A walking aid matching recommendation system for brain disease complicated with lower limb movement disorder

Abstract

The application belongs to the technical field of medical information processing and computer aided decision, and relates to a walking aid equipment matching recommendation system for brain disease complicated with lower limb movement disorder, which aligns sensing data to generate a synchronous disturbance response data stream through a multi-modal response data processing module; a neural control gain calculation module extracts reflection characteristics and performs weighted calculation to generate a gain index; a dynamic mechanical impedance logic analysis module performs condition judgment to generate an impedance index; a pathological feature space mapping module maps the indexes into two-dimensional pathological feature coordinate points; an equipment matching rule association module generates a digital equipment matching zoning map based on a knowledge base; and a personalized recommendation generation module generates a recommendation report containing equipment types and parameter configurations based on a position matching degree and a displacement interpolation algorithm. The application solves the problem that traditional experience recommendation is difficult to quantitatively match the neural and double physical pathological mechanisms of patients by constructing an auxiliary decision model based on data analysis.

Description

Technical Field

[0001] This invention belongs to the technical field of medical information processing and computer-aided decision-making, and relates to an assistive device matching and recommendation system for brain diseases complicated with lower limb motor disorders. Background Technology

[0002] Many lower limb motor disorders are caused by damage to the central nervous system and certain brain diseases. For example, damage to brain structures such as the amygdala, or brain diseases such as epilepsy, often leads to abnormalities in neural control pathways and their transmission of mechanical movement in the lower limbs, resulting in severe lower limb motor disorders (such as joint stiffness and hyperreflexia). Currently, the selection of assistive devices for these patients relies primarily on the physician's subjective experience and static assessments in clinical practice, such as gait observation and functional scale scores. However, this traditional approach has significant limitations in information technology: First, existing assessment indicators are mostly descriptive of external motor manifestations, lacking data models that can decouple and quantify the dynamic response of the central nervous system (such as the brain) to external disturbances from the mechanical impedance characteristics of the limbs themselves; second, existing data acquisition methods often only focus on data acquisition, lacking supporting algorithmic logic to process this data, and failing to map complex physiological parameters (such as neural control gains affected by brain diseases) into specific equipment selection criteria.

[0003] The lack of a standardized decision support system that can correlate the intrinsic neuromechanical pathological mechanisms of patients with brain diseases and other conditions with the functional attributes of assistive devices often makes it difficult to accurately match device recommendations to the specific pathological deficiencies of patients, such as abnormal neurological reflexes or insufficient mechanical stiffness. Therefore, how to utilize information processing technology to construct a quantitative physiological characteristic model and achieve automated and refined device recommendations based on knowledge base rules is a pressing issue that needs to be addressed in the field of rehabilitation assistive device fitting. Summary of the Invention

[0004] To address the aforementioned problems, this invention provides an assistive device matching and recommendation system for brain diseases complicated by lower limb motor disorders.

[0005] A system for matching and recommending assistive devices for patients with brain diseases complicated by lower limb motor dysfunction, comprising: The multimodal response data processing module is configured to receive sensing data of the target limb after it is physically disturbed at a preset gait phase point, and to perform time axis alignment processing on the data to generate a synchronous disturbance response data stream containing electromyographic activity signals and motion posture signals. The neural control gain calculation module is configured to extract reflection signal features from the synchronous perturbation response data stream, and generate a neural control gain index that quantifies the responsiveness of the nervous system by weighting the gain of the reflection signal relative to the baseline data. The dynamic mechanical impedance logic analysis module is configured to execute condition judgment logic, compare the neural control gain index with a preset threshold, and selectively activate the passive stiffness analysis algorithm or the active reflection stiffness analysis algorithm based on the comparison result, and generate the dynamic mechanical impedance index using the synchronous disturbance response data stream. The pathological feature spatial mapping module is configured to construct a two-dimensional data coordinate system with dynamic mechanical impedance index and neural control gain index as dimensions, and map the calculated index data to two-dimensional pathological feature coordinate points in this coordinate system. The device matching rule association module is configured to retrieve a pre-stored knowledge base of assistive devices and pathological features, define multiple non-overlapping device recommendation areas in the coordinate system based on the rules in the knowledge base, and generate a digital device matching partition map. The personalized recommendation generation module is configured to retrieve the position of two-dimensional pathological feature coordinate points in the device matching partition map. By calculating the geometric matching degree and position offset of the coordinate points relative to the center of the region, it generates a personalized mobility aid recommendation report containing device type and parameter interpolation configuration suggestions.

[0006] In a further embodiment of the present invention, the multimodal response data processing module performs the following steps when generating the synchronization disturbance response data stream: Gait cycles are monitored and preset gait phase points are identified through sensor data streams; When a gait phase point is detected, a physical disturbance execution command is triggered and a synchronization time window is set; Within the synchronous time window, electromyographic activity data, joint angle change data, and reaction force data are acquired in parallel. Using the physical disturbance trigger time as the time reference, the acquired data is aligned to generate a synchronization disturbance response data stream.

[0007] In a further embodiment of the present invention, the neural control gain calculation module performs the following calculation steps: Preprocess the electromyographic activity data in the synchronous perturbation response data stream to obtain the envelope; Within the preset short latency time window and long latency time window, the short latency reflection signal and the long latency reflection signal are identified and extracted respectively. Calculate the reflection gain ratio of each reflected signal relative to the baseline data before the disturbance; The neural control gain index is calculated and generated based on a preset weighting model by using the reflection gain ratio and the latency of the reflected signal.

[0008] In a further embodiment of the present invention, the dynamic mechanical impedance logic analysis module performs the following logic steps: The neural control gain index is numerically compared with the preset neural reflex threshold. If the index is below the threshold, the passive stiffness analysis algorithm is invoked to calculate the passive mechanical stiffness using the synchronous disturbance response data stream; If the index is not lower than the threshold, the active reflection stiffness analysis algorithm is invoked to calculate the active reflection stiffness using the synchronous disturbance response data stream; The calculation results are uniformly assigned as the dynamic mechanical impedance index.

[0009] In a further embodiment of the present invention, the pathological feature spatial mapping module performs the following steps: Establish a two-dimensional Cartesian coordinate system, defining the first coordinate axis as the dynamic mechanical impedance index axis and the second coordinate axis as the neural control gain index axis; Set the coordinate axis range to cover the preset range of human lower limb joint stiffness and nerve reflex data; Using the dynamic mechanical impedance index as the abscissa and the neural control gain index as the ordinate, two-dimensional pathological feature coordinate points are generated in the coordinate system.

[0010] In a further embodiment of the present invention, the device matching rule association module performs the following steps: Read the assisted mobility device-pathological feature knowledge base to obtain device recommendation rules that include the definition of indicator ranges; Based on the dynamic mechanical impedance index range and neural control gain index range defined by each rule, a geometric region is divided in the coordinate system as the device recommendation area. Each device recommendation area is data-bound with the corresponding mobility aid device type, function description, and recommendation priority information; All device recommendation areas are integrated to form a device matching partition map.

