Data processing device and system for neuro-electric stimulation rehabilitation device

Abstract

The application provides a data processing device and system of a neural electrical stimulation rehabilitation device, the data processing device comprising: a processor, a signal acquisition circuit and an electrical stimulation electrode interface; wherein the processor comprises: a data acquisition module for acquiring multi-modal assessment data of a patient; an assessment prediction module for obtaining an adaptability score and an initial recovery prediction level; a parameter generation module for generating initial stimulation parameters; a determination module for determining a rehabilitation control parameter set of the patient; a feedback optimization module for generating a first instruction and adjusting parameters of a quantitative screening model based on the first instruction; the assessment prediction module is further used for updating the multi-modal assessment data to obtain current assessment data, re-performing assessment prediction on the current assessment data, and outputting a current recovery prediction level; and the feedback optimization module is further used for updating the rehabilitation control parameter set based on the current recovery prediction level.

Description

Technical Field

[0001] This application relates to the field of medical big data processing technology, and more specifically, to a data processing device and system for a neurostimulation rehabilitation device. Background Technology

[0002] In the integration of neuromodulation therapy (such as spinal cord stimulation (SCS)) and rehabilitation medicine, how to formulate and dynamically adjust clinical pathways has always been a clinical challenge. Traditional rehabilitation control parameters rely heavily on the physician's experience and lack objective quantitative indicators. Furthermore, the parameters of neurostimulation rehabilitation equipment are often fixed after being set at the beginning of treatment.

[0003] However, preoperative screening, intraoperative localization, and postoperative rehabilitation are often completed by different doctors or in different systems, resulting in a lack of data sharing. For example, data on electrode placement and stimulation thresholds during surgery are often not transmitted to the rehabilitation therapist, leading to blind and experience-dependent determination of parameters for neurostimulation rehabilitation equipment and resulting in a poor patient experience.

[0004] Therefore, there is an urgent need for a data processing device for neurostimulation rehabilitation equipment that can achieve personalized customization of parameters and has a closed-loop feedback optimization mechanism. Summary of the Invention

[0005] This application addresses the shortcomings of existing methods by proposing a data processing device and system for neurostimulation rehabilitation equipment, which enables personalized customization and dynamic optimization of the parameters of the neurostimulation rehabilitation equipment.

[0006] In a first aspect, embodiments of this application provide a data processing device for a neurostimulation rehabilitation device, used for perioperative rehabilitation regulation of implanted neurostimulation. The data processing device includes: a processor, a signal acquisition circuit, and an electrical stimulation electrode interface; wherein... The electrical stimulation electrode interface is used to output graded electrical stimulation to the implanted first electrode. The signal acquisition circuit is used to acquire intraoperative electromyographic response data corresponding to the graded electrical stimulation. The intraoperative electromyographic response data includes: the threshold current that induces the target muscle group to produce the minimum measurable response, the saturation current that induces the non-target muscle group to produce a side effect response, and the target muscle group activation specificity index. The processor includes: The data acquisition module is used to acquire patients' preoperative multimodal assessment data; The assessment and prediction module is used to input the acquired multimodal rehabilitation assessment data into the quantitative screening model, and output the fitness score and initial recovery prediction level through the quantitative screening model. The parameter generation module is used to determine the effective parameter control window of the neuro-electrical stimulation rehabilitation device based on the threshold current, saturation current and target muscle group activation specificity index in the intraoperative electromyographic response data, and to generate initial stimulation parameters based on the effective parameter control window. The determination module is used to determine the patient's set of rehabilitation control parameters based on the fitness score, the initial recovery prediction level, the initial stimulation parameters, and preset rehabilitation control parameter generation rules. The feedback optimization module is used to generate a first instruction based on the initial recovery prediction level and the actual motor function recovery data of the patient when the neurostimulation rehabilitation device is running based on the rehabilitation control parameter set, and to adjust the parameters of the quantitative screening model based on the first instruction; The assessment and prediction module is also used to update the multimodal assessment data based on the actual motor function recovery data to obtain the current assessment data, and to reassess and predict the quantitative screening model after iteratively updating the input parameters of the current assessment data, and output the current recovery prediction level. The feedback optimization module is also used to update the set of rehabilitation control parameters based on the current recovery prediction level.

[0007] Secondly, embodiments of this application provide a data processing method for a neurostimulation rehabilitation device, used for perioperative rehabilitation regulation of implanted neurostimulation, the method comprising: Acquire preoperative multimodal assessment data of patients, input it into a quantitative screening model, and output fitness scores and initial recovery prediction levels; Based on the application of graded electrical stimulation on the implanted first electrode, corresponding intraoperative electromyographic response data are obtained. The intraoperative electromyographic response data includes: the threshold current that induces the target muscle group to produce the minimum measurable response, the saturation current that induces the non-target muscle group to produce the side effect response, and the target muscle group activation specificity index. Based on the threshold current, the saturation current, and the target muscle group activation specificity index, the effective parameter control window of the neuro-electrical stimulation rehabilitation device is determined, and initial stimulation parameters are generated based on the effective parameter control window. The rehabilitation control parameter set for the patient is determined based on the fitness score, the initial recovery prediction level, the initial stimulation parameters, and the preset rehabilitation control parameter generation rules. Based on the initial recovery prediction level and the actual motor function recovery data of the patient when the neurostimulation rehabilitation device operates based on the rehabilitation control parameter set, a first instruction is generated; Based on the first instruction, the parameters of the quantitative screening model are adjusted, and the current evaluation data is input into the quantitative screening model for re-evaluation and prediction. The current recovery prediction level is output. The current evaluation data is obtained by updating the multimodal evaluation data based on the actual motor function recovery data. The set of rehabilitation control parameters is updated based on the current recovery prediction level.

[0008] Thirdly, embodiments of this application also disclose a data processing system, including the data processing apparatus as described in the first aspect.

[0009] Fourthly, embodiments of this application also disclose a computer-readable storage medium storing a computer program that, when executed by a processor, implements one or more of the data processing methods described in the embodiments of the first aspect of this application.

[0010] Fifthly, embodiments of this application also disclose a computer program product, including a computer program that, when executed by a processor, implements one or more of the data processing methods described in the embodiments of the first aspect of this application.

[0011] The beneficial technical effects of the technical solutions provided in this application include: The solution in this application uses a quantitative screening model to process the patient's multimodal assessment data, outputting an fitness score and an initial recovery prediction level, thus achieving accurate prediction of the patient's treatment potential. Simultaneously, it generates precise initial stimulation parameters by combining intraoperative electromyographic response data, and generates a set of rehabilitation control parameters for the patient based on the fitness score, recovery prediction level, and initial stimulation parameters. This allows the neurostimulation rehabilitation device to operate based on this set of rehabilitation control parameters, thereby achieving personalized customization of the neurostimulation rehabilitation device parameters.

[0012] Additional aspects and advantages of this application will be set forth in part in the description which follows, and will become apparent from the description or may be learned by practice of this application. Attached Figure Description

[0013] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a schematic diagram of the structure of a data processing device for a neurostimulation rehabilitation device provided in an embodiment of this application; Figure 2 A schematic flowchart illustrating a data processing method for a neurostimulation rehabilitation device provided in this application embodiment; Figure 3This is a schematic diagram of the structure of a data processing system provided in an embodiment of this application. Detailed Implementation The embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the embodiments described below with reference to the accompanying drawings are exemplary descriptions for explaining the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions of the embodiments of this application.

