Graphene film-based intelligent lumbar support control method and rehabilitation device

Abstract

This application relates to the technical field of lumbar support, and in particular to a smart lumbar support control method and rehabilitation device based on graphene membrane. The method includes: multimodal data acquisition: acquiring multimodal physiological and motor signals of the user's lumbar region; intelligent state analysis: inputting the multimodal signals into a pre-trained intelligent control model for analysis and processing, and outputting a comprehensive assessment result of the user's lumbar load, muscle tension, and thermal comfort, analyzing the user's current lumbar physiological state, specific activity patterns, and need for heat therapy; comprehensive adaptive control: generating coordinated control commands based on the analyzed lumbar physiological state, specific activity patterns, spinal traction needs, and need for heat therapy. The device includes a processor and a memory communicatively connected to the processor. This application achieves real-time, comprehensive understanding of the user's lumbar state, and accordingly performs synchronous, adaptive, and precise control of the two core functions of support and heat therapy.

Description

Technical Field

[0001] This application relates to the technical field of lumbar support, and in particular to an intelligent lumbar support control method and rehabilitation device based on graphene film. Background Technology

[0002] Lumbar spine health issues, especially chronic low back pain and postpartum diastasis recti, have become prevalent conditions affecting the quality of life of modern people, particularly women. Common interventions include physical orthotics and thermotherapy, with wearing lumbar support devices being one of the most widely used methods. Traditional support devices primarily provide restraint and stability through physical pressure, while some products incorporate heat therapy to promote blood circulation and relieve muscle tension. With advancements in materials science and electronic information technology, utilizing novel functional materials to achieve smarter and more comfortable rehabilitation interventions has become an industry trend. Graphene, due to its excellent electrical and thermal conductivity, as well as its ability to release 5-15μm far-infrared rays that resonate with human cells, is widely used in smart wearables and healthcare, providing an ideal material basis for developing next-generation intelligent lumbar support devices.

[0003] Currently, various lumbar support or physiotherapy devices combining graphene and smart elements are available in existing technologies. For example, a lumbar support can provide heating therapy by incorporating parallel graphene electrothermal films within its main body, allowing for precise temperature control of different areas via a mobile app. In terms of orthopedic and intelligent monitoring, existing technologies also include integrated spinal orthopedic modules, intelligent detection modules (for monitoring pressure and temperature), and infrared thermotherapy modules utilizing nano-graphene technology.

[0004] Despite some progress in existing technology, significant shortcomings remain, especially when applied to women's compression lumbar support belts that require multi-functional integration (combining support, heating, and restraint). First, the level of functional integration and intelligence is insufficient: most existing products offer only single functions, or simply add heating and support, lacking the ability to adaptively and collaboratively adjust based on the user's real-time physiological state (such as muscle tension, core temperature, and posture) and environment. Second, human-computer interaction and personalized adaptation capabilities are weak: the support intensity and heating mode of existing devices are often preset or manually adjusted, failing to accurately match the personalized needs of users in different activities (such as sitting, walking, and abdominal crunches during rehabilitation) or different health stages (such as different periods of postpartum recovery). Summary of the Invention

[0005] This application provides a smart lumbar support control method and rehabilitation device based on graphene film, which can at least partially solve the above-mentioned technical problems.

[0006] In the first aspect, this application provides a smart lumbar support control method and rehabilitation device based on graphene film, which adopts the following technical solution:

[0007] A smart lumbar support control method based on graphene film includes the following steps:

[0008] Multimodal data acquisition: Acquire multimodal physiological and motor signals of the user's waist and abdomen, including at least circumferential strain signals of the waist and abdomen, body surface temperature signals, and electromyographic signals;

[0009] Intelligent state analysis: The multimodal signals are input into a pre-trained intelligent control model for analysis and processing, and a comprehensive evaluation result of the user's lumbar load, muscle tension and thermal comfort is output. Based on the output result, the user's current lumbar physiological state, specific activity mode and need for implementing heat therapy are analyzed.