[0011] In a further embodiment of the present invention, the personalized recommendation generation module performs the following steps: Search the target device recommendation area containing two-dimensional pathological feature coordinate points in the device matching partition map; Calculate the matching degree value of the coordinate points of two-dimensional pathological features relative to the geometric center of the target device recommendation area; Based on the relative positions of two-dimensional pathological feature coordinate points within the target equipment recommendation area, configuration suggestions for key functional parameters of the equipment are generated using an interpolation algorithm. Integrate device information, matching scores, and configuration suggestions to generate a recommendation report.

[0012] A further aspect of the present invention involves calculating the reflection gain ratio, specifically including: The average amplitude of the reflected signal is obtained by calculating the ratio of the integral value of the reflected signal over the duration to the duration. The baseline average amplitude is obtained by calculating the ratio of the integral value of electromyographic activity during the stable period before the physical disturbance is applied to the duration of that stable period. The reflection gain is obtained by calculating the ratio of the average amplitude of the reflection to the average amplitude of the baseline.

[0013] A further aspect of the present invention, specifically including the calculation of passive mechanical stiffness or active reflection stiffness, comprises: Extract the reaction force data and joint angle change data within the corresponding analytical algorithm time window from the synchronous disturbance response data stream; Obtain the effective lever arm length of the target limb, and calculate the joint torque data using the reaction force data and the effective lever arm length; Obtain the peak change of joint torque data and the peak change of joint angle data within the time window; The absolute value of the ratio of the peak change in joint torque data to the peak change in joint angle data is used as the stiffness value.

[0014] A further aspect of the present invention includes calculating the matching degree value specifically as follows: Obtain the coordinates of the geometric center point of the target device recommendation area, as well as the width and height of the area in the two coordinate axis directions; Calculate the deviations of the coordinate points of the two-dimensional pathological features from the geometric center point on the two coordinate axes respectively; The deviation is normalized by using half of the width and half of the height respectively, to obtain the normalized absolute value of the deviation; The larger of the absolute values ​​of the normalized deviations corresponding to the two coordinate axes is selected as the regional comprehensive deviation. Calculate the difference between the value 1 and the regional comprehensive deviation to obtain the matching degree; if the difference is less than 0, the matching degree is recorded as 0.

[0015] In summary, the present invention has the following beneficial technical effects: 1. By analyzing the synchronous perturbation response data stream through the system's data processing module, the complex lower limb motor response can be decoupled into neural control gain indicators and dynamic mechanical impedance indicators through algorithms. This processing method enables the extraction of feature parameters from the mixed signal, providing a quantitative data foundation for subsequent computer-aided decision-making.

[0016] 2. By constructing a coordinate system with dynamic mechanical impedance and neural control gain as dimensions, and using a knowledge base to generate a device matching partition map, continuous physiological parameters are transformed into visualized spatial location information. This mechanism establishes a standardized classification logic based on geometric region determination, providing a clear mathematical basis for the device selection process.

[0017] 3. By calculating the geometric matching degree between feature coordinate points and the center of the recommendation area, parameter suggestions are automatically generated using an interpolation algorithm. This data operation based on relative position can not only recommend device types but also provide specific parameter configuration schemes, reducing subjective errors from human experience and significantly improving the matching efficiency between mobility aids and the patient's physiological state. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. The drawings are used to provide a further understanding of the present invention.

[0019] Figure 1 This discloses a schematic diagram of the framework in the embodiments of this application.

[0020] Figure 2 This discloses a flowchart of an embodiment of this application. Detailed Implementation

[0021] The following is in conjunction with the appendix Figure 1 - Figure 2 A preferred description of the present invention is provided below.

[0022] See attached document Figure 1 - Figure 2 This invention proposes an assistive device matching and recommendation system for lower limb motor disorders complicated by brain diseases, comprising the following modules: The multimodal response data processing module is configured to receive sensing data of the target limb after it is physically disturbed at a preset gait phase point, and to perform time axis alignment processing on the data to generate a synchronous disturbance response data stream containing electromyographic activity signals and motion posture signals. The neural control gain calculation module is configured to extract reflection signal features from the synchronous perturbation response data stream, and generate a neural control gain index that quantifies the responsiveness of the nervous system by weighting the gain of the reflection signal relative to the baseline data. The dynamic mechanical impedance logic analysis module is configured to execute condition judgment logic, compare the neural control gain index with a preset threshold, and selectively activate the passive stiffness analysis algorithm or the active reflection stiffness analysis algorithm based on the comparison result, and generate the dynamic mechanical impedance index using the synchronous disturbance response data stream. The pathological feature spatial mapping module is configured to construct a two-dimensional data coordinate system with dynamic mechanical impedance index and neural control gain index as dimensions, and map the calculated index data to two-dimensional pathological feature coordinate points in this coordinate system. The device matching rule association module is configured to retrieve a pre-stored knowledge base of assistive devices and pathological features, define multiple non-overlapping device recommendation areas in the coordinate system based on the rules in the knowledge base, and generate a digital device matching partition map. The personalized recommendation generation module is configured to retrieve the position of two-dimensional pathological feature coordinate points in the device matching partition map. By calculating the geometric matching degree and position offset of the coordinate points relative to the center of the region, it generates a personalized mobility aid recommendation report containing device type and parameter interpolation configuration suggestions.

[0023] In one embodiment of the present invention, the multimodal response data processing module performs the following steps when generating a synchronization disturbance response data stream: Gait cycles are monitored and preset gait phase points are identified through sensor data streams; When a gait phase point is detected, a physical disturbance execution command is triggered and a synchronization time window is set; Within the synchronous time window, electromyographic activity data, joint angle change data, and reaction force data are acquired in parallel. Using the physical disturbance trigger time as the time reference, the acquired data is aligned to generate a synchronization disturbance response data stream.

[0024] Specifically, this is implemented in a personalized mobility aid recommendation system comprised of a central processing unit, a wearable sensor array, and physical perturbation actuators. First, inertial measurement units and plantar pressure sensors deployed on the target limb of the patient with lower limb movement disorders collect and process real-time temporal data of angular velocity and plantar pressure. The gait phase recognition module built into the central processing unit analyzes the temporal data based on a finite state machine model. To ensure low latency and high accuracy in phase recognition, a preset refresh rate of 100 Hz-200 Hz is set for sampling and processing. This range is based on the fact that the normal human gait frequency is approximately 1 Hz-2 Hz. According to the Nyquist sampling theorem and real-time control requirements, a refresh rate above 100 Hz ensures that the phase delay is controlled within 10 ms, meeting the timeliness requirements of neural reflex assessment.

