Multimodal fusion system and wearable device for muscle assessment and early intervention

Abstract

The application discloses a multi-modal fusion system and wearable device for muscle evaluation and early intervention, comprising: a data acquisition module for acquiring biomechanical signals and recovery period physiological signals at the end of muscle movement, and for acquiring multi-modal physiological signals; a trend analysis module for extracting muscle function recovery indicators from the recovery period physiological signals, establishing a muscle function baseline based on the muscle function recovery indicators, and generating a decline warning information according to the muscle function baseline and the current muscle function recovery indicators; an evaluation module for obtaining a cause classification result of low muscle strength; an intervention generation module for generating a training task according to the cause classification result; and a feedback output module for acquiring training feedback signals when a user performs the training task, and generating and outputting action explanation instructions. The application can realize early detection and intervention of sarcopenia, accurately output a cause classification result, and output a targeted training task through the cause classification result.

Description

Technical Field

[0001] This application relates to the field of medical technology, and in particular to a multimodal fusion system and wearable device for muscle assessment and early intervention. Background Technology

[0002] Sarcopenia is an age-related, progressive, systemic loss of skeletal muscle mass and function, which can significantly increase the risk of falls, fractures, disability, and even death. Early identification of muscle weakness and targeted intervention are key strategies to slow the progression of sarcopenia and shorten the period of disability in older adults.

[0003] The current clinical diagnosis and treatment system for sarcopenia is mainly based on internationally recognized diagnostic guidelines, and its technical approach exhibits the following characteristics:

[0004] Current community screening methods mostly use a single grip strength test, which outputs a suspicious or normal label by comparing the result with a fixed threshold (e.g., men <28kg, women <18kg). This method can only identify patients who have entered the stage of strength decline. A large number of elderly people whose grip strength is still within the normal range but who have already experienced rapid functional decline are passively waiting because they do not meet the referral criteria, thus missing the best intervention window.

[0005] In addition, after clinical diagnosis of sarcopenia, no technical means are provided to differentiate the etiology. Patients who also present with low grip strength may have functional impairments due to different physiological mechanisms such as decreased neural recruitment efficiency, insufficient muscle endurance, and limited cardiopulmonary support for exercise load. Current technology cannot complete etiological identification in out-of-hospital settings, resulting in highly homogenized intervention programs that lack targeting. Summary of the Invention

[0006] This application aims to at least partially solve one of the technical problems in the aforementioned technologies.

[0007] To achieve the above objectives, the first aspect of this application proposes a multimodal fusion system for muscle assessment and early intervention, comprising: a data acquisition module for acquiring biomechanical signals characterizing a user's muscle load, acquiring physiological signals during the recovery period at the end of a muscle movement, and acquiring multimodal physiological signals during a user assessment test according to an active assessment instruction; a trend analysis module for identifying the end time of the muscle movement based on the biomechanical signals, extracting muscle function recovery indicators from the physiological signals during the recovery period, establishing a muscle function baseline based on the temporal changes of the muscle function recovery indicators, and generating deterioration warning information based on the muscle function baseline and the current muscle function recovery indicators; an assessment module for receiving the active assessment instruction triggered by the user after the deterioration warning information is output, analyzing the multimodal physiological signals to obtain an etiological classification result of muscle weakness; an intervention generation module for generating a training task based on the etiological classification result; and a feedback output module for acquiring training feedback signals when the user performs the training task, and generating and outputting movement correction instructions based on the training feedback signals.

[0008] In addition, the multimodal fusion system and wearable device for muscle assessment and early intervention proposed in this application may also have the following additional technical features:

[0009] As a further description of the above technical solution: the data acquisition module includes: a mechanical sensing unit, used to acquire the pressure value generated on the skin when the wrist flexor tendon contracts through a flexible tactile sensor array embedded in the inner side of the wearable device strap, and to detect the end of the muscle movement based on the rate of change of the pressure value; the mechanical sensing unit is also used to convert the pressure value into an equivalent grip force value through a mapping relationship; a recovery period physiological signal acquisition unit, used to continuously acquire surface electromyography (EMG) signals of the forearm flexor muscle region for a preset duration after the end of the detected muscle movement, to constitute the recovery period physiological signal; and a multimodal physiological sensing unit, used to simultaneously activate the surface EMG electrodes, photoplethysmography (PPG) sensor, and blood oxygen sensor built into the wearable device according to the active assessment command, to acquire the multimodal physiological signal; wherein, the multimodal physiological signal includes the surface EMG signal acquired by the surface EMG electrodes, the PPG signal acquired by the PPG sensor, and the blood oxygen saturation signal acquired by the blood oxygen sensor.