[0014] Those skilled in the art will understand that, unless specifically stated otherwise, the terms "described" and "the" as used herein may also include plural forms. It should be further understood that the term "comprising" as used in this application's specification means the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude implementations of other features, information, data, steps, operations, elements, components, and / or combinations thereof supported by this art. It should be understood that when we say an element is "connected" or "coupled" to another element, the element may be directly connected or coupled to the other element, or it may mean that the element and the other element are connected through an intermediate element. Furthermore, "connected" or "coupled" as used herein may include wireless connections or wireless coupling. The term "and / or" as used herein refers to at least one of the items defined by the term; for example, "A and / or B" may be implemented as "A," or as "B," or as "A and B."

[0015] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0016] To facilitate understanding of the technical solution of this application, the following terms will be introduced.

[0017] Electromyography (EMG) is a bioelectrical signal generated by muscles during rest or contraction. It is formed by the temporal and spatial superposition of action potentials of motor units and can be detected by electrodes to assess neuromuscular function. This signal reflects the electrical activity of muscle contraction, including information such as activation timing and contraction intensity, and is mainly used in rehabilitation medicine, sports science, and human-computer interaction.

[0018] Electromyographic signals can be classified according to different acquisition methods as follows: Surface electromyography (sEMG): Non-invasively acquired through electrodes attached to the skin surface, with an amplitude range of 0.01-5 mV and a frequency range of 10-500 Hz; Needle electrode electromyography (EMG) signals: By collecting signals through needle electrodes inserted into muscles, more precise information on deep muscle activity can be obtained.

[0019] Electrical stimulation, also known as electrical stimulation signal, is an electrical signal used to stimulate the target muscle group.

[0020] The technical solution of this application and how it solves the above-mentioned technical problems are described in detail below with specific embodiments. It should be noted that the following embodiments can be referenced, borrowed, or combined with each other, and the same terms, similar features, and similar implementation steps in different embodiments will not be described again.

[0021] The data processing device and system of the neurostimulation rehabilitation equipment proposed in this application will be described in detail below with reference to the accompanying drawings.

[0022] This application provides a data processing device for a neurostimulation rehabilitation device. This data processing device is used for perioperative rehabilitation regulation of implanted neurostimulation. Figure 1 As shown, the data processing device 10 includes a processor 1, a signal acquisition circuit 2, and an electrical stimulation electrode interface 3. The electrical stimulation electrode interface 3 is used to output graded electrical stimulation to the implanted first electrode; the signal acquisition circuit 2 is used to acquire intraoperative electromyographic response data corresponding to the graded electrical stimulation.

[0023] Optionally, the processor 1 includes: a data acquisition module 11, an evaluation and prediction module 12, a parameter generation module 13, a determination module 14, and a feedback optimization module 15, wherein, The data acquisition module 11 is used to acquire the patient's preoperative multimodal assessment data; the assessment prediction module 12 is used to input the acquired multimodal rehabilitation assessment data into the quantitative screening model, and output the fitness score and initial recovery prediction level through the quantitative screening model calculation and processing; the parameter generation module 13 is used to generate initial stimulation parameters based on the intraoperative electromyographic response data acquired by the signal acquisition circuit 2; the determination module 14 is used to determine the patient's rehabilitation control parameter set according to the fitness score, the initial recovery prediction level, the initial stimulation parameters, and the preset rehabilitation control parameter generation rules; the feedback optimization module 15 is used to generate a first instruction based on the initial recovery prediction level and the actual motor function recovery data of the patient when the neurostimulation rehabilitation device operates based on the rehabilitation control parameter set, and adjust the parameters of the quantitative screening model based on the first instruction.

[0024] The aforementioned assessment and prediction module 12 is further configured to update the multimodal assessment data based on the actual motor function recovery data to obtain the current assessment data, and to re-assess and predict the quantitative screening model after iteratively updating the input parameters of the current assessment data, and output the current recovery prediction level; the aforementioned feedback optimization module 15 is further configured to update the rehabilitation control parameter set based on the current recovery prediction level.

[0025] In this embodiment, the processes of the feedback optimization module 15 generating the first instruction, adjusting the parameters of the quantitative screening model, the evaluation and prediction module 12 re-evaluating and predicting, and the feedback optimization module 15 updating the rehabilitation control parameter set are configured to be executed periodically or triggered cyclically within the rehabilitation cycle to dynamically optimize the rehabilitation control parameter set according to the patient's real-time recovery status.

[0026] In this embodiment, by introducing a closed-loop mechanism that generates a first instruction based on actual recovery data to adjust model parameters, dynamic optimization of the rehabilitation control parameter set is achieved, thereby enabling precise control of the neurostimulation rehabilitation device.

[0027] In some embodiments, the intraoperative electromyographic response data includes: a threshold current for inducing a minimum measurable response in the target muscle group, a saturation current for inducing a side effect response in non-target muscle groups, and a target muscle group activation specificity index; the parameter generation module 13 is specifically used for: Based on the threshold current, the saturation current, and the target muscle group activation specificity index, the effective parameter control window of the neuro-electrical stimulation rehabilitation device is determined, and the initial stimulation parameters are generated based on the effective parameters.

[0028] This application provides a closed-loop system that spans the entire rehabilitation process. This system, through the coordinated logic of preoperative assessment, intraoperative electromyography (EMG) calibration, and postoperative rehabilitation feedback, transforms rehabilitation from a static, open-loop process into a dynamic, adaptive one. Specifically: 1. The preoperative quantitative assessment stage serves as the benchmark for closed-loop pre-input. In this stage, the data acquisition module collects patients' preoperative multimodal assessment data, feeds it into the quantitative screening model, and outputs fitness scores and initial recovery prediction levels.

[0029] This stage allows for the quantification of the patient's neurological damage level and tolerance potential to electrical stimulation using preoperative central nervous system imaging and nerve conduction parameters, serving as a benchmark for the rehabilitation plan. Without preoperative assessment, rehabilitation parameters lack individualized adaptation, and only general standardized parameters can be used, which can easily lead to stimulation overload or insufficient therapeutic effect.

[0030] 2. The unique calibration step in this application is the stage where graded stimulation during surgery yields triple electromyographic indices and generates safe initial parameters. During this stage, the electrical stimulation electrode interface outputs graded electrical stimulation to the implanted electrode, and the signal acquisition circuit simultaneously acquires three types of intraoperative specific electromyographic data: threshold current, saturation current, and activation specificity index. The parameter generation module determines the effective parameter control window based on the joint determination of these three data points and outputs the initial stimulation parameters.

[0031] Because preoperative assessment can only predict neural tolerance potential but cannot determine the individualized safe stimulation range; and because implanted electrodes directly contact the spinal cord / brain nerves with extremely narrow current tolerance, preoperative imaging alone cannot distinguish individual differences in muscle response. Therefore, it is necessary to define the safety window by measuring three types of indicators in real time during surgery, to translate the preoperative macroscopic assessment into executable stimulation hardware parameters, and to link the preoperative theoretical assessment with actual postoperative rehabilitation.