[0025] The system determines the current gait phase in real time. The preset gait phase point is a key time node selected based on clinical assessment needs. In this embodiment, the heel strike is preferred as the trigger point. The determination logic is as follows: when the pressure sensor reading in the heel area exceeds a preset threshold (e.g., 50 N / s) within Δt = 10 ms after the swing phase ends, and the foot-to-ground angle detected by the inertial measurement unit is in the 0°-15° range, heel strike is determined. This setting effectively filters out false touch signals during the foot swing process. When the module detects that the gait has entered the preset gait phase point, it generates a trigger command, which is divided into two paths: The first signal is transmitted via Bluetooth Low Energy to a physical perturbation actuator bound to the joint of the target limb. This actuator is preferably a linear resonant actuator or voice coil motor encapsulated within a brace. It drives the actuator to generate a physical perturbation with specific energy characteristics. The parameters of this physical perturbation are set as follows: waveform is a sinusoidal pulse, duration 50 ms-100 ms, preferably 80 ms, vibration frequency 80 Hz-120 Hz, preferably 100 Hz, and peak acceleration amplitude controlled between 0.5 g and 1.5 g. The parameters are set based on the following: 80 Hz-120 Hz is the sensitive frequency range for stimulating type Ia afferent fibers of the muscle spindle, effectively inducing the stretch reflex; while controlling the duration to within 100 ms and the amplitude to less than 1.5 g ensures that the stimulation intensity is below the threshold, meaning it can be perceived by the proprioceptor but is insufficient to cause macroscopic kinematic instability or fall risk in the limb, thus ensuring patient safety.

[0026] The second path serves as a synchronization timestamp, initiating a preset synchronization time window and triggering the synchronous acquisition of multimodal response signals. This synchronization time window is set from 50 ms before the disturbance occurs as a baseline reference to 200 ms after the disturbance occurs. The rationale is that short-latency spinal reflexes typically occur between 30 ms and 50 ms, long-latency cortical long-circuit reflexes occur between 50 ms and 100 ms, while early mechanical motor responses are typically delayed to 100 ms and 150 ms. A 200 ms window is sufficient to cover the complete neuromechanical response chain while avoiding interference from the next action cycle.

[0027] Within this synchronous time window, the system's data acquisition unit collects electromyographic activity data in parallel from surface electromyography (EMG) sensor arrays deployed on the surface of key muscle groups of the target limb, such as the tibialis anterior and gastrocnemius muscles, at a first preset sampling rate, such as 500 Hz-2000 Hz. It should be understood that key muscle groups specifically refer to the agonist and antagonist muscles that play a major role in the selected gait phase. For example, when the heel strikes, the focus is mainly on the tibialis anterior, which is responsible for controlling ankle dorsiflexion, and the gastrocnemius, which is responsible for plantar flexion cushioning. The setting of 500 Hz-2000 Hz is to preserve the high-frequency components in the EMG signal and prevent aliasing. At the same time, at a second preset sampling rate, such as 50 Hz-500 Hz, the system collects angle change data and reaction force data of the target joint from the inertial measurement unit and the plantar pressure sensor.

[0028] After the data acquisition task is completed, the central processing unit performs time alignment processing on the electromyographic activity data and motion posture sensor data according to the synchronization timestamp, for example, by using linear interpolation or timestamp matching algorithms, to ensure the precise correspondence of each data point on the time axis, and finally generates a structured multi-channel time series data set. This set is defined as a synchronized perturbation response data stream containing neural activity and mechanical motion information, and is temporarily stored in system memory for subsequent use.

[0029] The gait phase point is determined based on the zero-crossing point of the angular velocity output by the inertial measurement unit (IMU) or the abrupt change point of the plantar pressure sensor reading. Electromyographic (EMG) activity data consists of raw EMG signal sequences after bandpass filtering and notch filtering. Motion posture sensor data includes three-dimensional joint angle time-series data output by the IMU through an attitude calculation algorithm, and ground reaction force time-series data obtained by integrating the plantar pressure sensor array. The synchronization disturbance response data stream is organized into a data matrix, where each row represents a sampling time point and each column represents data from an independent sensor channel.

[0030] For example, suppose we are assessing ankle stability during walking in a patient with right-sided hemiplegia (i.e., lower limb motor dysfunction) due to brain disease (e.g., complex partial epilepsy secondary to amygdala sclerosis) causing damage to specific brain structures, leading to conduction abnormalities. Because the motor cortex and subcortical structures of the brain are chronically affected by abnormal discharges (such as epileptic seizures), this patient not only exhibits physical lower limb muscle weakness but also abnormal stretch reflexes due to a lack of descending brain structural regulation. For this complex neuromechanical pathological state, the system sets the preset gait phase point to the instant the right heel strikes the ground. When the pressure value monitored by the heel pressure sensor deployed in the insole of the patient's right foot jumps from a swing phase level below 5 N to 15 N, the heel strike event is determined to have occurred, and the current moment is recorded. Recorded as a synchronization timestamp. Immediately send a command to the linear resonant actuator fixed to the lateral side of the patient's right ankle joint, causing it to generate a vibration pulse with a frequency of 100 Hz and a duration of 80 ms. Simultaneously, initiate the... arrive A synchronization time window with a total duration of 250 ms was established. Within this time window, the following data were acquired in parallel: surface electromyography (EMG) activity data of the tibialis anterior and gastrocnemius muscles, collected at a sampling rate of 2000 Hz, forming two EMG data sequences of length 500; angle change data of the ankle joint in the sagittal plane, collected at a sampling rate of 200 Hz, forming a 50-length angle data sequence; and total plantar reaction force data, also collected at a sampling rate of 200 Hz, forming a 50-length force data sequence. Finally, these four sets of data sequences were compared with... The time offsets are aligned and integrated into a unified, multi-channel synchronous perturbation response data stream file, which contains the complete process of the neural and mechanical responses around the patient's ankle joint from 50 ms before the perturbation to 200 ms after the perturbation.

[0031] In one embodiment of the present invention, the neural control gain calculation module performs the following calculation steps: Preprocess the electromyographic activity data in the synchronous perturbation response data stream to obtain the envelope; Within the preset short latency time window and long latency time window, the short latency reflection signal and the long latency reflection signal are identified and extracted respectively. Calculate the reflection gain ratio of each reflected signal relative to the baseline data before the disturbance; The neural control gain index is calculated and generated based on a preset weighting model by using the reflection gain ratio and the latency of the reflected signal.