[0010] As a further description of the above technical solution: the trend analysis module includes: an event detection unit, used to identify the moment when the muscle movement ends based on the biomechanical signal, wherein the moment when the muscle movement ends is the moment when the pressure value drops from above a first pressure threshold to below a second pressure threshold; a feature extraction unit, used to extract muscle function recovery indicators from the physiological signals during the recovery period, wherein the muscle function recovery indicator is the recovery time required for the root mean square amplitude of electromyography to decay by 50% from the termination moment; a baseline establishment unit, used to store all the muscle function recovery indicators within a user sliding time window, and calculate a first statistical feature value within the sliding time window as the muscle function baseline; and an early warning determination unit, used to calculate a second statistical feature value for all the muscle function recovery indicators within the current monitoring period, and calculate the ratio of the second statistical feature value to the muscle function baseline, wherein when the ratio is greater than a preset threshold for a continuous preset duration, the decline early warning information is generated.

[0011] As a further description of the above technical solution: both the first statistical characteristic value and the second statistical characteristic value are medians or percentiles.

[0012] As a further description of the above technical solution: the evaluation module includes: an index generation unit, used to analyze the multimodal physiological signals to obtain a neural recruitment efficiency index, a muscle endurance index, and a cardiopulmonary support index; and an attribution classification unit, used to logically combine and determine the neural recruitment efficiency index, the muscle endurance index, and the cardiopulmonary support index, and output the etiological classification result.

[0013] As a further description of the above technical solution: the analysis of the multimodal physiological signals to obtain the neural recruitment efficiency index, muscle endurance index, and cardiopulmonary support index includes: extracting the root mean square amplitude at the time of synchronous sampling with the equivalent grip strength value from the surface electromyography signal, calculating the ratio of the root mean square amplitude to the equivalent grip strength value, and using the ratio as the neural recruitment efficiency index; extracting the decline slope of the median frequency during the evaluation test from the surface electromyography signal as the muscle endurance index; extracting the heart rate change from the photoplethysmography signal, extracting the decline slope of blood oxygen saturation during the evaluation test from the blood oxygen saturation signal, and extracting the grip strength change from the equivalent grip strength value, and generating the cardiopulmonary support index based on the ratio of the heart rate change to the grip strength change and the decline slope of blood oxygen saturation.

[0014] As a further description of the above technical solution: the logical combination determination, outputting the etiological classification result includes: if the neural recruitment efficiency index is lower than the efficiency threshold, the etiological classification result is poor neuromuscular control; if the muscle endurance index is lower than the endurance threshold and the neural recruitment efficiency index is not lower than the efficiency threshold, the etiological classification result is insufficient muscle endurance; if the cardiopulmonary support index is lower than the support threshold and the neural recruitment efficiency index is not lower than the efficiency threshold and the muscle endurance index is not lower than the endurance threshold, the etiological classification result is cardiopulmonary function limitation.

[0015] As a further description of the above technical solution: the intervention generation module is used to call the training task from the preset training library according to the etiological classification result, wherein the training library includes control training tasks corresponding to poor neuromuscular control; repetition training tasks corresponding to insufficient muscle endurance; and intermittent training tasks corresponding to cardiopulmonary function limitations.

[0016] As a further description of the above technical solution: the feedback output module includes: a signal acquisition unit, used to acquire the root mean square amplitude of surface electromyography and the equivalent grip force value in real time when the user performs the training task, as the training feedback signal; and a movement correction unit, used to calculate the ratio of the equivalent grip force value to the root mean square amplitude of surface electromyography at the same moment, generate a force efficiency, generate a force efficiency score based on the force efficiency, compare the force efficiency score with a preset quality threshold, and generate and output a movement explanation instruction when the force efficiency score is lower than the preset quality threshold.

[0017] According to the multimodal fusion system and wearable device for muscle assessment and early intervention proposed in this application, a baseline of muscle function is established by using electromyography recovery time collected in daily, imperceptible data. This changes the diagnosis from a single-point threshold to a continuous trend warning, enabling early detection and early intervention of sarcopenia. Through active assessment, neural recruitment efficiency index, muscle endurance index, and cardiopulmonary support index can be obtained. Through logical combination and judgment, three etiological classification results can be accurately output. Furthermore, targeted training tasks can be output based on the etiological classification results.

[0018] To achieve the above objectives, a second aspect of this application proposes a wearable device including a multimodal fusion system for muscle assessment and early intervention.

[0019] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0020] 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:

[0021] Figure 1 This is a system block diagram of a multimodal fusion system for muscle assessment and early intervention according to an embodiment of this application;

[0022] Figure 2 This is a flowchart of a multimodal fusion system for muscle assessment and early intervention according to an embodiment of this application;

[0023] Figure 3 This is a flowchart of the etiological typing results output according to one embodiment of this application. Detailed Implementation

[0024] The embodiments of this application are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0025] The following description, in conjunction with the accompanying drawings, describes a multimodal fusion system and wearable device for muscle assessment and early intervention according to embodiments of this application.