[0010] Comprehensive adaptive regulation: Based on the analyzed lumbar physiological state, specific activity patterns, spinal traction requirements, and thermotherapy requirements, a coordinated control command is generated to synchronously adjust the mechanical deformation parameters, axial traction parameters, and thermotherapy parameters of the graphene electrothermal drive element integrated in the flexible substrate.

[0011] By employing the above technical solution, multimodal signals from the user's lumbar region are first collected via sensors, including circumferential strain, surface temperature, and electromyography (EMG) signals. These signals are input into a pre-trained intelligent control model. The model processes the signals and outputs a comprehensive assessment of lumbar load, muscle tension, and thermal comfort. Based on this assessment, the system analyzes the user's current lumbar physiological state, ongoing activity patterns, whether spinal traction is needed, and the required intensity of thermotherapy. Finally, based on all these analyses, a unified control command is generated, simultaneously adjusting three parameters of the graphene electrothermal drive element: mechanical deformation (for support and shaping), axial traction (for spinal stretching), and thermotherapy output (for heat therapy).

[0012] The system integrates three core rehabilitation techniques—support, traction, and thermotherapy—through a unified intelligent analysis and decision-making center for coordinated control. Based on real-time, multi-dimensional physiological data, the system proactively assesses the spinal biomechanical state (including whether traction is needed) and issues synchronous adjustment commands. This enables comprehensive, integrated, and adaptive intervention for lumbar spine problems, rather than simply adding up individual or fragmented functions. It achieves a real-time, comprehensive understanding of the user's lumbar condition and accordingly performs synchronized, adaptive, and precise control of the two core functions: support and thermotherapy.

[0013] Optionally, the intelligent state analysis step further includes:

[0014] Action pattern recognition: Based on the temporal variation characteristics of the strain signal and electromyographic signal, the intelligent control model identifies the specific actions that the user is performing, including stillness, walking, sitting, bending over or crunching.

[0015] By employing the aforementioned technical solution, specifically utilizing the temporal characteristics of strain and electromyographic signals, an intelligent control model identifies the user's current specific actions (stillness, walking, sitting, bending over, abdominal crunches), achieving precise identification of the user's dynamic behavioral intentions. This identification result provides crucial contextual information for subsequent regulation, enabling the application of support, traction, and thermotherapy to be highly correlated with the user's real-time activities, laying the foundation for contextualized and predictive intervention; it allows support and thermotherapy strategies to evolve from universal responses to contextualized responses highly correlated with specific actions, laying the foundation for meeting personalized needs under different activity modes; and it enables intelligent state analysis to not only assess static physiological indicators (load, tension) but also understand dynamic behavioral patterns, significantly improving the dimensionality and accuracy of state analysis, providing richer evidence for generating more precise regulatory instructions.

[0016] Optionally, when the user is detected performing abdominal crunches, the integrated adaptive control step includes:

[0017] Dynamic deformation assistance: The graphene electrothermal drive element is controlled to generate a preset curling deformation inward in the corresponding area on the front side of the user's abdomen to provide dynamic auxiliary support.

[0018] Spinal extension assistance: During the recovery phase of the crunch, a graphene electrothermal drive element located in the posterior region of the lumbar spine is controlled to produce a gentle, axial extension deformation to assist the spine in extending segment by segment and reduce pressure on the posterior intervertebral discs.

[0019] Synergistic thermotherapy enhancement: Simultaneously increase the far-infrared thermotherapy output power of the corresponding area to cooperate with muscle movement and promote metabolism.

[0020] By employing the above technical solution, when the system detects that a user is performing an abdominal crunch, the comprehensive adaptive control process is specifically triggered by three simultaneous operations. First, the graphene element in front of the abdomen is controlled to produce a specific deformation that curls inward into the abdominal cavity. Second, during the recovery phase of the crunch (when the body returns to a lying or sitting position), the graphene element behind the lumbar spine is controlled to produce a gentle, extending deformation along the spinal axis. Third, the far-infrared heating power in the corresponding area is increased; firstly, the curling deformation in the front of the abdomen dynamically increases intra-abdominal pressure, providing hydraulic support to the lumbar spine from the front and guiding the contraction of the rectus abdominis muscles. Secondly, the axial extending deformation provided during the recovery phase actively assists the spine in extending smoothly segment by segment, helping to balance intervertebral disc pressure, especially reducing the load on the posterior annulus fibrosus. Finally, the simultaneous local heat therapy promotes muscle metabolism and relieves potential stiffness. The combination of these three elements achieves a leap from passive protection to actively enhancing training effects and optimizing movement biomechanics.