[0032] Specifically, the central processing unit (CPU) retrieves the synchronization perturbation response data stream from system memory and initiates the neural reflex feature analysis process. First, the processor extracts the electromyographic activity data column corresponding to the surface electromyography (EMG) sensor from the data matrix of the synchronization perturbation response data stream. To eliminate baseline drift and power frequency interference, the processor performs full-wave rectification on the extracted raw EMG signal and applies a fourth-order Butterworth low-pass filter for smoothing to obtain a clear EMG signal envelope. The cutoff frequency is adjustable from 10 Hz to 500 Hz, preferably 50 Hz to obtain a smooth envelope.

[0033] Subsequently, using the synchronization timestamp of the physical perturbation as a reference point, pre-defined short-latency and long-latency reflex analysis windows were defined on the electromyographic signal envelope. The short-latency window is typically set between 20 and 45 ms after the perturbation, primarily capturing monosynaptic or oligosynaptic reflex activity at the spinal cord level. The long-latency window is set between 50 and 100 ms after the perturbation, capturing signals involving more complex long-circulation pathways through the cerebral cortex and potentially involving subcortical brain structures such as the amygdala. At this stage, because brain diseases such as epilepsy significantly interfere with the normal acquisition and motor integration process of the brain, the long-latency reflex signals extracted by the system can accurately reflect the degree of damage to the descending motor control pathways of the lower limbs caused by brain diseases or structural abnormalities.

[0034] Within the short latency reflex analysis window, peak values ​​are searched, and the first significant peak identified is designated as the short latency reflex signal. Similarly, within the long latency reflex analysis window, long latency reflex signals are searched and designated. Specifically, the starting point of the reflex signal is defined as the moment when the value of the electromyographic signal envelope first continuously exceeds the sum of the baseline mean and three times the standard deviation.

[0035] Next, the calibrated short-latency and long-latency reflection signals are quantized, specifically including: calculating the amplitude of the peak of the two reflection signals relative to the baseline level before the disturbance; obtaining the integral value of each signal envelope by numerical integration over the reflection duration; and recording the time difference from the occurrence of the disturbance to the starting point of the reflection signal as the latency period.

[0036] Simultaneously, the integral value of electromyographic activity and the duration of this stable period are calculated within the synchronization time window and before the disturbance. Further, the ratio of the integral value of the reflex signal to the duration of the reflex (i.e., the average amplitude of the reflex), and the ratio of the integral value of the baseline electromyographic activity to the duration of the baseline (i.e., the average amplitude of the baseline) are calculated, and the ratio of these two is used to quantify the reflex gain. Finally, the calculated reflex gain and multiple parameters, including latency, are substituted into a pre-defined weighted fusion model to calculate and generate a single scalar value. This value is defined as a neural control gain index that quantifies the nervous system's ability to respond to external stimuli and is stored for subsequent steps.

[0037]

[0038]

[0039] In the above formula, The reflection gain represents any reflection, and the short latency reflection gain can be calculated separately. With long latency reflection gain . , , and These are the durations of the reflection analysis window and the baseline analysis window, respectively, which eliminate the bias caused by the varying lengths of the time windows and are used to normalize the baseline signal. This is the envelope function of the electromyographic signal obtained after rectification and smoothing filtering. The start and end time interval of the identified reflected signal. This is the stable time interval used to calculate baseline activity before the disturbance occurs; its value is determined by the synchronization time window defined above. This represents the final generated neural control gain index. and These are the latency durations of long-latency reflexes and short-latency reflexes, respectively, in milliseconds (ms). These are the preset dimensionless weighting coefficients. Weighting coefficients with the reciprocal dimension of time, unit: ms -1 The weighting coefficients are set based on the physiological significance of different muscle groups and the importance of each parameter in pathological assessment, using an expert knowledge base or historical data regression analysis. The weighting coefficients aim to comprehensively assess the rapid response capability and regulatory complexity of neural pathways. For example, assuming that long latency reflexes are more important in assessing neural control in the human brain due to structural brain lesions or brain diseases, weighting coefficients can be set... The long latency gain reflects cortical regulation affected by abnormal discharges (such as epilepsy) and has the highest weight. Short latency gain reflects spinal reflexes; latency, as a negative indicator, is set... A longer incubation period indicates a slower reaction time; These coefficients are stored in the system's configuration file, ensuring that the indicators positively reflect the agility and strength of the nervous system, and can be adjusted by clinicians according to specific application scenarios.

[0040] For example, continuing from the previous example, the central processing unit retrieves the synchronization perturbation response data stream and extracts the electromyographic data sequence of the tibialis anterior muscle. After full-wave rectification and low-pass filtering with a cutoff frequency of 10 Hz, a smooth electromyographic signal envelope is obtained. First, analyze the time period before the disturbance occurs. The baseline electromyographic activity integral value for this interval is calculated to be 0.0025 mV·s, and the baseline average amplitude is... mV.

[0041] Subsequently, in the short latency reflection analysis window Within s, the start of The latency of the reflected wave of s ms, which is within the time interval The signal integral value within s is mV·s. Therefore, the average amplitude of short-latency reflection is... mV. Therefore, the short latency reflection gain. .

[0042] Next, within the long latency reflectance analysis window, the response originating from... The reflected wave of s, for which data adjustments were made to optimize subsequent calculations, terminates at 10.585 s, with a latency duration of [missing information]. ms, the signal integral value within this interval is mV·s. The average amplitude of long-latency reflection is then... mV. Therefore, the long latency reflection gain. .

[0043] Finally, the preset weight coefficients are called. Substitute the values ​​into the formula to calculate the neural control gain index: Ultimately, a neural control gain index of 1.5 was generated. It's important to clarify that for a healthy human brain, this neural reflex pathway index typically remains in a low and stable range. However, for this patient with epilepsy and other brain diseases, the structural lesions in their brain lead to a decreased inhibitory capacity of the central nervous system on the peripheral nervous system, resulting in an abnormal amplification of the reflex. Therefore, this relatively high scalar value of 1.5 is essentially a precise quantification of the pathophysiological state within the brain (such as neural remodeling under brain disease conditions) in the lower limb neuromuscular response. Subsequently, the system associates this value with the patient ID and stores it in the processing cache.

[0044] In one embodiment of the present invention, the dynamic mechanical impedance logic analysis module performs the following logic steps: The neural control gain index is numerically compared with the preset neural reflex threshold. If the index is below the threshold, the passive stiffness analysis algorithm is invoked to calculate the passive mechanical stiffness using the synchronous disturbance response data stream; If the index is not lower than the threshold, the active reflection stiffness analysis algorithm is invoked to calculate the active reflection stiffness using the synchronous disturbance response data stream; The calculation results are uniformly assigned as the dynamic mechanical impedance index.

[0045] Specifically, after completing the calculation of the neural control gain index, the central processing unit immediately initiates the conditional analysis process of dynamic mechanical impedance characteristics. It reads the neural control gain index calculated in the previous round from the processing buffer and compares it with a preset neural reflex threshold. It should be noted that this neural reflex threshold is a critical value used to distinguish between active and sluggish responses in the nervous system, and its setting is based on the statistical distribution of neural control gain in a large-scale healthy population or a patient group with a specific pathological type. In this embodiment, based on the aforementioned weighting model, the preset neural reflex threshold is 0.2.