[0026] like Figures 1 to 3 As shown in the embodiments of this application, the multimodal fusion system and wearable device for muscle assessment and early intervention may include a data acquisition module for acquiring biomechanical signals characterizing the user's muscle load, acquiring physiological signals during the recovery period at the end of muscle movement, and acquiring multimodal physiological signals during the user's assessment test process according to active assessment instructions.

[0027] The trend analysis module is used to identify the end time of muscle movement based on biomechanical signals, extract muscle function recovery indicators from physiological signals during the recovery period, establish a muscle function baseline based on the temporal changes of muscle function recovery indicators, and generate degeneration early warning information based on the muscle function baseline and the current muscle function recovery indicators.

[0028] The assessment module is used to receive active assessment instructions triggered by the user after the decline warning information is output, so as to analyze multimodal physiological signals and obtain etiological classification results of muscle weakness.

[0029] The intervention generation module is used to generate training tasks based on etiological classification results.

[0030] The feedback output module is used to collect training feedback signals when the user performs training tasks, and to generate and output action correction instructions based on the training feedback signals.

[0031] It should be noted that in this application, the data acquisition module collects physiological signals during the recovery period after muscle movement in everyday life scenarios without being noticed; the trend analysis module extracts muscle function recovery indicators and establishes an individualized functional baseline based on these signals, upgrading the traditional single-point threshold diagnosis to a continuous trend warning based on the degree of baseline deviation, thereby triggering early intervention when strength is still within the normal range but has already shown continuous decline.

[0032] When users receive a decline warning, they can proactively conduct an assessment. The assessment module analyzes multimodal physiological signals, quantifies functional indices in three dimensions: neural recruitment efficiency, muscle endurance, and cardiopulmonary support, and outputs etiological classification results through logical combination. This enables accurate identification of the cause of muscle weakness, thus avoiding the problem of homogenization in subsequent intervention programs.

[0033] In addition, the intervention generation module generates targeted training tasks based on etiological subtypes, so that the rehabilitation plan is accurately matched with the type of physiological dysfunction; the feedback output module collects training feedback signals in real time during the execution of training tasks and outputs movement correction instructions, forming a closed-loop training mechanism.

[0034] Therefore, this application constructs a complete technical solution from early warning and etiological identification to targeted intervention and real-time correction, effectively promoting the improvement of sarcopenia management from passive diagnosis to proactive and precise intervention.

[0035] It should be noted that, in this embodiment of the application, the evaluation module is used to receive an active evaluation instruction triggered by the user after the decline warning information is output, so as to analyze the multimodal physiological signals and obtain the etiological classification results of muscle weakness. The user-initiated method here can be a variety of user-initiated methods such as manual click, voice trigger, and remote trigger by the supporting APP.

[0036] Understandably, this application transforms the electromyography recovery time index collected in daily, imperceptible data into a personalized muscle function baseline that can be tracked for a long time. It also enables early warning of sarcopenia decline through baseline deviation, achieving a qualitative leap from "single muscle fatigue assessment" to "continuous trend monitoring of early sarcopenia". This solves the core technical pain point of existing technologies, which "can only identify late-stage patients with single-point fixed threshold diagnosis, missing the best intervention window".

[0037] In one embodiment of this application, the data acquisition module includes: a mechanical sensing unit, used to acquire the pressure value generated on the skin when the wrist flexor tendon contracts through a flexible tactile sensor array embedded in the inner side of the wearable device strap, and to detect the end of the muscle movement based on the rate of change of the pressure value. The mechanical sensing unit is also used to convert the pressure value into an equivalent grip force value through a mapping relationship.

[0038] It should be noted that the mechanical sensing unit continuously collects pressure values. Real-time monitoring The rate of change, when From above the first pressure threshold Rapidly dropped below the second pressure threshold When a muscle movement is considered complete, this condition can be expressed as:

[0039]

[0040] in, For the rate of change, The rate of change threshold, This refers to the interval time.

[0041] It is understandable that the collected pressure values It is converted into an equivalent grip strength value through a preset linear mapping relationship. The formula is as follows: ,in: These are calibration coefficients, obtained by calibrating the wearable device using a standard hand grip dynamometer, used to eliminate dimensional differences between pressure and grip strength. This is a correction constant used to compensate for systematic errors caused by installation position deviations.

[0042] The recovery period physiological signal acquisition unit is used to continuously acquire surface electromyographic signals of the forearm flexor muscle area for a preset duration (e.g., 3 seconds, 5 seconds, 10 seconds, 15 seconds, etc.) after the muscle movement is detected to complete, so as to constitute the recovery period physiological signal.