[0021] Optionally, the multimodal data acquisition step further includes acquiring humidity signals from the skin surface of the waist; the comprehensive adaptive control step further includes:

[0022] Wearable microenvironment management: When the humidity signal exceeds a preset threshold, the graphene electrothermal drive element is controlled to switch to a low-power ventilation and dehumidification mode, which promotes air circulation and moisture evaporation while maintaining basic heat sensation.

[0023] By adopting the above technical solution, the humidity signal of the skin surface of the waist is additionally collected during the data acquisition process. When the humidity value exceeds the set safety threshold, the control process will control the graphene electrothermal drive element to switch to a special working mode. This mode operates at a lower power, aiming to maintain the basic warmth while using heat to promote air flow and accelerate the evaporation of moisture on the skin surface.

[0024] The system introduces intelligent management of the wearable microenvironment. Its direct effect is to significantly improve the comfort of long-term wear of the device and reduce the risk of skin discomfort, allergies, or inflammation caused by heat and humidity. By ensuring wearable comfort, it increases users' willingness and compliance to continue using the device, thereby ensuring that core rehabilitation interventions (support, traction, and thermotherapy) can play a long-term and continuous role.

[0025] Optionally, the intelligent state analysis step further includes physiological adaptive modeling:

[0026] Based on the user's historical motion curves, the intelligent control model incorporates a fatigue-adaptation dynamic model. This model quantifies the physiological adaptation gain of the user's lumbar muscle groups to a specific activity pattern by analyzing the user's repeated exposure history to that pattern. The model also quantifies and evaluates the user's tolerance and effectiveness gain to the traction intervention by analyzing the user's immediate response and relaxation rate of electromyography signals when the axial traction waveform intervention is applied.

[0027] Fatigue risk assessment: Based on the long-term continuous electromyographic and postural signals, the static load accumulation of the lumbar muscles is analyzed;

[0028] Active intervention: When the accumulated level exceeds the safety threshold, a preset intervention program to relieve muscle fatigue is automatically triggered and executed. The program includes periodic support tension and relaxation cycles and pulsed heat therapy.

[0029] By adopting the above technical solution and introducing physiological adaptation modeling based on the user's historical movement curves, the system can dynamically quantify the user's physiological adaptation gain to specific activity patterns and tolerance to traction intervention. Based on this, fatigue risk assessment is performed by combining long-term continuous electromyography and posture signals, enabling personalized calculation of the cumulative static load on the lumbar muscles. When the assessed cumulative load exceeds a personalized safety threshold, the system automatically triggers a preset active intervention program. This approach promotes the evolution of intervention logic from "real-time symptom response" to "predictive protection based on long-term personalized status." The system can quantitatively assess the risk of chronic fatigue accumulation that users may not easily perceive and proactively intervene before it causes pain or strain. This predictive intervention based on a personalized risk model helps break the vicious cycle of "fatigue-compensation-strain," thereby preventing the occurrence or aggravation of chronic low back pain at a more proactive stage, achieving a paradigm shift from "passive treatment" to "active protection."

[0030] Optionally, the periodic support tension-relaxation cycle specifically includes: controlling graphene electrothermal drive elements in different areas of the waist and abdomen to alternately deform and relax according to a preset time sequence and waveform, forming a dynamic wave-like pressure massage around the waist; and the cycle and amplitude of this cycle are adjusted linearly or non-linearly according to the value of the fatigue accumulation degree.

[0031] The waveform includes an axial traction waveform: controlling the drive elements located in the symmetrical partitions on both sides of the lumbar spine to produce synchronous deformation perpendicular to the body surface.