[0046] If the system determines that the neural control gain index is lower than the neural reflex threshold, this indicates a significant defect or sluggish response in the patient's neural reflex pathway. At this point, the system activates passive stiffness analysis logic. Under this logic, angle change data and reaction force data of the joints synchronized with the application of the disturbance are extracted from the synchronous disturbance response data stream. Particularly within the short latency reflex time window, the ratio of the peak change in joint torque generated by the reaction force to the peak change in joint angle within this time window is calculated. This ratio is used as the passive mechanical stiffness characterizing the limb's physical compliance.

[0047] Conversely, if the system determines that the neural control gain index is not lower than the neural reflex threshold, it indicates that the patient's nervous system still has the ability to regulate external disturbances. At this time, the system activates the active reflex stiffness analysis logic, which includes the contribution of the muscle contraction force effect driven by long-latency reflexes to the total joint stiffness, thereby assessing the dynamic response characteristics of the entire neuromuscular system. Under this logic, joint angle change data and reaction force data are also extracted, but the calculation window is aligned to the time period during which the long-latency reflex occurs in the electromyographic activity data. By calculating the ratio of the peak change in total reaction force generated by the combined action of neural reflexes to the peak change in joint angle during this time period, the active reflex stiffness reflecting the neuromuscular coupling characteristics is obtained.

[0048] Ultimately, regardless of the analytical logic executed by the system, the passive mechanical stiffness or active reflection stiffness calculated will be uniformly represented and stored as a new scalar value, namely the dynamic mechanical impedance index, for subsequent steps to perform two-dimensional feature mapping.

[0049] The joint torque is calculated by multiplying the reaction force by the effective lever arm length. The effective lever arm length can be preset based on the patient's foot size or dynamically calculated using the pressure center position obtained from a plantar pressure sensor. The stiffness calculation formula is as follows:

[0050]

[0051]

[0052] In the above formula, The calculated mechanical stiffness can be either passive mechanical stiffness. or active reflection stiffness . This represents the peak change in joint torque. This represents the change in peak reaction force measured after the disturbance. The effective lever arm length is the vertical distance from the point of application of the reaction force to the center of joint rotation. This represents the corresponding peak joint angle change. and These are the peak reaction force and peak joint angle detected within a specific analysis window, respectively. and These are the baseline reaction force and baseline joint angle measured moment before the disturbance occurs. Under the passive stiffness analysis logic, the peak value search window is synchronized with the short latency reflection analysis window; under the active reflection stiffness analysis logic, the peak value search window is synchronized with the long latency reflection analysis window.

[0053] For example, continuing from the previous steps, the calculated neural control gain index is read as 1.5. The internally preset neural reflex threshold is 0.2, which is derived from clinical database statistics and used to distinguish whether the neural response is basically normal. Since 1.5 is not lower than 0.2, the active reflex stiffness analysis logic is activated. Then, ankle joint angle change data and plantar reaction force data are retrieved from the synchronous perturbation response data stream. First, the baseline state is determined, and the baseline reaction force is read at t=10.499 s, just before the perturbation occurs. 15 N, baseline ankle angle The value is 5°. Next, the analysis window is locked on the time period during which the long latency reflection occurs, i.e. Within this window, the peak value of the reaction force was found to occur at t=10.580 s, with a value of N; simultaneously, the peak value of the joint angle also appears around this moment, with a value of The system calls the preset effective lever arm length for the patient. That is, 15 centimeters. Then, substitute the values ​​into the formula to calculate the active reflection stiffness: Nm / °. Finally, the calculated 20 Nm / ° was used as a dynamic mechanical impedance index, and stored together with the patient ID and the neural control gain index 1.5 to construct two-dimensional pathological features.

[0054] In one embodiment of the present invention, the pathological feature spatial mapping module performs the following steps: Establish a two-dimensional Cartesian coordinate system, defining the first coordinate axis as the dynamic mechanical impedance index axis and the second coordinate axis as the neural control gain index axis; Set the coordinate axis range to cover the preset range of human lower limb joint stiffness and nerve reflex data; Using the dynamic mechanical impedance index as the abscissa and the neural control gain index as the ordinate, two-dimensional pathological feature coordinate points are generated in the coordinate system.

[0055] Specifically, after generating the dynamic mechanical impedance index and the neural control gain index, the system's data processing module calls a graphics rendering engine or data visualization library to construct a two-dimensional pathological feature topology map for characterizing the pathological state. This process first logically establishes a two-dimensional Cartesian coordinate system in the system's allocated video memory or system memory.

[0056] The dynamic mechanical impedance index is designated as the first dimension of the coordinate system, usually set as the horizontal axis, i.e., the X-axis. Its range is set according to the stiffness range that the human lower limb joints may exhibit in healthy and various pathological states. For example, it can be set to 0-50 Nm / °, where the low value range of 0-10 Nm / ° represents limb softness or joint instability, the medium value range of 10 Nm / °-25 Nm / ° represents the normal reflex range, and the high value range of 25 Nm / °-50 Nm / ° represents joint stiffness or muscle spasm.

[0057] Meanwhile, the neural control gain index is designated as the second-dimensional coordinate axis, usually set as the vertical axis, i.e., the Y-axis. Its range covers the interval from severe inhibition to hyperactivity of neural reflexes. For example, it can be set to a dimensionless range of -5 to 20, where negative values ​​or close to 0 represent sluggish or absent neural response, such as peripheral nerve injury, 0.2-5 represents the normal reflex range, and 5-20 represents hyperreflexia, such as upper motor neuron syndrome.

[0058] Subsequently, the specific dynamic mechanical impedance index value calculated for the current patient in the previous steps is read from the processing buffer as the X-coordinate value, and the neural control gain index value is read as the Y-coordinate value. This pair of coordinate values ​​is then mapped in a two-dimensional Cartesian coordinate system. This involves locating and rendering a visual data point on the graphical interface, or generating a node object containing this coordinate information at the data structure level. The unique position of this data point in the topology map intuitively reflects the coupling state of the patient's neurological response and limb biomechanical characteristics under disturbance. This successfully mapped and generated data point is defined as a unique two-dimensional pathological feature coordinate point identifying the patient's current neuromechanical coupling state. The data structure of this coordinate point, along with its corresponding patient ID, is stored in the system's feature database for subsequent region localization and device matching.

[0059] The two-dimensional pathological feature coordinates are data structures containing two floating-point values, corresponding to their projections onto two coordinate axes, and are associated with the patient's unique identifier. The core of this step lies in transforming abstract physiological parameters into visualized spatial locations, laying the foundation for subsequent geometric region-based classification and matching.