[0043] The multimodal physiological sensing unit is used to simultaneously activate the surface electromyography electrodes, photoplethysmography pulse wave sensor, and blood oxygen sensor built into the wearable device according to the active assessment command, so as to collect multimodal physiological signals.

[0044] The multimodal physiological signals include surface electromyography signals acquired by surface electromyography electrodes, photoplethysmography signals acquired by photoplethysmography pulse wave sensors, and blood oxygen saturation signals acquired by blood oxygen sensors.

[0045] It should be noted that surface electromyography (EMG), photoplethysmography (PPG), and blood oxygen saturation signals are recorded synchronously with time stamp alignment, forming a complete set of multimodal physiological signals. This means that all sensors share the same high-precision clock source, and sampling trigger commands are synchronously issued by the main control unit, ensuring that the absolute error of the timestamps of the multi-channel signals is ≤1ms. For signals with different sampling rates, a linear interpolation resampling method is used to align all signals to the same time axis, eliminating synchronization errors caused by differences in sampling rates.

[0046] In addition, to further ensure the accuracy of the collected data, it is necessary to preprocess the surface electromyography (EMG) signal, photoplethysmography (PPG) signal, and blood oxygen saturation signal. Specifically, for the EMG signal, a 20-500Hz Butterworth bandpass filter and a 50Hz power frequency notch filter are used. Motion artifacts are identified using the sliding window threshold method, and artifact segments are replaced by linear interpolation of adjacent valid data. For the PPG and blood oxygen signals, a 0.5-8Hz bandpass filter is used, baseline drift is removed using the moving average method, and motion artifacts are suppressed using an adaptive filtering algorithm to ensure that the effective signal-to-noise ratio is ≥20dB.

[0047] In one embodiment of this application, the trend analysis module includes: an event detection unit, used to identify the moment when muscle movement ends based on biomechanical signals. The moment when the muscle movement ends The pressure value at the end of the muscle movement is higher than the first pressure threshold. Drop below the second pressure threshold At that moment.

[0048] The feature extraction unit is used to extract muscle function recovery indicators from physiological signals during the recovery period. The muscle function recovery indicator is the recovery time required for the root mean square amplitude of electromyography to decrease by 50% from the termination time. That is, feature extraction unit calculation Root mean square amplitude of electromyography at time tw The calculation formula is:

[0049]

[0050] Where G is The number of sampling points near the given time (a preset fixed value, such as 50), where Δt is the sampling interval. for Time of the first The amplitude of surface electromyography signal at each sampling point.

[0051] Then determine the amplitude threshold at which the root mean square amplitude of electromyography (EMG) decays by 50% (denoted as ). ),Right now .

[0052] It should be noted that an amplitude threshold of 50% is a common choice in this field. From a physiological perspective, the time point at which the electromyographic amplitude decays by 50% after muscle contraction ends corresponds precisely to the core cycle of skeletal muscle cells recovering from the excitation-contraction coupling state to the resting state. This can stably reflect the muscle's energy metabolism and neural recruitment recovery ability. Compared with other ratios, an amplitude threshold of 50% can better reflect the true physiological data feedback.

[0053] Find the condition RMS(t) = The time (denoted as) (Then muscle function recovery index) for: .

[0054] The baseline establishment unit is used to store all muscle function recovery indicators within a user's sliding time window (denoted as L, e.g., 30 days). Where M represents the number of valid data acquisitions within the sliding time window, and the first statistical characteristic value within the sliding time window is calculated as the baseline of muscle function, where the first statistical characteristic value is the median, i.e.

[0055]

[0056] Here, B represents the user's baseline muscle function.

[0057] The early warning judgment unit is used to calculate the second statistical characteristic value of all muscle function recovery indicators within the current monitoring period (denoted as W, e.g., 7 days), which can also be expressed using the median.

[0058]

[0059] Where H is the number of samples in the monitoring period, This is the second statistical characteristic value.

[0060] And calculate the ratio of the second statistical characteristic value to the baseline of muscle function. ,Right now

[0061]

[0062] When the ratio The value is greater than the preset threshold for a continuous preset duration D (e.g., 5 days or 7 days). If the value is 1.3 (for example), it is determined that the user's muscle function is in a continuous decline, and a decline warning message is generated.

[0063] For clarity, in the embodiments of this application, the first statistical characteristic value and the second statistical characteristic value may use percentiles in addition to the median.

[0064] In one embodiment of this application, the evaluation module includes an index generation unit for analyzing multimodal physiological signals to obtain neural recruitment efficiency index, muscle endurance index, and cardiopulmonary support index.