[0032] By employing the above technical solution, graphene elements in different zones of the waist and abdomen are controlled to alternately deform and relax according to a set time and waveform sequence, creating an effect of pressure waves moving around the waist. The rhythm and intensity of this massage are automatically adjusted according to the real-time fatigue level. In particular, the waveform includes an axial traction waveform, which, by controlling the elements in symmetrical zones on both sides of the spine, synchronously generates deformation perpendicular to the body surface (i.e., roughly along the spinal axis).

[0033] First, dynamic wave-like massage can more effectively promote deep blood circulation and muscle relaxation. Second, its parameters adapt to the level of fatigue, allowing for a precise match between stimulation intensity and needs. Third, and most importantly, by integrating axial traction waveforms, conventional relaxation massage simultaneously possesses intermittent, dynamic spinal axial micro-traction functionality. This axial force embedded in the circumferential massage gently and intermittently increases the intervertebral space, creating mechanical conditions for decompression and nutrient exchange in the intervertebral discs, achieving a synergistic effect of muscle relaxation and spinal decompression.

[0034] Optionally, in the intelligent state analysis step, the comprehensive evaluation result includes the net fatigue load based on the physiological adaptation model;

[0035] When calculating the static load accumulation level in real time, the original load generated by the current activity mode is subtracted from the real-time physiological adaptation gain corresponding to that mode to obtain the net fatigue load, and intervention is triggered or adjusted based on this net load value.

[0036] By employing the aforementioned technical solutions and distinguishing between "original load" and "net fatigue load," the system can more accurately identify the user's true fatigue risk. For example, for routine activities to which the user has developed physiological adaptation, the system will reduce intervention accordingly to avoid overprotection; while for activities to which the user has not yet adapted or where the load suddenly increases, the system will increase sensitivity and provide timely support. Crucially, this model quantifies and models the user's personalized tolerance and effectiveness response to traction intervention. This allows traction, an operation requiring precise dosage, to be optimized based on system learning, providing a core decision-making basis for subsequent personalized and precise traction control under safety domain constraints, significantly improving the system's intelligence and safety.

[0037] Optionally, the comprehensive adaptive control step also includes multi-objective collaborative optimization: a multi-objective optimization model with the constructed support effect index, thermotherapy comfort index, user tolerance index and traction safety index as the core;

[0038] The target value of the support effect index is positively correlated with the net fatigue load; the threshold of the user tolerance index is positively correlated with the physiological adaptability gain; the preferred range of the thermotherapy comfort index is dynamically calculated based on the body surface temperature signal and the metabolic improvement requirements of the net fatigue load; the traction safety index is jointly determined by the user's tolerance gain to traction intervention, real-time spinal curvature status, and electromyographic tension, and is used to constrain the range of axial traction parameters.

[0039] The method further includes, based on the value and trend of the net fatigue load, utilizing the reversible switching characteristic of the elastic modulus of the graphene-based phase change elastomer film in the graphene electrothermal drive element near 28.5°C, dynamically adjusting the rigidity or flexibility of the film when performing the wave-like pressure massage, so as to match the pressing texture required for different fatigue stages.

[0040] By adopting the above technical solutions, the system supports an efficacy index, a thermotherapy comfort index, a user tolerance index, and a newly added traction safety index. The calculation benchmarks or constraints of these indices are dynamically correlated with personalized parameters such as the resulting physiological adaptation gain, net fatigue load, and traction tolerance gain. For example, the traction safety index is directly determined by the user's traction tolerance gain, real-time spinal curvature, and electromyographic tension, and is used to limit the magnitude and duration of the traction force. The system solves this optimization model to find the optimal parameter combination under multiple constraints (such as ensuring traction safety), and accordingly controls the massage texture (adjusting the firmness through the phase transition properties of the graphene film) and the traction waveform.

[0041] This transforms complex clinical decision-making problems into computable optimization problems, ensuring the safety and personalization of interventions. The optimization model is no longer based on fixed rules, but on a continuously evolving personalized physiological model. In particular, the introduction of the traction safety index strictly constrains this powerful intervention method within the safety boundary based on the user's real-time physiological state, achieving automated intelligent traction with controllable risks.