[0060] For example, following the aforementioned steps, after completing the index calculation, the topology map construction process is initiated. First, a two-dimensional coordinate system is established, with the horizontal axis X representing the dynamic mechanical impedance index axis, displaying a range of 0 to 50 Nm / °; and the vertical axis Y representing the neural control gain index axis, displaying a range of -5 to 20. Next, the calculation results for the patient are read from the buffer, namely, a dynamic mechanical impedance index value of 20 Nm / ° and a neural control gain index value of 1.5. Subsequently, these values ​​are mapped as coordinates, generating a data point with coordinates (20, 1.5) in the two-dimensional coordinate system. This point located at coordinate (20, 1.5) is the generated two-dimensional pathological feature coordinate point for the patient. The horizontal coordinate value of 20 Nm / ° indicates that the patient's limb exhibits moderate mechanical resistance, while the vertical coordinate value of 1.5 indicates that its neural reflex gain is at a relatively low level compared to normal. This two-dimensional pathological feature coordinate point is saved by the system and used for the next step of device matching area localization.

[0061] In one embodiment of the present invention, the device matching rule association module performs the following steps: Read the assisted mobility device-pathological feature knowledge base to obtain device recommendation rules that include the definition of indicator ranges; Based on the dynamic mechanical impedance index range and neural control gain index range defined by each rule, a geometric region is divided in the coordinate system as the device recommendation area. Each device recommendation area is data-bound with the corresponding mobility aid device type, function description, and recommendation priority information; All device recommendation areas are integrated to form a device matching partition map.

[0062] Specifically, based on the constructed two-dimensional pathological feature topology map, a pre-configured region division and calibration process is executed to generate decision-making criteria for device matching. This process is not executed in real time for a single patient, but is completed during system initialization or knowledge base updates. First, a pre-defined assistive device-pathological feature knowledge base stored in non-volatile memory is accessed. This knowledge base is a structured database, pre-constructed and calibrated by domain experts based on extensive clinical case data, rehabilitation medicine theory, and device engineering parameters. Each record defines a device recommendation rule, including a region identifier, the corresponding assistive device type, functional description, recommendation priority, and the applicable ranges of dynamic mechanical impedance and neural control gain indices.

[0063] The system's data processing module reads each record in the knowledge base. For each record, it extracts the lower and upper limits of the defined dynamic mechanical impedance index and the lower and upper limits of the neural control gain index. These four boundary values ​​uniquely define rectangular or polygonal regions on the two-dimensional pathological feature topology map. This geometric region defined by the boundary parameters is then data-bound with its corresponding device type, functional description, and recommendation priority to form a device recommendation area. It should be noted that the device recommendation area is a logical partition on the topology map, and its boundaries are strictly defined by the quantitative index thresholds in the knowledge base. By traversing the entire knowledge base, multiple non-overlapping device recommendation areas are generated on the topology map. This non-overlapping characteristic is achieved by carefully designing the boundary values ​​of each region during knowledge base construction, avoiding the possibility of any coordinate point falling into multiple regions simultaneously, thus ensuring the uniqueness of the matching results. In one embodiment of the present invention, the following are specific examples of partitioning rules: Rule 1: For regions with high mechanical impedance and low neural gain, such as: At this time, the patient's joints are stiff and lack neuromodulation ability, and the corresponding device type is a high-strength, high-torque exoskeleton, which is recommended as a high priority.

[0064] Rule 2: For regions with low mechanical impedance and high neural gain, such as: At this time, the patient's joints are loose but accompanied by hyperreflexia. This condition is common in complications of brain diseases caused by damage to brain structure or frequent epileptic seizures (such as early spastic hemiplegia). The corresponding device type is an electrostimulation complex device with spasticity inhibition function. The recommendation is based on the fact that when the patient's lower limbs show high neural gain due to abnormal brain discharge (a characteristic of epilepsy), the external electrostimulation complex device can be used for inverse neural modulation to inhibit abnormal reflexes caused by brain diseases, rather than simply applying mechanical support.

[0065] Rule 3: Target regions with moderate mechanical impedance and moderate neural gain, such as: At this stage, the patient retains some residual function, and the corresponding device type is an on-demand assistive soft exoskeleton, which provides lightweight assistance without restricting residual movement.

[0066] By traversing the entire knowledge base, multiple non-overlapping device recommendation zones are generated on the topology graph. Finally, all these device recommendation zones, carrying complete boundary parameters and device association information, are integrated into a single, fast-querying data structure. This data structure is defined as the device matching partition graph and loaded into the system's cache for subsequent real-time matching steps. In engineering implementation, the device matching partition graph can be a spatial index structure such as a hash table or quadtree, using the coordinates of the two-dimensional pathological feature topology graph as keys, enabling efficient retrieval of the device recommendation zone where the coordinate point is located and its associated information.

[0067] A typical record in the mobility aid device-pathological feature knowledge base might be defined as follows: For patients with a dynamic mechanical impedance index higher than 30 Nm / ° and a neural control gain index lower than 0, a high-support, high-torque exoskeleton is recommended, with a high priority. The device matching partition map, in engineering implementation, can be a spatial index structure such as a hash table or quadtree. Using the coordinates of the two-dimensional pathological feature topology map as keys, it can efficiently retrieve the device recommendation area where the coordinate point is located and its associated information.

[0068] For example, following the steps described above, we begin loading the mobility aids-pathological feature knowledge base to construct a device matching partitioning map. Assume the knowledge base contains the following three core rules: Rule 1: For regions with high mechanical impedance and low neural gain, the boundaries are set as follows: dynamic mechanical impedance index X-axis range [30, 50] Nm / °, neural control gain index Y-axis range [-5, 0], corresponding to the device type of strong support high torque exoskeleton, with a recommended priority of 1.

[0069] Rule 2: For regions with low mechanical impedance and high neural gain, the boundaries are set as X-axis range [0, 15] and Y-axis range [5, 20]. The corresponding device type is an electrostimulation combination device with spasticity inhibition function, and the recommended priority is 1.

[0070] Rule 3: For regions with moderate mechanical impedance and moderate neural gain, the boundaries are set as X-axis range [15, 30) and Y-axis range [0, 5), the corresponding device type is on-demand assistive soft exoskeleton, and the recommended priority is 1.

[0071] First, rule one is processed, defining a rectangular region on the two-dimensional pathological feature topology map defined by coordinates (30, -5) and (50, 0), and binding it to the information of a high-torque, high-support exoskeleton. Next, rule two is processed, defining a region defined by coordinates (0, 5) and (15, 20), and binding it to the information of an electrostimulation composite device. Then, rule three is processed, defining a region defined by coordinates (15, 0) and (30, 5), and binding it to the information of an on-demand, assisted soft exoskeleton. After processing all rules, these three defined regions with complete attributes are combined to form a data list or index, i.e., a device matching partition map, and stored in memory, awaiting the next step of localization and retrieval.