[0065] Specifically, the neural recruitment efficiency index, muscle endurance index, and cardiopulmonary support index are obtained by extracting the root mean square amplitude at the time of synchronous sampling with the equivalent grip force value from the surface electromyography signal, calculating the ratio of the root mean square amplitude to the equivalent grip force value, and using this ratio as the neural recruitment efficiency index. The formula is as follows:

[0066]

[0067] in, To evaluate the root mean square amplitude of the surface electromyography signal during the testing period, To assess the average equivalent grip strength during the testing period, a lower N value indicates lower neural recruitment efficiency and a greater likelihood that muscle dysfunction stems from poor neuromuscular control.

[0068] The slope of the median frequency decline during the assessment test was extracted from the surface electromyography signal as a muscle endurance index. This is used to characterize the endurance level of sustained muscle contraction, and its formula is as follows:

[0069]

[0070] in, To test the median frequency of the surface electromyography signal at the initial moment, To evaluate the end time of the test (t= The median frequency of surface electromyography (EMG) signals is the Q value. The lower the Q value, the worse the muscle endurance, and the more likely the muscle dysfunction is to be due to insufficient muscle endurance.

[0071] Heart rate variability is extracted from photoplethysmography (PPG) signals, the rate of decrease in blood oxygen saturation during the assessment test is extracted from blood oxygen saturation signals, and grip strength variability is extracted from equivalent grip strength values. Based on the ratio of heart rate variability to grip strength variability and the rate of decrease in blood oxygen saturation, a cardiopulmonary support index is generated. This is used to characterize the cardiopulmonary function's ability to support muscle exercise load. It is calculated by combining changes in heart rate, blood oxygenation, and grip strength. The formula is as follows:

[0072]

[0073] in, To assess the change in heart rate during the test, it was extracted from the photoplethysmography (PPG) signal; To assess the change in grip strength during the test; To assess the change in blood oxygen saturation during the test; , Weighting coefficients ( + =1, a preset fixed value, such as =0.6, =0.4), used to balance the influence of the two indicators; The absolute value of the slope of the decrease in blood oxygen saturation is used to eliminate the influence of the negative sign and make the indicator positively correlated with cardiopulmonary support capacity. and These are the first and second dimension conversion coefficients, used to unify the dimensions of each item and scale the calculation results on both sides to the same numerical order of magnitude (e.g., =0.003、 =15); The lower the value, the weaker the cardiopulmonary function's ability to support exercise load, and the more likely muscle dysfunction is to stem from cardiopulmonary function limitations.

[0074] The attribution classification unit is used to logically combine and determine the neural recruitment efficiency index N, muscle endurance index M, and cardiopulmonary support index C, and output the etiological classification results.

[0075] Specifically: If the neural recruitment efficiency index N is lower than the efficiency threshold That is, N < This indicates a decrease in neural recruitment efficiency, and the etiological classification result is poor neuromuscular control.

[0076] If the muscle endurance index Q is lower than the endurance threshold Furthermore, the neural recruitment efficiency index N is not lower than the efficiency threshold. That is, Q < And N≥ If the etiological classification result is insufficient muscle endurance, then the etiological classification result is insufficient muscle endurance.

[0077] If the cardiopulmonary support index C is below the support threshold Furthermore, the neural recruitment efficiency index N is not lower than the efficiency threshold. Furthermore, the muscle endurance index Q is not lower than the endurance threshold. That is, C < And N≥ And Q≥ If the etiological classification result is cardiopulmonary function limitation, then the etiological classification result is cardiopulmonary function limitation.

[0078] In one embodiment of this application, the intervention generation module is used to call training tasks from a preset training library based on etiological classification results.

[0079] The training library includes control training tasks corresponding to neuromuscular insufficiency, such as precision grip strength control training and fine motor finger training. The core purpose is to improve the efficiency of nerve recruitment of muscles and improve neuromuscular control ability.

[0080] Training tasks that address insufficient muscle endurance, such as continuous grip strength training and segmented endurance training, aim to improve the muscle's ability to contract continuously and thus enhance muscle endurance.

[0081] Intermittent training tasks that correspond to cardiorespiratory limitations, such as intermittent grip strength training and rhythmic breathing training, aim to improve the cardiorespiratory capacity to support exercise load while maintaining muscle strength.

[0082] It should be noted that the difficulty, duration, and frequency of training tasks can be dynamically adjusted according to the user's age and basic muscle strength level to ensure the safety and effectiveness of training and to suit the physical characteristics of the elderly population.

[0083] In one embodiment of this application, the feedback output module includes a signal acquisition unit, used to acquire the root mean square amplitude of surface electromyography and the equivalent grip force value in real time when the user performs a training task, as a training feedback signal.