[0042] Optionally, it also includes a weight adaptation step, specifically including:

[0043] During the intervention period, the rate of decrease of the net fatigue load is calculated in real time as the support effect feedback value, the stability of the body surface temperature signal in the thermotherapy area is monitored as the thermotherapy comfort feedback value, and the proportion of high-frequency components representing discomfort in the electromyographic signal is used as the tolerance feedback value.

[0044] The above feedback values ​​are compared with their respective dynamic benchmark values, and the magnitude and rate of weight adjustment are individually constrained based on the user's individual recovery characteristics revealed by the fatigue-adaptation dynamic model.

[0045] By employing the above technical solution, three key feedbacks are monitored in real time: the rate of decrease in net fatigue load (support effect feedback), the stability of temperature in the thermotherapy area (thermotherapy comfort feedback), and the proportion of high-frequency components in electromyographic signals (tolerance feedback). These measured feedback values ​​are compared with the initial predicted target values ​​of the optimization model, and the weight allocation of each optimization objective (support, comfort, tolerance, and safety) in the next round of decision-making is dynamically adjusted based on the deviation. Furthermore, the magnitude and speed of this weight adjustment are constrained by the user's individual recovery characteristics revealed by the fatigue-adaptation dynamic model.

[0046] This endows the entire system with the ability to continuously self-optimize. It not only responds to real-time conditions but also reflects on and adjusts its decision preferences based on the effects of each intervention. This forms a two-layer learning and optimization architecture: the bottom layer is a long-term physiological characteristic model, and the upper layer is a short-term policy optimizer. The policy adjustments in the upper layer are constrained by the individual characteristics in the lower layer, ensuring that the system's evolution does not deviate from the user's personal comfort and safety boundaries, achieving a highly human-like, robust, and personalized adaptive system.

[0047] Secondly, the rehabilitation device provided in this application adopts the following technical solution:

[0048] A rehabilitation device includes: a processor, and a memory communicatively connected to the processor;

[0049] The memory is provided with a computer-readable storage medium, and a computer program is stored on the computer-readable storage medium.

[0050] When the processor processes the computer program stored on the computer-readable storage medium, it implements a graphene film-based intelligent lumbar support control method.

[0051] In summary, this application includes at least one of the following beneficial technical effects:

[0052] 1. The system integrates three core rehabilitation methods—support, traction, and thermotherapy—through a unified intelligent analysis and decision-making center for coordinated control. Based on real-time, multi-dimensional physiological data, the system proactively assesses the spinal biomechanical state (including whether traction is needed) and issues synchronous adjustment commands. This enables comprehensive, integrated, and adaptive intervention for lumbar spine problems, rather than the superposition of single or fragmented functions. It achieves real-time, comprehensive understanding of the user's lumbar condition and accordingly performs synchronized, adaptive, and precise control of the two core functions: support and thermotherapy.

[0053] 2. The user's personalized tolerance and effectiveness response to traction intervention were quantitatively modeled, which enabled the optimization of traction, an operation that requires precise dosage, based on system learning. This provided the core decision-making basis for the subsequent realization of personalized and precise traction control under safety domain constraints, and significantly improved the system's intelligence level and safety.

[0054] 3. It transforms complex clinical decision-making problems into computable optimization problems, ensuring the safety and personalization of interventions; the optimization model is no longer based on fixed rules, but on a continuously evolving personalized physiological model; in particular, the introduction of the traction safety index strictly constrains this powerful intervention method of traction within the safety boundary based on the user's real-time physiological state, realizing automated intelligent traction with controllable risks. Attached Figure Description

[0055] Figure 1 This is a flowchart of the intelligent lumbar support control method in Embodiment 1 of this application;

[0056] Figure 2 This is a flowchart of the intelligent lumbar support control method in Embodiment 6 of this application. Detailed Implementation

[0057] The following combination Figures 1 to 2 This application will be described in further detail.

[0058] This embodiment discloses an intelligent lumbar support control method based on graphene film.