[0072] In one embodiment of the present invention, the personalized recommendation generation module performs the following steps: Search the target device recommendation area containing two-dimensional pathological feature coordinate points in the device matching partition map; Calculate the matching degree value of the coordinate points of two-dimensional pathological features relative to the geometric center of the target device recommendation area; Based on the relative positions of two-dimensional pathological feature coordinate points within the target equipment recommendation area, configuration suggestions for key functional parameters of the equipment are generated using an interpolation algorithm. Integrate device information, matching scores, and configuration suggestions to generate a recommendation report.

[0073] Specifically, after the device matching partition map is generated and loaded, an automated localization and recommendation process is executed. First, the two-dimensional pathological feature coordinates of the current patient are retrieved from the feature database. Then, a retrieval algorithm is initiated, which takes the coordinates of the two-dimensional pathological feature coordinates as input and queries the spatial index structure defined by the device matching partition map. This query operation uniquely determines the device recommendation area to which the two-dimensional pathological feature coordinates fall by comparing the X and Y values ​​of the coordinates with the boundary parameters of each device recommendation area. After successfully locating the corresponding device recommendation area, the relative position of the coordinates within the area is further analyzed to quantify the accuracy of the recommendation. The geometric center point of the device recommendation area is calculated, and the matching degree is calculated based on the relative position between the center point and the two-dimensional pathological feature coordinates. Considering that the device recommendation area is a rectangular region, to ensure that all points within the region boundary have valid matching degree values, this embodiment uses an algorithm based on maximum axial deviation, approximating Chebyshev distance logic. It should be understood that this matching degree is a dimensionless scalar quantity used to visually demonstrate to clinicians the degree to which the recommended treatment plan matches the patient's current pathological characteristics. Its value is inversely proportional to distance; that is, the closer the coordinate point is to the center of the area, the higher the matching degree. Based on the matching degree, the key functional parameters of the mobility aids associated with that area are dynamically fine-tuned. Specifically, the recommended configuration of key functional parameters is calculated by interpolation based on the relative position of the coordinate point within the area. For example, if the value of the coordinate point on the dynamic mechanical impedance axis is too high, the system will recommend a higher initial assist torque value.

[0074] Finally, the device types and function descriptions contained in the located device recommendation area are integrated with the calculated matching degree and the key function parameter configuration suggestions after fine-tuning. Adjacent areas are searched to identify potential alternatives. Finally, this information is formatted to generate and output a structured personalized walking aid device recommendation report that includes the preferred walking aid device type, key function parameter configuration suggestions, and potential alternatives.

[0075] The matching degree is calculated using the following formula:

[0076] In the above formula, This represents the calculated matching degree, and its value ranges from 0 to 1. and The horizontal axis represents the two-dimensional pathological feature coordinate points, i.e., the dynamic mechanical impedance index; and the vertical axis represents the neural control gain index. and The x and y coordinates represent the geometric center point of the recommended area for the corresponding device. and These represent the width of the recommended area for the device along the X-axis and the height along the Y-axis, respectively. The formula calculates the absolute values ​​of the normalized deviations of the x and y coordinates separately and takes the maximum of the two as the basis for measuring the degree of deviation. This calculation method ensures that as long as the coordinate point is within the boundary of the rectangular area, i.e., the normalized deviations of both axes are no greater than 1, the matching degree calculation result is always greater than or equal to 0, avoiding the logical contradictions that occur when using Euclidean distance at the corners of the rectangle.

[0077] The retrieval algorithm can employ simple iterative comparison methods or more efficient spatial data structure queries, such as range queries based on quadtrees or R-trees. For rectangular regions, the region center is the intersection of its diagonals; for irregular polygonal regions, it is its geometric centroid. Potential alternatives are typically derived from the assistive devices corresponding to the regions adjacent to the current recommended device area and whose patient coordinates are closest to their boundaries, providing additional reference for clinical decision-making. The personalized assistive device recommendation report is a digital document that can be displayed or printed through the system interface. Its content clearly lists the various recommendation results, facilitating understanding and implementation by rehabilitation therapists or doctors.

[0078] For example, following the aforementioned steps, a recommendation report is generated for the patient. First, the patient's two-dimensional pathological feature coordinates are read as (20, 1.5). A search is performed in the device matching partition map. By comparing the coordinates, it is found that the point meets the conditions of X-axis range [15, 30) and Y-axis range [0, 5). Therefore, the device recommendation area to which it belongs is successfully located, which corresponds to the area of ​​rule three, that is, the recommended device is an on-demand assistive soft exoskeleton.

[0079] Next, calculate the geometric center point of the region, whose coordinates are... , ,Right now .

[0080] The width of the area ,high Therefore, the horizontal half-width is 7.5 and the vertical half-height is 2.5. Then, the corrected formula is used to calculate the matching degree: Calculate the absolute value of the transverse normalized bias: .

[0081] Calculate the absolute value of the longitudinal normalized bias: .

[0082] Determine the regional comprehensive deviation: Compare 0.333 and 0.4, and take the larger value of 0.4 as the regional comprehensive deviation.

[0083] Calculate the matching degree: .

[0084] Determine the matching degree as Based on this, the following key functional parameter configuration recommendations are generated: Since the patient's dynamic mechanical impedance index of 20 Nm / ° is located in the lower part of the region [15, 30], deviating from the center to the left, it is recommended that the initial assist torque be set to 33% of the device's usable range; since the neural control gain index of 1.5 is located in the lower part of the region [0, 5], deviating from the center below, it is recommended to disable or use low-intensity neural electrical stimulation modulation. For alternative solutions, the point (20, 1.5) is detected to be closest to the boundary of Y=0, and below this boundary is a region of high mechanical impedance and low neural gain. Therefore, the "strong support high torque exoskeleton" is listed as a potential alternative, but its matching priority is noted to be low. Finally, integrating the above information, a personalized mobility aid device recommendation report is output, the core content of which is: Preferred device: On-demand assistive soft exoskeleton; Matching degree: Recommended parameters: 33% auxiliary torque, neural modulation function off; Alternative option: high-torque exoskeleton with strong support.

[0085] Each of the modules can be implemented in whole or in part through software, hardware, or a combination thereof. It supports hardware embedded in or independent of the processor in the computer device, and also supports software stored in the memory of the computer device, so that the processor can call and execute the operations corresponding to each of the above modules.