[0032] 3. The postoperative bi-dimensional feedback iterative stage presents a closed-loop dynamic adaptive optimization process, including: First-layer model update: The device collects actual motion recovery data from the patient and generates the first instruction to adjust the weights of the quantitative screening model and correct the model prediction bias. Second-level baseline update: The original preoperative multimodal assessment dataset is updated in reverse with real recovery data to generate the current assessment data, which is then input into the updated model to obtain the updated recovery prediction level; The entire set of rehabilitation control parameters is updated synchronously, and the next rehabilitation cycle begins.

[0033] In other words, the preoperative assessment stage outputs the patient's neurological tolerance potential (macro-constraint), and the intraoperative electromyography calibration stage outputs an individualized safety stimulation window (hardware execution boundary). Both are input into the determination module to generate the initial set of rehabilitation parameters. The actual postoperative recovery results simultaneously correct the "preoperative assessment model and rehabilitation parameters" (long-term dynamic calibration), thereby realizing a collaborative mechanism of data exchange and bidirectional correction across the entire cycle of preoperative, intraoperative, and postoperative care.

[0034] Clearly, the "preoperative assessment, intraoperative electromyography, and postoperative rehabilitation feedback" system in this application constitutes a closed-loop collaborative system, which can achieve at least the following technical effects: 1. Addressing the clinical pain point of fragmented data across multiple stages: Current technology separates preoperative imaging, intraoperative electrode testing, and postoperative rehabilitation into independent devices or processes, making data sharing impossible. Doctors can only manually adjust parameters based on experience and multiple reports. The proposed solution utilizes a single system to achieve data sharing across the preoperative, intraoperative, and postoperative stages, automatically adjusting rehabilitation parameters and eliminating biases from human experience.

[0035] 2. Balancing the dual constraints of neurological safety and rehabilitation efficacy: Preoperative assessment predicts the risk of nerve damage, and intraoperative triple electromyography indicators lock in the effective stimulation window that does not damage the nerve. This dual constraint can avoid irreversible nerve damage caused by implanted electrical stimulation. At the same time, based on the patient's actual recovery, parameters are continuously iterated to dynamically match changes in nerve plasticity, thus taking into account long-term rehabilitation effects.

[0036] 3. Achieve continuous self-calibration of the assessment model and continuous improvement of prediction accuracy: In the existing scheme, the preoperative assessment data is fixed after being entered, and the patient's nerves continue to repair after surgery, and the initial assessment gradually becomes distorted; the scheme of this application will update the preoperative assessment baseline with long-term rehabilitation data feedback, and the quantitative screening model will be continuously corrected with the rehabilitation process. The recovery level prediction will become more and more in line with the patient's true state, and the rehabilitation plan will be adaptively and dynamically adjusted, without the need for doctors to take X-rays and reassess periodically.

[0037] 4. Capable of meeting individualized needs and adapting to unique clinical scenarios in the perioperative period of implantation: Compared with existing external or orthopedic rehabilitation solutions: 1) Existing external or orthopedic rehabilitation solutions do not require intraoperative safety calibration and do not pose a risk of nerve damage; the solution proposed in this application is designed with a three-layer collaborative closed loop for spinal cord / brain implanted electrodes, forming a dedicated standardized data processing link from preoperative screening, intraoperative safety testing, and long-term adaptive rehabilitation after surgery, which greatly reduces the threshold for clinical operation.

[0038] 2) Existing in vitro or orthopedic rehabilitation programs do not involve intraoperative safety window calculations, and stimulation parameters are only fine-tuned based on motor feedback. If directly transplanted to implantable devices, excessive current can easily damage the spinal cord / brain nerves. The program proposed in this application requires the simultaneous acquisition of threshold, saturation, and specificity indices during surgery to define an effective control window, thereby limiting the upper and lower limits of stimulation and completely avoiding the risk of irreversible nerve damage from implantable devices. This is a safety control effect that existing combinations cannot achieve at all.

[0039] 3) Existing extracorporeal or orthopedic rehabilitation programs only adjust parameters based on real-time rehabilitation motion data, without considering the patient's preoperative neurological damage. The same stimulation parameters have huge differences in efficacy for patients with different injuries. The program proposed in this application, on the other hand, is based on preoperative quantitative fitness scores to predict the patient's response potential to electrical stimulation. Rehabilitation parameters are combined with neurological tolerance constraints to achieve individualized matching and greatly improve rehabilitation response efficiency.

[0040] 5. Reduce the cost of long-term rehabilitation intervention: The entire closed-loop automatic cycle is iterated, with graded stimulation calibration performed only once during the initial surgery. Subsequent rehabilitation cycles automatically update the assessment model and rehabilitation parameters, reducing the workload of physicians in regular follow-ups and manual modification of stimulation parameters.

[0041] In some embodiments, when determining the effective parameter control window, the parameter generation module 13 is specifically used for: The difference between the saturation current and the threshold current is calculated to obtain the basic dynamic range; If the activation specificity index of the target muscle group is greater than or equal to the preset specificity threshold, the basic dynamic range is determined as the effective parameter control window; If the target muscle group activation specificity index is less than the specificity threshold, the basic dynamic range is attenuated based on the target muscle group activation specificity index to obtain the effective parameter control window.

[0042] In some embodiments, when the parameter generation module 13 generates the initial stimulation parameters based on the effective parameter control window, it is specifically used for: Based on the threshold current, the saturation current, and the preset safety factor, the initial stimulation amplitude is calculated such that the initial stimulation amplitude is between the threshold current and the saturation current. The window width is adjusted based on the effective parameters to determine the initial pulse width, wherein the window width and the initial pulse width are positively correlated. Based on the initial recovery prediction level, the initial stimulation frequency is determined within a safe upper limit of the stimulation frequency, wherein the safe upper limit is determined based on the window width; The initial stimulation amplitude, the initial pulse width, and the initial stimulation frequency are determined as the initial stimulation parameters; The safety factor represents the relative position of the initial stimulus intensity within the effective parameter control window.

[0043] In some embodiments, the multimodal assessment data includes: high-resolution MRI image data and somatosensory evoked potential data; the quantitative screening model employs a deep neural network to assess the degree of neurological function impairment and predict postoperative recovery ability.

[0044] In some embodiments, the determining module 14 is further configured to: Intraoperative CT images are rigidly registered with preoperative MRI images to generate fused images; A three-dimensional model of the spinal cord or brain is reconstructed based on the fused images, and the correspondence between electrode contacts and nerve segments is marked on the three-dimensional model. Based on the aforementioned correspondence, the implantation location and configuration parameters of the electrodes are determined; The first electrode is determined from the electrodes based on the configuration parameters.

[0045] In some embodiments, the determining module 14 is specifically used for: Based on the fitness score, the execution intensity coefficient of the operating parameters of the neuro-electrical stimulation rehabilitation device is determined; Based on the initial recovery prediction level, the time nodes for dividing the rehabilitation stage are determined; Based on the initial stimulus parameters, a target training action is determined by matching within a preset training action library. Using the execution intensity coefficient, the time nodes of the rehabilitation stage and the target training actions are weighted and adjusted to generate the rehabilitation control parameter set for the patient.