[0084] The motion correction unit is used to calculate the ratio of the equivalent grip force value to the root mean square amplitude of surface electromyography at the same moment, generate the force efficiency E, and generate the force efficiency score S based on the force efficiency E. The force efficiency score is compared with a preset quality threshold. When the force efficiency score is lower than the preset quality threshold, the unit generates and outputs the motion explanation instruction.

[0085] It should be noted that the power efficiency score S (out of 100) is generated based on the power efficiency E, and the scoring formula is as follows:

[0086]

[0087] in, This is the preset optimal force efficiency. This represents the lowest effective force efficiency.

[0088] Compare the force efficiency score S with the preset quality threshold (For example, 60 points) compare, if S < If the user's movements are incorrect or their force application is inefficient, then a movement explanation instruction will be generated and output. This instruction is a preset instruction that can guide the user to adjust their movements, ensure training effectiveness, and avoid muscle damage caused by incorrect movements.

[0089] It should be noted that, in order to adapt to the physiological differences of users of different ages, genders and physical conditions, and to reduce the risk of misjudgment caused by fixed thresholds, a multi-dimensional population stratification threshold matrix was first established based on large-sample clinical screening data of sarcopenia in the elderly population.

[0090] For example, age is used as the first dimension, dividing the population into three age groups: 60-69 years old, 70-79 years old, and 80 years old and above; gender is used as the second dimension, dividing the population into two groups: male and female; and body mass index (BMI) is used as the third dimension, dividing the population into four ranges: <18.5, 18.5-23.9, 24.0-27.9, and ≥28.0. These three dimensions are combined to form 24 hierarchical units, and each hierarchical unit corresponds to a set of exclusive initial thresholds.

[0091] When a user uses the device for the first time, they enter their basic physiological information, such as age, gender, height, and weight. The system automatically calculates the BMI index and matches it with the corresponding stratified unit, then retrieves the initial threshold set corresponding to that unit as the initial judgment benchmark for system operation.

[0092] A wearable device includes a multimodal fusion system for muscle assessment and early intervention.

[0093] Specifically, when a user uses the wearable device described in this application, data collection and baseline establishment are completed seamlessly without the user's active intervention. That is, when the user performs actions such as gripping or lifting, the wrist flexor tendons contract, and the wearable device's biomechanical sensing unit continuously collects the pressure values ​​generated by the wrist flexor tendon contraction on the skin in real time. and monitor simultaneously The rate of change.

[0094] After the user's action ends, the wrist flexor tendon relaxes, and the pressure value rapidly decreases from above the first pressure threshold. The device calculates the rate of change through the mechanical sensing unit, and when the condition is met... When the muscle movement ends, the end time is recorded. Among them, the above-mentioned The value is 0.8. 5 kPa It takes 10ms. The value is 2 kPa.

[0095] The recovery period physiological signal acquisition unit continuously acquires surface electromyographic signals of the user's forearm flexor muscle area at a preset time of 10 seconds, forming the recovery period physiological signal.

[0096] The feature extraction unit of the trend analysis module starts working and calculates... Root mean square amplitude of electromyography at time tw For example, calculated =80μV; then determine the threshold for 50% attenuation. =0.5×80=40μV, find the moment when the root mean square amplitude of electromyography is 40μV. Calculate recovery time ,For example = +4s, then =4s, used as an indicator of muscle function recovery.

[0097] The baseline establishment unit of the trend analysis module stores all valid data collected by the user within 30 days, according to a sliding time window of L=30 days. The data (assuming 80 valid acquisitions) is used to calculate a baseline B=3.5s, which serves as the user's muscle function baseline and is continuously and dynamically updated.

[0098] The early warning judgment unit of the trend analysis module calculates in real time the current monitoring period W=7 days. The median data A (assuming H = 12 valid acquisitions, A = 4.6s) is calculated using the formula R = 4.6 / 3.5 ≈ 1.31. Since R = 1.31 > (e.g., 1.3), and the condition is met for a duration of D=5 days. At this time, the system generates a decline warning message.

[0099] When the user receives a device vibration or a warning notification from the display screen, the user triggers an active assessment command. The surface electromyography electrodes, photoplethysmography (PPG) sensor, and blood oxygen sensor are simultaneously activated, and the multimodal physiological sensing unit collects data for 60 seconds. Multimodal physiological signals (60s) including surface electromyography signals, photoplethysmography pulse wave signals, and blood oxygen saturation signals are time-stamped to ensure data synchronization.

[0100] The index generation unit of the evaluation module analyzes the collected multimodal physiological signals and calculates the three major indices respectively.