[0059] Example 1: Refer to Figure 1 A smart lumbar support control method based on graphene film includes the following steps:

[0060] Multimodal data acquisition: Acquire multimodal physiological and motor signals of the user's waist and abdomen, including at least circumferential strain signals of the waist and abdomen, body surface temperature signals, and electromyographic signals;

[0061] Intelligent state analysis: The multimodal signals are input into a pre-trained intelligent control model for analysis and processing, and a comprehensive evaluation result of the user's lumbar load, muscle tension and thermal comfort is output. Based on the output result, the user's current lumbar physiological state, specific activity mode and need for implementing heat therapy are analyzed.

[0062] Comprehensive adaptive regulation: Based on the analyzed lumbar physiological state, specific activity patterns, spinal traction requirements, and thermotherapy requirements, a coordinated control command is generated to synchronously adjust the mechanical deformation parameters, axial traction parameters, and thermotherapy parameters of the graphene electrothermal drive element integrated in the flexible substrate.

[0063] Multimodal data acquisition: Real-time acquisition of multimodal physiological and motion signals of the user's waist and abdomen through a sensor array integrated on a flexible substrate.

[0064] Circumferential strain signal of the waist and abdomen Measurements were taken using distributed graphene resistance strain sensing units. Let the relative resistance change of the i-th sensing unit at time t be... Its relationship with the strain it receives The relationship is determined by the pre-calibrated sensitivity coefficient. Decide: The system fits the strain values ​​of N units around the waist and abdomen to obtain the circumferential strain distribution function describing the overall binding tightness and shape of the waist belt. , where θ is the orbital angle coordinate.

[0065] Body surface temperature signal The resistance value was measured using a graphene temperature sensing unit. With temperature The relationship is: ,in The resistance is given at a reference temperature T0, and α is the temperature resistivity of graphene. This is achieved by measuring... The contact point temperature was calculated by reverse calculation. The temperature distribution is obtained by fusing data from multiple points.

[0066] electromyographic signals Electromyographic signals from target muscle groups such as the erector spinae and rectus abdominis in the lumbar region are acquired using surface electromyography electrodes. The raw signals are amplified and bandpass filtered (e.g., 20-500Hz), and their root mean square (RMS) values ​​are calculated as an indicator of muscle activation level. ,in This is the length of the time window, for example, 100ms.

[0067] Intelligent state analysis: This involves analyzing the collected multi-dimensional signal vectors. The input is fed into a pre-trained deep neural network model. This model (such as a combination of a Long Short-Term Memory (LSTM) network and an attention mechanism) analyzes the temporal and spatial correlations of the signal and outputs a comprehensive state evaluation vector. .

[0068] The lumbar load index L(t) assesses the overall mechanical load on the lumbar intervertebral discs and articular surfaces, with a value range of [0,1]. Its calculation incorporates strain signals. Symmetry (reflecting whether the posture is neutral) and amplitude (reflecting external load), as well as electromyographic signals. The co-activation level. The formula can be expressed as: ,in To extract posture and load characteristics from strain distribution, It is a characteristic of electromyographic coordinated contraction. The spatiotemporal feature vector representing posture and load is extracted from the strain distribution ε(θ,t) using a convolutional neural network. To obtain multi-channel electromyography signals The feature vector representing the muscle co-activation mode is obtained by calculating the cross-correlation matrix, etc. As weight, For bias terms, The Sigmoid activation function maps the output to the interval [0,1].

[0069] Muscle tension M(t): assesses the real-time tension and fatigue state of muscles, with a value ranging from [0,1]. Its calculation is primarily based on the median frequency shift ΔMF and amplitude decay rate of the electromyographic signal. The formula can be expressed as:

[0070] .

[0071] Thermal comfort C(t): assesses the user's subjective comfort level with the current heat therapy, ranging from -1 to 1. It is calculated based on body surface temperature. Ambient temperature (This can be estimated using a temperature sensor) and the rate of temperature change. The formula can be expressed as: ,in The target temperature set by the user.