[0046] In some embodiments, when the feedback optimization module 15 updates the set of rehabilitation control parameters based on the current recovery prediction level, it is specifically used for: Determine the update intensity coefficient corresponding to the current recovery prediction level; Based on the updated intensity coefficient, the training load parameters in the already executed rehabilitation control parameter set are adjusted; and / or, Based on the rehabilitation stage corresponding to the current recovery prediction level, a first training action is determined from a preset training action library, and the current training action is replaced with the first training action.

[0047] In some embodiments, the fitness score characterizes the patient’s expected response to neuromodulation therapy; The fitness score is obtained by the quantitative screening model performing the following operations: Obtain physiological index data and imaging feature data from the multimodal assessment data; Calculate the similarity between the physiological indicator data and the preset ideal treatment indications; The fitness score is generated by weighted calculation based on the similarity and the extent of neural damage in the imaging feature data.

[0048] In some embodiments, when generating the first instruction, the feedback optimization module 15 is specifically used for: Feature extraction is performed on the actual motor function recovery data to generate an actual recovery feature vector; The recovery prediction level output by the quantization screening model is mapped to a prediction feature vector; Calculate the deviation value between the actual recovery feature vector and the predicted feature vector, where the deviation value represents the loss function value between the prediction result and the actual control effect; The first instruction is generated based on the deviation value, and the first instruction is used to adjust the weight parameters of the quantization screening model through the backpropagation algorithm.

[0049] The data processing device provided in this application processes the patient's multimodal assessment data through a quantitative screening model, outputting an fitness score and an initial recovery prediction level, thus achieving accurate prediction of the patient's treatment potential. Simultaneously, it generates precise initial stimulation parameters by combining intraoperative electromyographic response data, and generates a set of rehabilitation control parameters for the patient based on the fitness score, recovery prediction level, and initial stimulation parameters. This allows the neurostimulation rehabilitation device to operate based on this set of rehabilitation control parameters, thereby achieving personalized customization of the neurostimulation rehabilitation device parameters.

[0050] In addition, a closed-loop mechanism was introduced to generate the first instruction based on actual recovery data to adjust the model parameters, thereby realizing the dynamic optimization of the rehabilitation control parameter set and improving the accuracy and adaptability of the neurostimulation rehabilitation equipment.

[0051] The data processing device of this application embodiment can execute the data processing method provided in this application embodiment. The implementation principle is similar. The actions performed by each module and unit in the data processing device in each embodiment of this application are corresponding to the steps in the data processing method in each embodiment of this application. For detailed functional descriptions of each module of the data processing device, please refer to the descriptions in the corresponding data processing methods shown below, which will not be repeated here.

[0052] In some embodiments, a data processing method for a neurostimulation rehabilitation device is provided. For example... Figure 2 As shown, the method includes: S1. Obtain the patient's preoperative multimodal assessment data, input it into the quantitative screening model, and output the fitness score and initial recovery prediction level.

[0053] In some embodiments, multimodal assessment data can be understood as a collection of patient physiological information obtained through different modal medical testing methods, such as including but not limited to: high-resolution MRI (magnetic resonance imaging) image data, somatosensory evoked potential data, patient medical record text data, and biochemical test indicators.

[0054] In some embodiments, the acquired multimodal rehabilitation assessment data is input into a quantitative screening model, and the model is processed to output fitness scores and initial recovery prediction levels.

[0055] In some embodiments, the quantization screening model is a pre-trained machine learning model, such as a deep neural network (DNN) or a convolutional neural network (CNN), which is able to extract feature vectors from high-dimensional multimodal evaluation data and map them to two output dimensions: fitness score and initial recovery prediction level, to assess the degree of neurological function impairment and predict postoperative recovery potential.

[0056] In some embodiments, the fitness score is a quantified probability value or score representing the patient's expected response to neuromodulation therapy. Optionally, the fitness score can be normalized to a range of 0 to 1, with a higher score indicating a better expected response.

[0057] In some embodiments, the fitness score is obtained by the quantitative screening model performing the following operations: Obtain physiological index data and imaging feature data from the multimodal assessment data; Calculate the similarity between the physiological indicator data and the preset ideal treatment indications; The fitness score is generated by weighted calculation based on the similarity and the extent of neural damage in the imaging feature data.

[0058] In some embodiments, physiological indicators may include, but are not limited to, age, disease duration, etc., and imaging features may include, but are not limited to, the extent of nerve damage (e.g., length of damage). Damaged area Including the coordinates of the damaged location, the reciprocal of the Euclidean distance between physiological indicators and ideal treatment criteria can be used as the similarity score. S Optionally, the fitness score can be determined based on the following formula. S :

[0059] in, , To quantify the weight parameters of the screening model, This represents the weighting coefficient for the location of nerve damage.

[0060] In some embodiments, the initial recovery prediction grade is used to predict the patient's potential for postoperative motor function recovery, such as the grade predicted according to the ASIA disability grading system. Optionally, the initial recovery prediction grade may include, but is not limited to: Grade A (corresponding to excellent), Grade B (corresponding to good), Grade C (corresponding to moderate), and Grade D (corresponding to poor).

[0061] S2. Based on the intraoperative electromyographic response data obtained by applying graded electrical stimulation to the first electrode, generate initial stimulation parameters.

[0062] In some embodiments, graded electrical stimulation can be output to the first electrode through the electrical stimulation electrode interface, while the corresponding intraoperative electromyographic response data can be acquired through the signal acquisition circuit to generate initial stimulation parameters.

[0063] In some embodiments, the first electrode refers to an electrode contact implanted in the patient's body (such as in the epidural space of the spinal cord or a specific nucleus in the brain) for applying electrical stimulation pulses.

[0064] In some embodiments, the implantation location of the electrode can be determined by image fusion navigation technology, and a specific contact point can be selected as the first electrode.

[0065] In some embodiments, the above method may further include: Intraoperative CT images are rigidly registered with preoperative MRI images to generate fused images; A three-dimensional model of the spinal cord or brain is reconstructed based on the fused images, and the correspondence between electrode contacts and nerve segments is marked on the three-dimensional model. Based on the aforementioned correspondence, the implantation location and configuration parameters of the electrodes are determined; The first electrode is determined from the electrodes based on the configuration parameters.

[0066] In this embodiment, rigid registration of intraoperative CT images with preoperative MRI images can be understood as: using translation and rotation algorithms, the high-definition preoperative MRI images are overlaid on the intraoperative CT images to obtain a fused image that has both accurate coordinates and a clear neural display of the navigation map. This fused image includes complete structures such as bones, nerves, and blood vessels.

[0067] In some embodiments, a 3D model reconstruction is further performed based on the fused image. The reconstruction process may employ voxel reconstruction or surface rendering.

[0068] Optionally, voxel reconstruction directly utilizes the three-dimensional volume data of fused images. By setting different grayscale thresholds, isosurfaces of different tissues such as the spinal cord, brain parenchyma, and cerebrospinal fluid are extracted to construct a three-dimensional anatomical model.

[0069] Optionally, the surface rendering algorithm first extracts the contour lines of the region of interest, and then generates a smooth 3D surface model by fitting triangular facets.

[0070] In some embodiments, on the reconstructed 3D model, image segmentation algorithms are used to automatically identify the positions of electrode contacts, and an anatomical atlas database is used to calibrate the correspondence between electrode contacts and nerve segments. For example, in a spinal cord stimulation scenario, the spatial distance between the electrode contacts and specific neural structures such as the dorsal horn of the spinal cord and the dorsal root entry area can be calculated based on the spatial coordinates of the electrode contacts, and marked on the model with different colors or labels. After calibration, the electrode implantation location and configuration parameters can be further determined based on the correspondence between electrode contacts and nerve segments.

