantable or non-implantable neural interface device, and regionally decode signals of different channels in combination with a distributed processing unit to generate an action intention signal matrix; a neural-muscle activation and recording module is used to synchronously trigger the activation of corresponding neural regions and residual limb muscles based on the action intention signal matrix through the coordinated work of a neural stimulation device and a muscle stimulation device, record neural response data and muscle dynamic output data, and form a neural-muscle synchronous action data stream; a real-time monitoring and dynamic error analysis module is used to monitor muscle mechanical parameters and dynamic bioelectric signals in real time, compare and analyze the monitoring data with the neural-muscle synchronous action data stream, and extract dynamic error signals; a parameter optimization module is used to optimize neural stimulation parameters and muscle stimulation parameters through a multimodal data fusion model using the dynamic error signal; a neural-muscle synergistic control module is used to generate a complete neural-muscle synergistic control solution, drive the patient's residual limb or prosthesis to complete the target action in real time, and monitor the matching of the action execution effect with the target demand through sensors.

[0014] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program instructions are executed by the processor, the steps of the intelligent assistance method for nursing rehabilitation as described in the first aspect of the present invention are implemented.

[0015] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program instructions are executed by a processor, the steps of the intelligent assistance method for nursing and rehabilitation as described in the first aspect of the present invention are implemented.

[0016] The beneficial effects of the present invention are as follows: the present invention collects residual neural signals at the amputation site through a multi-channel implantable or non-implantable neural interface device, and realizes regionalized signal decoding in combination with a distributed processing unit; based on the decoded action intention signal matrix, the coordinated work of the neural stimulation device and the muscle stimulation device is utilized to realize synchronous neural-muscle activation; the stimulation parameters are optimized through a multimodal data fusion model, and a real-time driven neural-muscle collaborative control scheme is constructed, which effectively solves the problems of poor neural signal decoding accuracy and neural-muscle response mismatch in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0018] Figure 1 Flowchart of intelligent assistance method for nursing rehabilitation.

[0019] Figure 2 This is a structural diagram of the intelligent assistance system for nursing rehabilitation. DETAILED DESCRIPTION

[0020] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0021] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0022] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0023] Example 1: Reference Figure 1 , which is the first embodiment of the present invention, and which provides an intelligent auxiliary method for nursing rehabilitation, such as Figure 1 As shown, including: S1: Obtain the residual neural signals at the amputation site through a multi-channel implantable or non-implantable neural interface device, combine with a distributed processing unit to perform regional decoding on the signals of different channels, and generate a matrix of action intention signals.

[0024] S1.1: According to the anatomical characteristics of the amputation site and the residual nerve distribution, different signal acquisition channels are selected through a multi-channel implantable or non-implantable neural interface device, and regional division is performed according to the spatial relationship of the signal acquisition channels in the nerve distribution to generate regional identification data.

[0025] The regional divisions include: Through clustering algorithms (such as density peak clustering or community partitioning based on graph structure), channel signals are divided into multiple regions according to spatial position and signal characteristics.

[0026] Furthermore, the neural signals collected through the multi-channels are processed in a distributed manner, and the signals in each region are pre-processed, including band-pass filtering and noise suppression based on an adaptive algorithm, to generate denoised signal data.

[0027] S1.2: Extract frequency domain and time domain features of the denoised signal data and combine the regional identification data to form a regional feature vector matrix .

[0028] Among them, it includes combining the features of each regional signal into a regional feature vector matrix, in which each row represents a region and each column represents different features; the features include signal amplitude, frequency distribution and phase change characteristics.

[0029] S1.3: Based on the regional feature vector matrix, the interaction pattern between regions is calculated, including correlation, coherence and directionality analysis, to generate regional interaction pattern data.

[0030] Preferably, the area is calculated by mutual information and Signal correlation between: ; in, For Region and The mutual information between For Region The signal value of For Region The signal value of For Region and The probability density of the joint distribution of signals in , For Region The marginal probability density of the signal in is, For Region The marginal probability density of the signal in .

[0031] Analyze the coherence between regions in the frequency domain: ; in, It is a region and The cross power spectral density is For Region and In frequency The coherence value on reflects the synchronization of two regions at a specific frequency. and Respectively, the regions and In frequency The autopower spectral density on .

[0032] Calculate the directionality of the signal based on dynamic causal modeling (DCM): ; in, For Region and The signal directionality index, the larger the value, the more obvious the impact.