[0072] Based on the state assessment vector S(t), the system determines the current physiological state of the lumbar region (e.g., high load and high risk, moderate fatigue, thermal comfort), identifies the current activity pattern (e.g., sitting, walking) by analyzing the temporal patterns of strain and electromyography, and infers the intensity of demand for heat therapy based on L(t) and M(t). .

[0073] Integrated adaptive control: Generates coordinated control commands based on the output of intelligent state analysis.

[0074] Mechanical deformation parameters : Controlling the driving voltage or current of the graphene electrothermal drive element to produce the desired curvature change. . Calculations are made based on L(t) and the identified posture. For example, when L(t) is high and the posture is identified as bending over, a larger support curvature is set in the lumbar spine region.

[0075] Axial traction parameters When needed (e.g., after prolonged sitting with L(t) remaining at a high level), specific partition elements are controlled to produce deformation displacement perpendicular to the body surface. . Positively correlated with L(t), but subject to maximum safe displacement constraint.

[0076] Thermotherapy parameters : Control the heating power or duty cycle of the drive element to achieve the target temperature. . according to and basic comfort temperature calculate: , where γ is the gain coefficient, with a value of 0.8℃; The value is 36.5°C, which is slightly higher than the baseline temperature of the abdominal surface (about 33-35°C). It can provide a perceptible, mild warmth, promote local blood flow, and is far below the risk threshold for low-temperature burns (>44°C).

[0077] Wherein, the target curvature Based on the lumbar load index L(t) and postural characteristics The mapping yields: ,in This is a predefined or learned mapping function. The target displacement... Positively correlated with L(t), that is ,in This is the displacement gain coefficient. The preset maximum safe displacement is set at 2.0 mm. Referring to the safety principles of clinical non-surgical spinal traction, for the daily use of wearable devices, the single traction displacement should be much smaller than the clinical treatment amount (centimeter level). 2.0 mm of micro-traction is sufficient to produce beneficial mechanical changes in the intervertebral space, while greatly ensuring safety.

[0078] These parameters are synchronously adjusted through a low-level PID controller, and ultimately act on the graphene electrothermal drive element to achieve a synergistic output of support, traction, and thermotherapy.

[0079] For example, a user wears this support belt while working. The system continuously acquires signals X(t). When the user sits for a long time with their back bent forward, the strain distribution ε(θ,t) shows anterior compression and posterior stretching, and the EMG signal shows continuous activation of the erector spinae muscles, with increased L(t) and M(t) values. The condition analysis indicates a sedentary lifestyle with high fatigue load, requiring heat therapy. Rise. The control module generates an instruction: Increase the driving voltage of the drive element in the lumbar spine region, causing it to undergo a forward-pushing deformation (increase...). This provides additional support; simultaneously, it initiates low-intensity axial traction (set). (0.5mm), intermittently relieving intervertebral disc pressure; heating power is simultaneously fine-tuned to make... Maintaining a temperature slightly above the baseline (e.g., rising from 36.5℃ to 37.2℃) promotes blood circulation. When the user stands up and walks, the signal change triggers the system to switch to walking mode, reducing rigid support and increasing flexible following deformation.

[0080] This embodiment realizes real-time perception of the lumbar condition and preliminary coordinated control of support, traction, and heat therapy, providing a basic framework for subsequent intelligent expansion.

[0081] Example 2: This example differs from Example 1 in that the intelligent state analysis step further includes:

[0082] Action pattern recognition: Based on the temporal variation characteristics of the strain signal and electromyographic signal, the intelligent control model identifies the specific actions that the user is performing, including stillness, walking, sitting, bending over or crunching.

[0083] When the user is detected to be performing abdominal crunches, the comprehensive adaptive control steps include:

[0084] Dynamic deformation assistance: The graphene electrothermal drive element is controlled to generate a preset curling deformation inward in the corresponding area on the front side of the user's abdomen to provide dynamic auxiliary support.

[0085] Spinal extension assistance: During the recovery phase of the crunch, a graphene electrothermal drive element located in the posterior region of the lumbar spine is controlled to produce a gentle, axial extension deformation to assist the spine in extending segment by segment and reduce pressure on the posterior intervertebral discs.