[0071] Optional configuration parameters include: electrode contact combination (such as single-pole, bipolar or tripolar mode), contact activation sequence, etc.

[0072] Then, based on the location of the target nerve segment, the electric field distribution generated by different contact point combinations can be calculated, and a configuration scheme that maximizes coverage of the target area and minimizes the impact on non-target areas can be selected. Then, based on this configuration scheme, the most suitable contact point as the stimulation source is selected from the electrode array as the first electrode.

[0073] It should be understood that the selection of the first electrode is not fixed and can be dynamically adjusted based on feedback from intraoperative electromyographic response data. For example, if the initially selected first electrode induces a response in a non-target muscle group during the test, an adjacent contact point can be automatically selected as the new first electrode, thereby achieving intraoperative fine-tuning of the implantation position.

[0074] The image fusion navigation process described above can effectively improve the accuracy and safety of neural modulation.

[0075] In some embodiments, step S2 above may specifically include: S200 (not shown in the figure) obtains corresponding intraoperative electromyographic response data based on graded electrical stimulation applied to the implanted first electrode.

[0076] Optionally, the intraoperative electromyographic response data includes: the threshold current that induces the target muscle group to produce the minimum measurable response, the saturation current that induces the non-target muscle group to produce a side effect response, and the target muscle group activation specificity index.

[0077] In some embodiments, graded electrical stimulation refers to the application of electrical stimulation sequentially through the first electrode during the intraoperative testing phase, according to a preset current intensity gradient (e.g., in 0.5mA increments). Intraoperative electromyographic response data refers to the electrical signal response of the target muscle group recorded by a signal acquisition device simultaneously with the application of stimulation.

[0078] Optionally, the threshold current refers to the current value that can induce the target muscle group to produce the minimum measurable electromyographic response, which is usually manifested as a sudden increase in the amplitude of the electromyographic signal exceeding the background noise level.

[0079] Optionally, saturation current refers to the current value at which the current intensity increases to a certain level and begins to induce a response in non-target muscle groups or cause discomfort to the patient. This indicates that the stimulation range has spread to non-target nerve fibers.

[0080] Optionally, the target muscle group activation specificity index is used to quantify the precision of stimulation. It can be calculated by the ratio of the electromyographic response amplitude of the target muscle group to the response amplitude of the non-target muscle group. The higher the specificity index, the better the current focusing and the lower the risk of side effects.

[0081] For example, the specificity index C can be calculated based on the following formula: C = A target / (A target + Anon - target ), where A target For the integral of the electromyographic amplitude of the target muscle group, A non - target The integral of the electromyographic amplitude of the non-target muscle group.

[0082] S201 (not shown in the figure) determines an effective parameter control window based on the threshold current, the saturation current, and the target muscle group activation specificity index.

[0083] S202 (not shown in the figure) generates the initial stimulation parameters based on the effective parameter adjustment window.

[0084] In this embodiment, by introducing threshold current, saturation current and specificity index, the effective parameter control window can be precisely defined, avoiding problems such as insufficient stimulation intensity or excessive side effects.

[0085] In some embodiments, step S201 may specifically include: The difference between the saturation current and the threshold current is calculated to obtain the basic dynamic range; If the activation specificity index of the target muscle group is greater than or equal to the preset specificity threshold, the basic dynamic range is determined as the effective parameter control window; If the target muscle group activation specificity index is less than the specificity threshold, the basic dynamic range is attenuated based on the target muscle group activation specificity index to obtain the effective parameter control window.

[0086] In some embodiments, the basic dynamic range can be understood as the theoretically available current regulation range.

[0087] In some embodiments, a low specificity index indicates that the stimulation current diffuses easily, and directly using the baseline dynamic range may lead to side effects. Therefore, this embodiment introduces an attenuation mechanism. For example, a specificity threshold of 1.5 is set. When the measured specificity index is 2.0, it indicates good focusing. When the measured specificity index is 0.8, it indicates poor focusing, and attenuation processing will be initiated. Through this adaptive correction, the current adjustment range can be automatically reduced for patients with poor focusing, thereby avoiding the risk of side effects.

[0088] Therefore, the above-described embodiment improves the accuracy of the effective parameter control window and ensures the safety of neural modulation by correcting the basic dynamic range through a specificity index.

[0089] In some embodiments, step S202 may specifically include: Based on the threshold current, the saturation current, and the preset safety factor, the initial stimulation amplitude is calculated such that the initial stimulation amplitude is between the threshold current and the saturation current. The window width is adjusted based on the effective parameters to determine the initial pulse width, wherein the window width and the initial pulse width are positively correlated. Based on the initial recovery prediction level, the initial stimulation frequency is determined within a safe upper limit of the stimulation frequency, wherein the safe upper limit is determined based on the window width; The initial stimulation amplitude, the initial pulse width, and the initial stimulation frequency are determined as the initial stimulation parameters.

[0090] Optionally, the aforementioned safety factor characterizes the relative position of the initial stimulus intensity within the effective parameter control window. For example, the safety factor can be set between 0.3 and 0.7, but is not limited to this.

[0091] In some embodiments, the initial stimulus amplitude can be calculated as follows: Initial stimulus amplitude = Threshold current + (Effective parameter control window width × Safety factor).

[0092] For example, if the threshold current is 1.0mA, the effective window width is 2.0mA, and the safety factor is 0.5, then the initial stimulation amplitude is set to 2.0mA. This value is in the middle of the safety window, which can ensure the therapeutic effect while leaving room for adjustment.

[0093] In some embodiments, the positive correlation can be understood as a linear relationship, for example: initial pulse width = preset reference pulse width + k × window width, where k is a positive coefficient.

[0094] In some embodiments, the pulse width can be determined based on the principle of charge balance. Optionally, a wider window width indicates a larger range of tolerable charge amounts. Therefore, a wider pulse width can be set to reduce current density and decrease the risk of tissue damage.

[0095] In some embodiments, the stimulation frequency can be set with reference to the initial recovery prediction level. If the prediction level is high, it indicates good neural excitability, and a lower frequency (e.g., 30Hz) can be used; if the prediction level is low, a higher frequency (e.g., 50Hz) may be needed to maintain the therapeutic effect.

[0096] It should be noted that in this embodiment, the frequency setting is limited by a safety upper limit. Optionally, this safety upper limit is determined by the maximum safe charge amount derived from the window width, in order to prevent high-frequency stimulation from causing tissue heating.

[0097] In this embodiment, the initial stimulation parameters may include the initial stimulation amplitude, the initial pulse width, and the initial stimulation frequency.

[0098] S3. Determine the patient's set of rehabilitation control parameters based on the fitness score, the initial recovery prediction level, the initial stimulation parameters, and the preset rehabilitation control parameter generation rules.