[0033] S1.4: Combine the regional feature vector matrix and the regional interaction pattern data, calculate the action intention information of the multi-channel neural signal through a distributed decoding algorithm, and generate the action intention signal matrix.

[0034] Specifically, variational inference is used to decode regional interaction pattern data and the regional eigenvector matrix ; The decoded action intention signal is represented by a matrix, and the action intention signal matrix is ​​obtained .

[0035] It should be noted that the present invention can achieve high-precision generation of action intention signals by mining signal characteristics such as wavelet decomposition, mutual information calculation, dynamic causal modeling, etc., and introducing variational inference in the decoding algorithm.

[0036] It can be seen that the present invention improves the decoding accuracy of residual neural signals by introducing high-precision algorithms such as clustering algorithms, variational inference and dynamic causal modeling, effectively reduces signal noise interference and inter-regional signal aliasing problems, and enables the generated action intention signal matrix to truly reflect the patient's movement intention, thereby achieving the effect of improving the accuracy of action intention generation and the efficiency of neural signal utilization.

[0037] S2: Based on the action intention signal matrix, through the coordinated work of the nerve stimulation device and the muscle stimulation device, the activation of the corresponding nerve area and the residual limb muscle is synchronously triggered, the nerve response data and the muscle dynamic output data are recorded, and a nerve-muscle synchronous action data stream is formed.

[0038] S2.1: Based on the action intention signal matrix, combined with the patient's personalized nerve conduction characteristics and muscle response characteristics, nerve stimulation parameters and muscle stimulation parameters are generated through parameter mapping, including stimulation frequency, amplitude and pulse width, to form a stimulation parameter data set.

[0039] S2.2: Distribute the stimulation parameter data set to the multi-channel output terminal of the neural stimulation device, determine the regional identification of the stimulation point through the spatial positioning algorithm, activate the neural area corresponding to the action intention signal matrix, and generate neural activation data, wherein the neural activation data is calculated as follows: ; in, is the neural activation data, indicating the activated neural area and its response strength, is the total number of channels of the neurostimulation device, For stimulation equipment The output intensity of each channel, For the spatial positioning algorithm, calculate the Regions In the stimulation parameters The activation level below.

[0040] It should be noted that the spatial localization algorithm takes into account the three-dimensional anatomical structure of the nerve area and uses a high-dimensional interpolation function to dynamically adjust the localization accuracy of the stimulation point.

[0041] S2.3: Based on the neural activation data, an electrical stimulation signal synchronized with the neural stimulation is applied to the target muscle group through the muscle stimulation device to induce a muscle action response and generate muscle dynamic output data.

[0042] S2.4: Collect the response signals of the nerve area and the dynamic output data of the target muscle through embedded sensors, including muscle contraction strength, displacement and electrophysiological signals, and fuse the dynamic output data to form a real-time response data stream of nerve-muscle .

[0043] S2.5: The real-time response data stream and action intention signal matrix Synchronous correlation analysis is performed, and signal delays are eliminated through a time correction algorithm to generate a neuromuscular synchronous action data stream containing neural activation data and muscle dynamic output data.

[0044] It should be noted that the time correction algorithm uses dynamic time warping (DTW) and adaptive window function to enhance signal alignment accuracy and reduce the impact of delay on subsequent action control. The specific operations include: First, respond to data streams in real time and action intention signal matrix After normalization, we get and , ensure that their numerical ranges are consistent for subsequent alignment calculations; Extract the characteristic points of the signal (such as peak points, key turning points) respectively to generate characteristic time series and ; After normalization, construct and The Euclidean distance matrix ; The cumulative cost matrix C is calculated recursively: ; in, From the starting point to the point The minimum cumulative alignment cost of and They are and The time index of Further, from Backtrack to the end point and find the optimal time-aligned path P.

[0045] Preferably, in order to avoid excessive computational redundancy in DTW in long time series, an adaptive window function is used to limit the search range of DTW: By estimating and The global trend of constructing an adaptive window function , constrain the search area of ​​the alignment path P: ; in, , is an adjustable parameter, represents the local standard deviation of the time series, for In the The local signal value at a time point.

[0046] Preferably, when calculating the cumulative cost matrix C, only the The position is calculated to improve efficiency.

[0047] Further, according to the optimal path P, and , and low-pass filter the synchronized data stream to eliminate the high-frequency noise caused by alignment, and obtain the time-corrected neuromuscular synchronized action data stream .