[0086] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A system for matching and recommending assistive devices for patients with brain diseases complicated by lower limb motor dysfunction, characterized in that, include: The multimodal response data processing module is configured to receive sensing data of the target limb after it is physically disturbed at a preset gait phase point, and to perform time axis alignment processing on the data to generate a synchronous disturbance response data stream containing electromyographic activity signals and motion posture signals. The neural control gain calculation module is configured to extract reflection signal features from the synchronous perturbation response data stream, and generate a neural control gain index that quantifies the responsiveness of the nervous system by weighting the gain of the reflection signal relative to the baseline data. The dynamic mechanical impedance logic analysis module is configured to execute condition judgment logic, compare the neural control gain index with a preset threshold, and selectively activate the passive stiffness analysis algorithm or the active reflection stiffness analysis algorithm based on the comparison result, and generate the dynamic mechanical impedance index using the synchronous disturbance response data stream. The pathological feature spatial mapping module is configured to construct a two-dimensional data coordinate system with dynamic mechanical impedance index and neural control gain index as dimensions, and map the calculated index data to two-dimensional pathological feature coordinate points in this coordinate system. The device matching rule association module is configured to retrieve a pre-stored knowledge base of assistive devices and pathological features, define multiple non-overlapping device recommendation areas in the coordinate system based on the rules in the knowledge base, and generate a digital device matching partition map. The personalized recommendation generation module is configured to retrieve the position of two-dimensional pathological feature coordinate points in the device matching partition map. By calculating the geometric matching degree and position offset of the coordinate points relative to the center of the region, it generates a personalized mobility aid recommendation report containing device type and parameter interpolation configuration suggestions.

2. The assistive device matching and recommendation system for brain diseases complicated with lower limb motor dysfunction according to claim 1, characterized in that, When generating the synchronization disturbance response data stream, the multimodal response data processing module performs the following steps: Gait cycles are monitored and preset gait phase points are identified through sensor data streams; When a gait phase point is detected, a physical disturbance execution command is triggered and a synchronization time window is set; Within the synchronous time window, electromyographic activity data, joint angle change data, and reaction force data are acquired in parallel. Using the physical disturbance trigger time as the time reference, the acquired data is aligned to generate a synchronization disturbance response data stream.

3. The assistive device matching and recommendation system for lower limb motor dysfunction complicated by brain disease according to claim 1, characterized in that, The neural control gain calculation module performs the following calculation steps: Preprocess the electromyographic activity data in the synchronous perturbation response data stream to obtain the envelope; Within the preset short latency time window and long latency time window, the short latency reflection signal and the long latency reflection signal are identified and extracted respectively. Calculate the reflection gain ratio of each reflected signal relative to the baseline data before the disturbance; The neural control gain index is calculated and generated based on a preset weighting model by using the reflection gain ratio and the latency of the reflected signal.

4. The assistive device matching and recommendation system for lower limb motor dysfunction complicated by brain disease according to claim 1, characterized in that, The dynamic mechanical impedance logic analysis module executes the following logic steps: The neural control gain index is numerically compared with the preset neural reflex threshold. If the index is below the threshold, the passive stiffness analysis algorithm is invoked to calculate the passive mechanical stiffness using the synchronous disturbance response data stream; If the index is not lower than the threshold, the active reflection stiffness analysis algorithm is invoked to calculate the active reflection stiffness using the synchronous disturbance response data stream; The calculation results are uniformly assigned as the dynamic mechanical impedance index.

5. The assistive device matching and recommendation system for lower limb motor dysfunction complicated by brain disease according to claim 1, characterized in that, The pathological feature spatial mapping module performs the following steps: Establish a two-dimensional Cartesian coordinate system, defining the first coordinate axis as the dynamic mechanical impedance index axis and the second coordinate axis as the neural control gain index axis; Set the coordinate axis range to cover the preset range of human lower limb joint stiffness and nerve reflex data; Using the dynamic mechanical impedance index as the abscissa and the neural control gain index as the ordinate, two-dimensional pathological feature coordinate points are generated in the coordinate system.

6. The assistive device matching and recommendation system for lower limb motor dysfunction complicated by brain disease according to claim 1, characterized in that, The device matching rule association module performs the following steps: Read the assisted mobility device-pathological feature knowledge base to obtain device recommendation rules that include the definition of indicator ranges; Based on the dynamic mechanical impedance index range and neural control gain index range defined by each rule, a geometric region is divided in the coordinate system as the device recommendation area. Each device recommendation area is data-bound with the corresponding mobility aid device type, function description, and recommendation priority information; All device recommendation areas are integrated to form a device matching partition map.

7. The assistive device matching and recommendation system for brain diseases complicated with lower limb motor dysfunction according to claim 1, characterized in that, The personalized recommendation generation module performs the following steps: Search the target device recommendation area containing two-dimensional pathological feature coordinate points in the device matching partition map; Calculate the matching degree value of the coordinate points of two-dimensional pathological features relative to the geometric center of the target device recommendation area; Based on the relative positions of two-dimensional pathological feature coordinate points within the target equipment recommendation area, configuration suggestions for key functional parameters of the equipment are generated through interpolation algorithms. Integrate device information, matching scores, and configuration suggestions to generate a recommendation report.

8. The assistive device matching and recommendation system for brain diseases complicated with lower limb motor dysfunction according to claim 3, characterized in that, The calculation of the reflection gain ratio specifically includes: The average amplitude of the reflected signal is obtained by calculating the ratio of the integral value of the reflected signal over the duration to the duration. The baseline average amplitude is obtained by calculating the ratio of the integral value of electromyographic activity during the stable period before the physical disturbance is applied to the duration of that stable period. The reflection gain is obtained by calculating the ratio of the average amplitude of the reflection to the average amplitude of the baseline.

9. The assistive device matching and recommendation system for lower limb motor dysfunction complicated by brain disease according to claim 4, characterized in that, Calculating passive mechanical stiffness or active reflection stiffness specifically includes: Extract the reaction force data and joint angle change data within the corresponding analytical algorithm time window from the synchronous disturbance response data stream; Obtain the effective lever arm length of the target limb, and calculate the joint torque data using the reaction force data and the effective lever arm length; Obtain the peak change of joint torque data and the peak change of joint angle data within the time window; The absolute value of the ratio of the peak change in joint torque data to the peak change in joint angle data is used as the stiffness value.

10. The assistive device matching and recommendation system for lower limb motor dysfunction complicated by brain disease according to claim 7, characterized in that, The calculation of the matching degree specifically includes: Obtain the coordinates of the geometric center point of the target device recommendation area, as well as the width and height of the area in the two coordinate axis directions; Calculate the deviations of the coordinate points of the two-dimensional pathological features from the geometric center point on the two coordinate axes respectively; The deviation is normalized by using half of the width and half of the height respectively, to obtain the normalized absolute value of the deviation; The larger of the absolute values ​​of the normalized deviations corresponding to the two coordinate axes is selected as the regional comprehensive deviation. Calculate the difference between the value 1 and the regional comprehensive deviation to obtain the matching degree; if the difference is less than 0, the matching degree is recorded as 0.

Images