[0099] In some embodiments, the rules for generating rehabilitation control parameters can be understood as a pre-configured logical algorithm or expert knowledge base that defines how to map the patient's assessment results to a set of control parameters corresponding to a specific rehabilitation training plan. In this embodiment, an individualized set of rehabilitation control parameters is generated by comprehensively considering the patient's treatment potential score, predicted recovery level, and stimulation parameters determined intraoperatively. For example, if the fitness score is high, a higher intensity training load parameter can be configured in the rehabilitation control parameter set; if the initial stimulation parameters show that the patient responds well to stimulation at a specific frequency, then specific movement training at that frequency of stimulation is recommended in the rehabilitation control parameter set.

[0100] In some embodiments, the rehabilitation control parameter set may include parameters related to the selection of the training exercise library, the setting of training intensity, and the division of training cycles.

[0101] In some embodiments, step S3 above may specifically include: Based on the fitness score, the execution intensity coefficient of the operating parameters of the neuro-electrical stimulation rehabilitation device is determined; Based on the initial recovery prediction level, the time nodes for dividing the rehabilitation stage are determined; Based on the initial stimulus parameters, a target training action is determined by matching within a preset training action library. Using the execution intensity coefficient, the time nodes of the rehabilitation stage and the target training actions are weighted and adjusted to generate the rehabilitation control parameter set for the patient.

[0102] In some embodiments, the execution intensity coefficient is used to regulate the overall load level of rehabilitation training, and there is a preset mapping relationship between it and the fitness score. For example, when the fitness score is greater than 0.8, the patient is identified as a high-response type, and the execution intensity coefficient is set to 1.2, indicating that the training load can be increased by 20% based on the standard rehabilitation control parameter set; when the fitness score is between 0.5 and 0.8, the patient is identified as a medium-response type, and the execution intensity coefficient is set to 1.0, that is, the training is based on the standard rehabilitation control parameter set; when the fitness score is less than 0.5, the patient is identified as a low-response type, and the execution intensity coefficient is set to 0.8, indicating that the training load needs to be reduced to avoid patient fatigue or resistance.

[0103] It should be understood that the above values ​​are merely examples and may be adjusted based on the experience of clinical experts or large sample statistical data in actual applications. Furthermore, these values ​​do not constitute any limitation on the solutions in the embodiments of this application.

[0104] In some embodiments, the initial recovery prediction level is based on the functional level that the patient is likely to achieve postoperatively, as predicted by the model. Different levels correspond to different neural plasticity windows. For example, for patients with higher prediction levels, whose neurological function recovery potential is greater, the time points of the acute phase, recovery phase, and consolidation phase can be compressed, such as setting the acute phase to 2 weeks postoperatively, to allow for a faster entry into the intensive training phase. For patients with lower prediction levels, the acute phase may be extended to 4 weeks postoperatively to ensure that the neural tissue has sufficient adaptation time. This dynamic division based on prediction levels avoids the problems of "overtraining" or "undertraining" caused by traditional fixed time points.

[0105] In some embodiments, the training action library contains pre-stored sets of actions for different nerve segment stimulation patterns. For example, if the stimulation frequency in the initial stimulation parameters is high (e.g., above 50 Hz), it usually corresponds to a tonic stimulation pattern, in which case resistance training actions that enhance muscle tone can be preferentially matched; if the stimulation frequency is low (e.g., below 30 Hz), it usually corresponds to a functional stimulation pattern, in which case fine motor training that promotes sensorimotor integration can be preferentially matched.

[0106] It should be noted that the matching here can be achieved by calculating the similarity between the stimulus parameter feature vector and the action label vector in the action library, and the action with the highest similarity is selected as the target training action.

[0107] In some embodiments, the execution intensity coefficient is used to weight and adjust the time nodes and target training actions of the rehabilitation phase to generate a set of rehabilitation control parameters for the patient. Specifically, this may include multiplying the execution intensity coefficient by the standard training duration or number of repetitions to obtain individualized training parameters. For example, if an action in the standard program needs to be repeated 10 times and the execution intensity coefficient is 1.2, then the adjusted number of repetitions is 12.

[0108] The solution described in this embodiment can generate a set of individualized rehabilitation control parameters that include specific training actions, training intensity, training frequency, and a phase schedule.

[0109] S4. Based on the initial recovery prediction level and the actual motor function recovery data of the patient when the neurostimulation rehabilitation device operates based on the rehabilitation control parameter set, a first instruction is generated.

[0110] In some embodiments, after the neuro-electrical stimulation rehabilitation device has been running for a period of time according to the rehabilitation control parameter set, the patient's actual motor function recovery data can be obtained, such as joint range of motion, muscle strength score, walking speed, etc.

[0111] In some embodiments, the first instruction is a signal used to trigger the optimization of the quantization screening model.

[0112] In some embodiments, step S4 above may specifically include: Feature extraction is performed on the actual motor function recovery data to generate an actual recovery feature vector; The initial recovery prediction level output by the quantization screening model is mapped to a prediction feature vector; Calculate the deviation value between the actual recovery feature vector and the predicted feature vector, where the deviation value represents the loss function value between the prediction result and the actual control effect; The first instruction is generated based on the deviation value, and the first instruction is used to adjust the weight parameters of the quantization screening model through the backpropagation algorithm.

[0113] In some embodiments, the first instruction can be understood as a parameter update instruction package containing gradient information. The gradient of the loss function with respect to the model weight parameters is calculated based on the bias value using the backpropagation algorithm.

[0114] Optionally, the weights of the quantization screening model can be updated using gradient descent. The first instruction contains either the calculated gradient value or the updated weights.

[0115] In some embodiments, actual motor function recovery data typically contains a large amount of unstructured or semi-structured raw signals. In order for the quantization screening model to process this data, feature extraction and standardization are required first.

[0116] Optionally, the feature extraction process may include time-domain feature extraction and frequency-domain feature extraction.

[0117] Time-domain features may include, but are not limited to, the signal's mean, variance, standard deviation, number of zero crossings, and root mean square (RMS) value. These features can intuitively reflect the amplitude and stability of motion. For example, the RMS value is often used to assess the degree of muscle activation.

[0118] Frequency domain features are extracted by performing a Fast Fourier Transform (FFT) on the time domain signal, and may include, but are not limited to, power spectral density, median frequency, and spectral entropy. These features can reflect the coordination and fatigue level of neural control.

[0119] Standardization refers to mapping the extracted feature values ​​to the same order of magnitude using a normalization algorithm. For example, the Z-score standardization method can be used, with the formula: z = (x - μ) / σ, where x is the original feature value, μ is the sample mean, and σ is the sample standard deviation. Standardization eliminates the impact of differences in feature dimensions on model training.

[0120] In some embodiments, the standardized data can be mapped into an actual recovery feature vector containing n dimensions such as muscle strength, joint range of motion, and walking speed. .

[0121] In some embodiments, the initial recovery prediction level can be mapped to the corresponding prediction feature vector based on a pre-stored correspondence between recovery prediction levels and standard feature vectors (e.g., the standard feature vectors corresponding to level A are [0.9, 0.8, 0.7], and the standard feature vectors corresponding to level B are [0.6, 0.5, 0.4], etc.). Then, the actual recovered feature vector is calculated based on the following formula. With predicted feature vectors Deviation values ​​between:

[0122] in, The dimension of the feature vector. and Represent the two vectors at the th... The values ​​in each dimension.