[0048] It can be seen that the present invention dynamically adjusts the accuracy of the stimulation point through a spatial positioning algorithm, and aligns the nerve and muscle response data in combination with a time correction algorithm, which significantly enhances the activation effect of the stimulation device on the target area, reduces response delays and misactivation of non-target areas, and ultimately achieves more efficient neuromuscular coordinated control of the patient's residual limbs or prostheses, thereby improving the rehabilitation effect.

[0049] S3: Through embedded high-precision force sensors, surface electromyography (sEMG) sensors and muscle displacement sensors, muscle mechanical parameters (such as stress, displacement) and dynamic bioelectric signals are monitored in real time, and the monitoring data are compared and analyzed with the neuromuscular synchronous action data stream to extract dynamic error signals.

[0050] First, embedded high-precision force sensors, surface electromyography (sEMG) sensors, and muscle displacement sensors are arranged in the target muscle group area, and the neuromuscular synchronous action data stream generated in step S2 is used to generate the , calibrate the measurement range, sensitivity and time synchronization parameters of each sensor to form the configuration data set of the monitoring node.

[0051] Through the configuration data set, muscle mechanical parameters (including stress and displacement) and surface electromyographic signal data are collected in real time during the execution of the action, a dynamic monitoring data stream of the muscle is generated, and the dynamic monitoring data stream is preliminarily filtered and noise suppressed.

[0052] Optionally, the sensor data is aligned with the time axis of the neuromuscular synchronous action data stream and the time deviation is corrected using a dynamic time warping (DTW) algorithm.

[0053] Furthermore, real-time monitoring data Synchronous neuromuscular movement data flow Perform comparative analysis and extract dynamic error signals: ; in, Including mechanical stress error, electrophysiological signal error and displacement error.

[0054] Furthermore, for the error signal Perform frequency domain and time domain analysis to extract dynamic error features, including: time domain features, such as mean, standard deviation and peak value; frequency domain features, such as power spectral density (PSD), main frequency components, etc.

[0055] Preferably, the error signal is calculated Spatiotemporal correlation between components, extracting synergistic characteristics: ; in, for and The spatiotemporal correlation of and represents the different components of the error signal, and They are the first Quantity and Quantity The time mean, is the length of the time series.

[0056] The analysis results are subjected to feature fusion to generate the final dynamic error signal .

[0057] Optionally, the fusion method may adopt an adaptive weight model or a correlation-based weight allocation model, etc.

[0058] S4: Utilizing the dynamic error signal, the neural stimulation parameters (including pulse frequency, pulse width and amplitude) and muscle stimulation parameters are optimized through a multimodal data fusion model to ensure that the neuromuscular response matches the patient's target movement requirements.

[0059] S4.1: Based on the dynamic error signal generated in step S3 and the neural response data recorded in step S2, feature mapping is used to extract correlation features, including the mapping relationship between the dynamic error index and the neural activation intensity, to generate a feature mapping matrix.

[0060] Specifically, the mutual information between the dynamic error signal and the neural response signal is calculated; The mutual information calculation results are integrated according to the time dimension to generate a feature mapping matrix.

[0061] S4.2: Combine the feature mapping matrix and muscle dynamic output data, input into the multimodal data fusion model, optimize the model parameters through iterative training, so that it can accurately predict the effect of stimulation parameters on neuromuscular response, and generate a model optimization weight matrix.

[0062] Among them, the multimodal data fusion model is constructed using a graph neural network (GNN) or a recurrent neural network (RNN), and the loss function of the model is optimized by the gradient descent method, and the optimized weight matrix is ​​finally output. .

[0063] S4.3: Optimize the weight matrix based on the model, input the patient's target action requirement data (such as expected stress, displacement and electromyographic response characteristics), and convert the weight matrix Associated with the target requirements, an optimized neural stimulation parameter matrix and a muscle stimulation parameter matrix are generated.

[0064] S4.4: Input the optimized neural stimulation parameter matrix and muscle stimulation parameter matrix into the corresponding stimulation device, combine the real-time collected neural response data and dynamic monitoring data stream, calculate the error signal, and verify the optimization effect. If the error exceeds the threshold, dynamically adjust the stimulation parameters to ensure that the neural-muscle response matches the target action requirements.

[0065] Preferably, the stimulation parameters are dynamically adjusted through a feedback control mechanism, and the stimulation parameters are gradually optimized until the error tolerance is met.

[0066] It should be noted that this step uses dynamic error signals to reflect the deviations in the neural and muscular control process. By mining the error features and extracting the synergistic characteristics through frequency domain and time domain analysis, it helps to discover abnormal control of the neuromuscular system during the rehabilitation process.