[0101] Extract the root mean square amplitude average during the evaluation period. =75μV, average equivalent grip strength value =0.38kg, and calculated according to the formula, N=75 / 0.38≈197.37 (normalized to 0.28), thus obtaining the neural recruitment efficiency index N.

[0102] Extract the median frequency at the initial evaluation time. =80Hz, End Time =77Hz, and we can calculate Q=(77-80) / 60=-0.05Hz / s, thus obtaining the muscle endurance index Q.

[0103] Extract heart rate variability during the assessment period =12 times / min, change in grip strength =0.08kg, change in blood oxygen saturation =-2%, for example, the weighting coefficient =0.6、 =0.4, first dimension conversion coefficient =0.003, second dimension conversion coefficient =15, and according to the formula, C=0.6×0.003×(12 / 0.08)+0.4×15×(-(-2) / 60)=0.45, so the cardiopulmonary support index C is obtained.

[0104] The attribution classification unit of the system evaluation module is determined according to the logical combination rule: given N=0.28< =0.3, satisfying "N< The condition of “…” is met, therefore the etiological classification result is neuromuscular dysregulation.

[0105] Based on the etiological classification of neuromuscular control dysfunction, the system intervention generation module calls the corresponding control training task from the preset training library—such as slowly clenching a fist, holding it for 5 seconds, and slowly releasing it, 10 seconds each time, for a total of 5 sets, with even force exertion.

[0106] During user training, the device's surface electromyography electrodes and flexible tactile sensor array collect training feedback signals in real time, namely the root mean square amplitude of surface electromyography and the equivalent grip force value.

[0107] The motion correction unit of the feedback output module calculates the user's force efficiency E in real time according to the formula, for example, at a certain moment. =0.4kg, =100μV, calculated as E=0.4 / 100=0.004kg / μV (this example value is adapted to the exertion level of elderly users, based on actual calibration); then, the exertion efficiency score S (out of 100) is generated according to the scoring formula, for example... =0.005kg / μV, =0.001kg / μV, due to <E< According to the formula, S = [(0.004-0.001) / (0.005-0.001)] × 100 = 75 points.

[0108] The force efficiency score S=75 points is compared with the preset quality threshold. =60 points for comparison. Since 75≥60, the force application is standardized and should be maintained without any additional instructions.

[0109] When a user exerts too much force and at a certain moment the calculated value is E=0.0008kg / μV≤Emin, S=0 minutes<60 minutes, the system generates an instruction to guide the user to adjust their movements: "Please slowly clench your fist, feel the force in your wrist, avoid sudden force, and apply grip force evenly."

[0110] In summary, the multimodal fusion system and wearable device for muscle assessment and early intervention according to the embodiments of this application establish a muscle function baseline by utilizing the electromyography recovery time collected in daily, imperceptible data. This changes the diagnosis from a single-point threshold to a continuous trend warning, enabling early detection and early intervention of sarcopenia. Through active assessment, it can obtain neural recruitment efficiency index, muscle endurance index, and cardiopulmonary support index. By logically combining and judging, it can accurately output three etiological classification results. Moreover, targeted training tasks can be output based on the etiological classification results.

[0111] In the description of this specification, 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 indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0112] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0113] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A multimodal fusion system for muscle assessment and early intervention, characterized in that, include: The data acquisition module is used to collect biomechanical signals characterizing the user's muscle load, and to collect physiological signals during the recovery period at the end of muscle movement. It is also used to collect multimodal physiological signals during the user's assessment test according to active assessment instructions. The trend analysis module is used to identify the end time of the muscle movement based on the biomechanical signals, extract muscle function recovery indicators from the physiological signals during the recovery period, establish a muscle function baseline based on the temporal changes of the muscle function recovery indicators, and generate degeneration warning information based on the muscle function baseline and the current muscle function recovery indicators. The assessment module is used to receive the active assessment command triggered by the user after the decline warning information is output, so as to analyze the multimodal physiological signals and obtain the etiological classification results of muscle weakness. An intervention generation module is used to generate training tasks based on the etiological classification results; The feedback output module is used to collect training feedback signals when the user performs the training task, and generate and output action explanation instructions based on the training feedback signals.

2. The multimodal fusion system for muscle assessment and early intervention according to claim 1, characterized in that, The data acquisition module includes: The mechanical sensing unit is used to collect the pressure value generated on the skin when the wrist flexor tendon contracts through a flexible tactile sensor array embedded in the inner side of the wearable device strap, and to detect the end of the muscle movement based on the rate of change of the pressure value. The mechanical sensing unit is also used to convert the pressure value into an equivalent grip force value through a mapping relationship. The recovery period physiological signal acquisition unit is used to continuously acquire surface electromyographic signals of the forearm flexor muscle region for a preset duration after the muscle movement is detected to constitute the recovery period physiological signal. The multimodal physiological sensing unit is used to simultaneously activate the surface electromyography electrode, photoplethysmography pulse wave sensor and blood oxygen sensor built into the wearable device according to the active assessment command, so as to collect the multimodal physiological signals. The multimodal physiological signals include surface electromyography signals acquired by the surface electromyography electrodes, photoplethysmography signals acquired by the photoplethysmography sensor, and blood oxygen saturation signals acquired by the blood oxygen sensor.