[0123] In this embodiment, when the Loss value is large, it indicates that there is a significant difference between the current prediction result of the quantitative screening model and the actual rehabilitation effect of the patient, and the first instruction will trigger a larger gradient update step size; when the Loss value is close to zero, it indicates that the quantitative screening model prediction is accurate, and the first instruction will trigger fine-tuning or maintain the current parameters.

[0124] The scheme in the above embodiments, through deviation calculation and backpropagation, achieves continuous optimization of the quantitative screening model and improves the accuracy of model prediction.

[0125] S5. Adjust the parameters of the quantitative screening model based on the first instruction, input the current evaluation data into the quantitative screening model to re-evaluate and predict, and output the current recovery prediction level.

[0126] Optionally, the current assessment data is obtained by updating the multimodal assessment data based on the actual motor function recovery data, for example, by replacing or supplementing the preoperative data with the latest rehabilitation data. The first instruction triggers the quantitative screening model to update its parameters, for example, by adjusting the weight matrix of the neural network through a backpropagation algorithm to correct the model's prediction bias. Subsequently, the updated current assessment data is input into the adjusted model, outputting the current recovery prediction level. This current recovery prediction level reflects the patient's latest recovery potential prediction and is more timely and accurate than the initial prediction.

[0127] S6. Update the set of rehabilitation control parameters based on the current recovery prediction level.

[0128] In some embodiments, the set of rehabilitation control parameters can be dynamically adjusted based on changes in the current recovery prediction level. For example, if the current level indicates that the patient's recovery is ahead of schedule, the training difficulty can be increased or new training movements can be introduced; if the level indicates that the recovery is lagging, the training intensity can be reduced or the stimulation parameter combination scheme can be adjusted.

[0129] In some embodiments, step S6 above may specifically include: Determine the update intensity coefficient corresponding to the current recovery prediction level; Based on the updated intensity coefficient, the training load parameters in the already executed rehabilitation control parameter set are adjusted; and / or, Based on the rehabilitation stage corresponding to the current recovery prediction level, a first training action is determined from a preset training action library, and the current training action is replaced with the first training action.

[0130] As the rehabilitation process progresses, the patient's actual recovery may deviate from the prediction, thus requiring dynamic updates to the rehabilitation control parameter set.

[0131] In some embodiments, an updated intensity coefficient corresponding to the current recovery prediction level is determined. This coefficient is determined in a similar manner to the determination of the execution intensity coefficient described above, except that the data source used needs to be updated to the latest assessment results. If the current recovery prediction level indicates that the patient's recovery progress is better than expected, the updated intensity coefficient will be adjusted upwards accordingly; otherwise, it will be adjusted downwards.

[0132] In some embodiments, training load parameters may include, but are not limited to, the number of training sets, the number of repetitions per set, the resistance level, and the rest interval. For example, if the updated intensity coefficient is increased by 0.2 compared to the initial value (the execution emphasis coefficient mentioned above), the resistance level of the leg press training can be automatically adjusted from 5kg to 6kg, or the number of training sets can be increased from 3 to 4. Simultaneously, patient fatigue feedback can be monitored; if the patient reports discomfort after the resistance level adjustment, a rollback mechanism is triggered to restore the original parameters and reassess.

[0133] In some embodiments, the scheme of determining a first training action from a preset training action library based on the rehabilitation stage corresponding to the current recovery prediction level and replacing the current training action with the first training action reflects the progressive nature of the rehabilitation control parameter set. For example, in the early stages of rehabilitation, patients may mainly perform passive cycling exercises in a supine position to maintain joint range of motion; as the current recovery prediction level increases and it is determined that the patient has entered the recovery period, active cycling exercises or standing balance exercises will be retrieved from the action library as the first training action to replace the original passive training action.

[0134] The solution in this application embodiment achieves precise matching between the rehabilitation control parameter set and the patient's real-time status, maximizing the patient's potential for neurological function recovery, while ensuring the continuity and safety of rehabilitation training.

[0135] In the above embodiments, the processes of generating the first instruction, adjusting the parameters of the quantitative screening model, re-evaluating and predicting, and updating the rehabilitation control parameter set in steps S4 to S6 are configured to be executed periodically or triggered cyclically within the rehabilitation cycle, so as to dynamically optimize the rehabilitation control parameter set according to the patient's real-time recovery status.

[0136] It should be noted that periodic execution refers to automatic execution at fixed time intervals (such as once a week or once every two weeks); triggered execution refers to execution triggered when a patient's key indicators change or reach a specific point.

[0137] It should also be noted that when the operation of generating the first instruction is executed repeatedly, the current recovery prediction level output by the quantization screening model is mapped to a predicted feature vector, and the deviation value is further calculated. The weight parameters of the quantization screening model here are adjusted based on the previous first instruction.

[0138] In this embodiment, a closed-loop optimization mechanism enables the parameters of the neurostimulation rehabilitation device to be dynamically optimized adaptively based on the patient's real-time recovery status, thereby achieving precise control of the neurostimulation rehabilitation device.

[0139] In summary, the embodiments of this application provide a data processing scheme for a neurostimulation rehabilitation device, which can achieve the following technical effects: 1. By processing patients' multimodal assessment data through a quantitative screening model, the fitness score and initial recovery prediction level are output, realizing an objective quantitative assessment of treatment fitness and recovery potential, overcoming the limitations of traditional reliance on doctors' subjective experience judgment.

[0140] 2. By generating initial stimulation parameters based on intraoperative electromyographic response data, and especially by determining the effective parameter adjustment window based on threshold current, saturation current and specificity index, the range of safe and effective stimulation parameters can be accurately located, avoiding the risk of ineffective stimulation or side effects.

[0141] 3. Through closed-loop optimization steps, actual motor function recovery data of patients are collected, the first instruction is generated to adjust the parameters of the quantitative screening model, and the rehabilitation control parameter set is updated based on the current recovery prediction level, thereby realizing the dynamic optimization of the rehabilitation control parameter set and enabling precise control of the nerve electrical stimulation rehabilitation equipment.

[0142] Based on the same principles as the methods shown in the embodiments of this application, the embodiments of this application also provide a data processing system, which includes the data processing apparatus provided in the above embodiments.

[0143] In an alternative embodiment, a data processing system, such as... Figure 3 As shown, Figure 3 The data processing system 20 shown includes a processor 21 and a memory 23. The processor 21 is communicatively connected to the memory 23, for example, via a bus 22. Optionally, the data processing system 20 may also include a transceiver 24, which can be used for data interaction between the data processing system and other electronic devices, such as sending and / or receiving data. It should be noted that in practical applications, the transceiver 24 is not limited to one type, and the structure of this data processing system 20 does not constitute a limitation on the embodiments of this application.

[0144] Processor 21 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof, including the chips or data processing devices described in any of the foregoing embodiments. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 21 may also be a combination that implements computing functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0145] Bus 22 may include a pathway for transmitting information between the aforementioned components. Bus 22 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Bus 22 may be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 3 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0146] The memory 23 may be a ROM (Read-Only Memory) or other type of static storage device capable of storing static information and instructions, RAM (Random Access Memory) or other type of dynamic storage device capable of storing information and instructions, or it may be an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read-Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto.

[0147] The memory 23 is used to store computer programs that execute the embodiments of this application, and the execution is controlled by the processor 21. The processor 21 is used to execute the computer programs stored in the memory 23 to implement the steps shown in the foregoing method embodiments.