[0067] This real-time monitoring and error extraction mechanism provides key feedback for subsequent parameter optimization, while laying a data foundation for closed-loop control and improving rehabilitation accuracy.

[0068] S5: Generate a complete neuromuscular synergistic control solution to drive the patient's residual limb or prosthesis to complete the target action in real time, and monitor the match between the action execution effect and the target requirements through sensors.

[0069] If the dynamic error signal continues to meet the set threshold, the closed-loop control process is terminated and the rehabilitation assessment report is output, completing the intelligent assistance process.

[0070] Furthermore, a rehabilitation assessment report is generated to record key data in the process of movement recovery, including movement execution effects, stimulation parameter optimization records and error change curves, and the results of rehabilitation progress analysis are output to provide a reference for subsequent rehabilitation training and parameter adjustment.

[0071] For further information, see Figure 2 This embodiment also provides an intelligent assistance system for nursing rehabilitation, including: A neural signal acquisition and decoding module is used to obtain residual neural signals at the amputation site through a multi-channel implantable or non-implantable neural interface device, and to perform regional decoding of signals from different channels in combination with a distributed processing unit to generate a matrix of action intention signals; A neuromuscular activation and recording module is used to synchronously trigger the activation of the corresponding nerve area and residual limb muscles based on the action intention signal matrix through the coordinated work of the nerve stimulation device and the muscle stimulation device, record the nerve response data and the muscle dynamic output data, and form a neuromuscular synchronous action data stream; Real-time monitoring and dynamic error analysis module, used to monitor muscle mechanical parameters and dynamic bioelectric signals in real time, compare and analyze the monitoring data with the neuromuscular synchronous action data stream, and extract dynamic error signals; A parameter optimization module, used to optimize nerve stimulation parameters and muscle stimulation parameters by using the dynamic error signal through a multimodal data fusion model; The neuromuscular synergistic control module is used to generate a complete neuromuscular synergistic control solution, drive the patient's residual limb or prosthesis to complete the target action in real time, and monitor the match between the action execution effect and the target requirements through sensors.

[0072] This embodiment also provides a computer device, which is suitable for the case of an intelligent assistance method for nursing rehabilitation, including a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the intelligent assistance method for nursing rehabilitation proposed in the above embodiment.

[0073] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0074] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the intelligent assistance method for nursing rehabilitation proposed in the above embodiment is implemented.

[0075] In summary, the present invention collects residual nerve signals at the amputation site through a multi-channel implantable or non-implantable neural interface device, and realizes signal regionalization decoding in combination with a distributed processing unit; based on the decoded action intention signal matrix, the coordinated work of the neural stimulation device and the muscle stimulation device is utilized to realize nerve-muscle synchronous activation; the stimulation parameters are optimized through a multimodal data fusion model, and a real-time driven neural-muscle collaborative control scheme is constructed, which effectively solves the problems of poor neural signal decoding accuracy and neural-muscle response mismatch in the prior art.

[0076] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An intelligent auxiliary method for nursing rehabilitation, characterized in that: include: S1: Obtain residual neural signals at the amputation site through a multi-channel implantable or non-implantable neural interface device, and perform regional decoding of signals from different channels in combination with a distributed processing unit to generate a matrix of action intention signals; S2: Based on the action intention signal matrix, the activation of the corresponding nerve area and residual limb muscle is synchronously triggered through the coordinated work of the nerve stimulation device and the muscle stimulation device, and the nerve response data and muscle dynamic output data are recorded to form a nerve-muscle synchronous action data stream; S3: Real-time monitoring of muscle mechanical parameters and dynamic bioelectric signals, comparative analysis of the monitoring data with the neuromuscular synchronous action data stream, and extraction of dynamic error signals; S4: optimizing nerve stimulation parameters and muscle stimulation parameters through a multimodal data fusion model using the dynamic error signal; S5: Generate a complete neuromuscular synergistic control solution to drive the patient's residual limb or prosthesis to complete the target action in real time, and monitor the match between the action execution effect and the target requirements through sensors.

2. The intelligent assistance method for nursing rehabilitation according to claim 1, characterized in that: The S1 comprises the following steps: According to the anatomical characteristics of the amputation site and the residual nerve distribution, different signal acquisition channels are selected through a multi-channel implantable or non-implantable neural interface device, and regional division is performed according to the spatial relationship of the signal acquisition channels in the nerve distribution to generate regional identification data; Perform frequency domain and time domain feature extraction, and combine regional identification data to form a regional feature vector matrix ; Based on the regional eigenvector matrix , calculate the interaction patterns between regions; Combining the regional feature vector matrix and regional interaction pattern data, the action intention information of multi-channel neural signals is calculated through a distributed decoding algorithm to generate an action intention signal matrix.