3. The multimodal fusion system for muscle assessment and early intervention according to claim 1, characterized in that, The trend analysis module includes: The event detection unit is used to identify the moment when the muscle movement ends based on the biomechanical signal. The moment when the muscle movement ends is the moment when the pressure value drops from above a first pressure threshold to below a second pressure threshold. The feature extraction unit is used to extract muscle function recovery indicators from the physiological signals during the recovery period. The muscle function recovery indicators are the recovery time required for the root mean square amplitude of electromyography to decrease by 50% from the termination time. The baseline establishment unit is used to store all the muscle function recovery indicators within the user's sliding time window, and to calculate the first statistical feature value within the sliding time window as the muscle function baseline. The early warning determination unit is used to calculate the second statistical feature value of all muscle function recovery indicators within the current monitoring period, and calculate the ratio of the second statistical feature value to the muscle function baseline. When the ratio is greater than a preset threshold for a continuous preset duration, the decline early warning information is generated.

4. The multimodal fusion system for muscle assessment and early intervention according to claim 3, characterized in that, Both the first and second statistical characteristic values ​​are medians or percentiles.

5. The multimodal fusion system for muscle assessment and early intervention according to claim 2, characterized in that, The evaluation module includes: An index generation unit is used to analyze the multimodal physiological signals to obtain neural recruitment efficiency index, muscle endurance index, and cardiopulmonary support index. The attribution classification unit is used to logically combine and determine the neural recruitment efficiency index, muscle endurance index, and cardiopulmonary support index, and output the etiological classification result.

6. The multimodal fusion system for muscle assessment and early intervention according to claim 5, characterized in that, The analysis of the multimodal physiological signals to obtain neural recruitment efficiency index, muscle endurance index, and cardiopulmonary support index includes: The root mean square amplitude of the surface electromyography signal at the time of synchronous sampling with the equivalent grip force value is extracted, the ratio of the root mean square amplitude to the equivalent grip force value is calculated, and the ratio is used as the neural recruitment efficiency index. The slope of the median frequency decrease during the evaluation test was extracted from the surface electromyography signal as a muscle endurance index. The heart rate change is extracted from the photoplethysmography signal, the rate of decrease of blood oxygen saturation during the evaluation test is extracted from the blood oxygen saturation signal, and the grip strength change is extracted from the equivalent grip strength value. The cardiopulmonary support index is generated based on the ratio of the heart rate change to the grip strength change and the rate of decrease of blood oxygen saturation.

7. The multimodal fusion system for muscle assessment and early intervention according to claim 5, characterized in that, The logical combination determination, which outputs the etiological classification result, includes: If the neural recruitment efficiency index is lower than the efficiency threshold, the etiological classification result is neuromuscular dysregulation. If the muscle endurance index is lower than the endurance threshold and the neural recruitment efficiency index is not lower than the efficiency threshold, then the etiological classification result is insufficient muscle endurance. If the cardiopulmonary support index is lower than the support threshold, the neural recruitment efficiency index is not lower than the efficiency threshold, and the muscle endurance index is not lower than the endurance threshold, then the etiological classification result is cardiopulmonary function limitation.

8. The multimodal fusion system for muscle assessment and early intervention according to claim 1, characterized in that, The intervention generation module is used to call the training task from a preset training library based on the etiological classification results, wherein... The training library includes control training tasks corresponding to neuromuscular dysfunction. Training tasks corresponding to insufficient muscle endurance; Intermittent training tasks corresponding to cardiopulmonary function limitations.

9. The multimodal fusion system for muscle assessment and early intervention according to claim 1, characterized in that, The feedback output module includes: The signal acquisition unit is used to acquire the root mean square amplitude of surface electromyography and the equivalent grip force value in real time when the user performs the training task, so as to serve as the training feedback signal. The motion correction unit is used to calculate the ratio of the equivalent grip force value to the root mean square amplitude of surface electromyography at the same moment, generate the force efficiency, generate a force efficiency score based on the force efficiency, compare the force efficiency score with a preset quality threshold, and generate and output a motion explanation instruction when the force efficiency score is lower than the preset quality threshold.

10. A wearable device, characterized in that, Includes the multimodal fusion system for muscle assessment and early intervention as described in any one of claims 1 to 9.

Images