[0148] This application provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it can implement the steps and corresponding content of the aforementioned method embodiments.

[0149] This application also provides a computer program product, including a computer program that, when executed by a processor, can implement the steps and corresponding content of the aforementioned method embodiments.

[0150] In the description of this application, the terms "center," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate directions or positional relationships based on the exemplary directions or positional relationships shown in the accompanying drawings. They are used to facilitate the description or simplification of the embodiments of this application and are not intended to indicate or imply that the device or component referred to must have a specific orientation or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.

[0151] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0152] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "joining" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal communication between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0153] In the description of this specification, specific features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0154] The above description is only a partial implementation of this application. It should be noted that for those skilled in the art, other similar implementation methods based on the technical concept of this application, without departing from the technical concept of this application, also fall within the protection scope of the embodiments of this application.

Claims

1. A data processing device for a neuro-electrical stimulation rehabilitation device, characterized in that, For perioperative rehabilitation regulation of implantable neurostimulation, the data processing device includes: a processor, a signal acquisition circuit, and an electrical stimulation electrode interface; wherein... The electrical stimulation electrode interface is used to output graded electrical stimulation to the implanted first electrode. The signal acquisition circuit is used to acquire intraoperative electromyographic response data corresponding to the graded electrical stimulation. The intraoperative electromyographic response data includes: the threshold current that induces the target muscle group to produce the minimum measurable response, the saturation current that induces the non-target muscle group to produce a side effect response, and the target muscle group activation specificity index. The processor includes: The data acquisition module is used to acquire patients' preoperative multimodal assessment data; The assessment and prediction module is used to input the acquired multimodal rehabilitation assessment data into the quantitative screening model, and output the fitness score and initial recovery prediction level through the quantitative screening model. The parameter generation module is used to determine the effective parameter control window of the neuro-electrical stimulation rehabilitation device based on the threshold current, saturation current and target muscle group activation specificity index in the intraoperative electromyographic response data, and to generate initial stimulation parameters based on the effective parameter control window. The determination module is used to determine the patient's set of rehabilitation control parameters based on the fitness score, the initial recovery prediction level, the initial stimulation parameters, and preset rehabilitation control parameter generation rules. The feedback optimization module is used to generate a first instruction based on the initial recovery prediction level and the actual motor function recovery data of the patient when the neurostimulation rehabilitation device is running based on the rehabilitation control parameter set, and to adjust the parameters of the quantitative screening model based on the first instruction; The assessment and prediction module is also used to update the multimodal assessment data based on the actual motor function recovery data to obtain the current assessment data, and to reassess and predict the quantitative screening model after iteratively updating the input parameters of the current assessment data, and output the current recovery prediction level. The feedback optimization module is also used to update the set of rehabilitation control parameters based on the current recovery prediction level.

2. The data processing apparatus according to claim 1, characterized in that, When determining the effective parameter control window, the parameter generation module is specifically used for: The difference between the saturation current and the threshold current is calculated to obtain the basic dynamic range; If the target muscle group activation specificity index is greater than or equal to the preset specificity threshold, the basic dynamic range is determined as the effective parameter control window; If the target muscle group activation specificity index is less than the specificity threshold, the basic dynamic range is attenuated based on the target muscle group activation specificity index to obtain the effective parameter control window.

3. The data processing apparatus according to claim 2, characterized in that, When generating the initial stimulation parameters based on the effective parameter control window, the parameter generation module is specifically used for: Based on the threshold current, the saturation current, and the preset safety factor, the initial stimulation amplitude is calculated such that the initial stimulation amplitude is between the threshold current and the saturation current. The window width is adjusted based on the effective parameters to determine the initial pulse width, wherein the window width and the initial pulse width are positively correlated. Based on the initial recovery prediction level, the initial stimulation frequency is determined within a safe upper limit of the stimulation frequency, wherein the safe upper limit is determined based on the window width; The initial stimulation amplitude, the initial pulse width, and the initial stimulation frequency are determined as the initial stimulation parameters; The safety factor represents the relative position of the initial stimulus intensity within the effective parameter control window.

4. The data processing apparatus according to any one of claims 1-3, characterized in that, The multimodal assessment data includes: high-resolution MRI image data and somatosensory evoked potential data; the quantitative screening model uses a deep neural network to assess the degree of neurological function impairment and predict postoperative recovery ability.

5. The data processing apparatus according to any one of claims 1-3, characterized in that, The determining module is also used for: Intraoperative CT images are rigidly registered with preoperative MRI images to generate fused images; A three-dimensional model of the spinal cord or brain is reconstructed based on the fused images, and the correspondence between electrode contacts and nerve segments is marked on the three-dimensional model. Based on the aforementioned correspondence, the implantation location and configuration parameters of the electrodes are determined; The first electrode is determined from the electrodes based on the configuration parameters.

6. The data processing apparatus according to claim 1, characterized in that, The determining module is specifically used for: Based on the fitness score, the execution intensity coefficient of the operating parameters of the neuro-electrical stimulation rehabilitation device is determined; Based on the initial recovery prediction level, the time nodes for dividing the rehabilitation stage are determined; Based on the initial stimulus parameters, a target training action is determined by matching within a preset training action library. Using the execution intensity coefficient, the time nodes of the rehabilitation stage and the target training actions are weighted and adjusted to generate the rehabilitation control parameter set for the patient.

7. The data processing apparatus according to claim 1 or 6, characterized in that, The feedback optimization module, when updating the rehabilitation control parameter set based on the current recovery prediction level, is specifically used for: Determine the update intensity coefficient corresponding to the current recovery prediction level; Based on the updated intensity coefficient, the training load parameters in the already executed rehabilitation control parameter set are adjusted; and / or, Based on the rehabilitation stage corresponding to the current recovery prediction level, a first training action is determined from a preset training action library, and the current training action is replaced with the first training action.

8. The data processing apparatus according to claim 1 or 6, characterized in that, The fitness score characterizes the patient's expected response to neuromodulation therapy; The fitness score is obtained by the quantitative screening model performing the following operations: Obtain physiological index data and imaging feature data from the multimodal assessment data; Calculate the similarity between the physiological indicator data and the preset ideal treatment indications; The fitness score is generated by weighted calculation based on the similarity and the extent of neural damage in the imaging feature data.

9. The data processing apparatus according to any one of claims 1-3, characterized in that, When generating the first instruction, the feedback optimization module is specifically used for: Feature extraction is performed on the actual motor function recovery data to generate an actual recovery feature vector; The recovery prediction level output by the quantization screening model is mapped to a prediction feature vector; Calculate the deviation value between the actual recovery feature vector and the predicted feature vector, where the deviation value represents the loss function value between the prediction result and the actual control effect; The first instruction is generated based on the deviation value, and the first instruction is used to adjust the weight parameters of the quantization screening model through the backpropagation algorithm.

10. The data processing apparatus according to any one of claims 1-3, characterized in that, The generation of the first instruction, the adjustment of the parameters of the quantitative screening model, the update of the reassessment prediction, and the update of the rehabilitation control parameter set are configured to be executed periodically or triggered cyclically within the rehabilitation cycle to dynamically optimize the rehabilitation control parameter set according to the patient's real-time recovery status.

11. A data processing system, characterized in that, Includes the data processing apparatus as described in any one of claims 1 to 10.

Images