3. The intelligent assistance method for nursing rehabilitation according to claim 2, characterized in that: The interaction pattern between the regions includes correlation, coherence and directionality analysis.

4. The intelligent auxiliary method for nursing rehabilitation according to claim 3, characterized in that: The generation of the action intention signal matrix includes: Decoding regional interaction pattern data and regional feature vector matrices using variational inference ; The decoded action intention signal is represented by a matrix, and the action intention signal matrix is ​​obtained. .

5. The intelligent assistance method for nursing rehabilitation according to claim 1, characterized in that: The S2 comprises the following steps: Based on the action intention signal matrix, combined with the patient's personalized nerve conduction characteristics and muscle response characteristics, nerve stimulation parameters and muscle stimulation parameters are generated through parameter mapping to form a stimulation parameter data set; Distributing the stimulation parameter data set to the multi-channel output terminal of the neural stimulation device, determining the regional identification of the stimulation point through a spatial positioning algorithm, activating the neural area corresponding to the action intention signal matrix, and generating neural activation data; Based on the nerve activation data, applying an electrical stimulation signal synchronized with the nerve stimulation to the target muscle group through a muscle stimulation device to induce a muscle action response and generate muscle dynamic output data; The embedded sensors collect the response signals of the nerve area and the dynamic output data of the target muscle, and fuse the dynamic output data to form a real-time response data stream of the nerve-muscle ; The real-time response data stream and action intention signal matrix Synchronous correlation analysis is performed, and signal delays are eliminated through a time correction algorithm to generate a neuromuscular synchronous action data stream containing neural activation data and muscle dynamic output data.

6. The intelligent auxiliary method for nursing rehabilitation according to claim 5, characterized in that: The time correction algorithm uses dynamic time warping and adaptive window function, including the following steps: Respond to data streams in real time and action intention signal matrix After normalization, we get and ; Extract the characteristic points of the signal respectively and generate characteristic time series and ; structure and The Euclidean distance matrix ; Calculate the cumulative cost matrix C recursively; Find the optimal time-aligned path P; By estimating and The global trend of constructing an adaptive window function , constrain the search area of ​​the alignment path P; Align according to the optimal path P and , and low-pass filter the synchronized data stream to obtain the time-corrected neuromuscular synchronized action data stream .

7. The intelligent auxiliary method for nursing rehabilitation according to claim 6, characterized in that: The comparing and analyzing the monitoring data with the neuromuscular synchronous action data stream to extract the dynamic error signal comprises: Real-time monitoring data Synchronous neuromuscular movement data flow Perform comparative analysis and extract dynamic error signals ; For error signal Perform frequency domain and time domain analysis to extract dynamic error characteristics; Calculate the error signal The spatiotemporal correlation between the components is used to extract the synergistic characteristics, and the analysis results are fused to generate the final dynamic error signal. .

8. An intelligent assistance system for nursing rehabilitation, based on the intelligent assistance method for nursing rehabilitation according to any one of claims 1 to 7, characterized in that: The intelligent assistance system comprises: A neural signal acquisition and decoding module is used to obtain residual neural signals at the amputation site through a multi-channel implantable or non-implantable neural interface device, and to perform regional decoding of signals from different channels in combination with a distributed processing unit to generate a matrix of action intention signals; A neuromuscular activation and recording module is used to synchronously trigger the activation of the corresponding nerve area and residual limb muscles based on the action intention signal matrix through the coordinated work of the nerve stimulation device and the muscle stimulation device, record the nerve response data and the muscle dynamic output data, and form a neuromuscular synchronous action data stream; Real-time monitoring and dynamic error analysis module, used to monitor muscle mechanical parameters and dynamic bioelectric signals in real time, compare and analyze the monitoring data with the neuromuscular synchronous action data stream, and extract dynamic error signals; A parameter optimization module, used to optimize nerve stimulation parameters and muscle stimulation parameters by using the dynamic error signal through a multimodal data fusion model; The neuromuscular synergistic control module is used to generate a complete neuromuscular synergistic control solution, drive the patient's residual limb or prosthesis to complete the target action in real time, and monitor the match between the action execution effect and the target requirements through sensors.

Images