[0086] Synergistic thermotherapy enhancement: Simultaneously increase the far-infrared thermotherapy output power of the corresponding area to cooperate with muscle movement and promote metabolism.

[0087] This embodiment refines the state analysis and regulation based on embodiment 1 to identify and assist specific rehabilitation actions.

[0088] Enhanced intelligent state analysis: Building upon the intelligent state analysis steps, the intelligent control model additionally trains an action pattern recognition branch. This branch extracts the temporal joint features of strain and electromyographic signals, outputting an action probability vector. These correspond to being at rest, walking, sitting, bending over, and crunching, respectively. When When the threshold is set to 0.85, it is determined to be a crunch exercise. The model further outputs the movement phase, such as the upward contraction phase, the peak hold phase, and the downward recovery phase; The value is set to ensure the accuracy of action recognition and avoid misclassifying similar actions (such as sit-ups) as crunches, thereby ensuring the safety and accuracy of the assistance. It is a standard robustness setting based on the confidence of the classification model.

[0089] Comprehensive adaptive control for crunches: Upward contraction phase: Control the front abdominal area (corresponding angle) A graphene element is used to apply a voltage. This produces a deformation that bends inwards towards the abdomen, with a curvature of... and This is proportional to the temperature of the target area for hyperthermia, providing additional support. Simultaneously, the target temperature for hyperthermia in this area increases. (For example, 2.0°C).

[0090] Downward recovery phase: Controlling the components in the posterior region of the lumbar spine, applying voltage. This produces a gradual, elongated deformation along the spinal axis, with a displacement of... It is negatively correlated with the speed of retraction and is designed to assist in the smooth extension of the spine and reduce pressure on the posterior intervertebral disc.

[0091] This embodiment achieves accurate recognition and bionic assistance for specific rehabilitation movements, elevating intelligent support to the level of movement coordination.

[0092] Example 3: This example differs from Example 2 in that the multimodal data acquisition step further includes acquiring humidity signals from the skin surface of the waist; the comprehensive adaptive control step further includes:

[0093] Wearable microenvironment management: When the humidity signal exceeds a preset threshold, the graphene electrothermal drive element is controlled to switch to a low-power ventilation and dehumidification mode, which promotes air circulation and moisture evaporation while maintaining basic heat sensation.

[0094] This embodiment adds environmental comfort management and proactive intervention based on long-term monitoring to the existing embodiment 2.

[0095] Added data collection of skin surface humidity signal RH(t) (relative humidity, in %). The capacitance value was measured using a graphene-based capacitive humidity sensor. It has an approximately linear relationship with relative humidity: , where β is the sensitivity.

[0096] Microenvironment management: When When the humidity reaches 75%, it is considered a high humidity environment. The system will switch the heating power to a low-power pulse mode, with a power of [value missing]. The pulse period is This mode utilizes intermittent hot airflow to promote air circulation; The value is determined based on ergonomics. When the relative humidity of the microenvironment remains above 70%-75%, the skin's evaporative heat dissipation efficiency decreases significantly, leading to a marked increase in discomfort and the risk of dermatitis. Setting this threshold allows for timely activation of dehumidification, ensuring wearing comfort.

[0097] This embodiment adds intelligent management of the wearable microenvironment and realizes predictive proactive intervention based on the risk of fatigue accumulation.

[0098] Example 4: This example differs from Example 3 in that the intelligent state analysis step further includes physiological adaptive modeling:

[0099] Based on the user's historical motion curves, the intelligent control model incorporates a fatigue-adaptation dynamic model. This model quantifies the physiological adaptation gain of the user's lumbar muscle groups to a specific activity pattern by analyzing the user's repeated exposure history to that pattern. The model also quantifies and evaluates the user's tolerance and effectiveness gain to the traction intervention by analyzing the user's immediate response and relaxation rate of electromyography signals when the axial traction waveform intervention is applied.

[0100] Fatigue risk assessment: Based on the long-term continuous electromyographic and postural signals, the static load accumulation of the lumbar muscles is analyzed;

[0101] Active intervention: When the accumulated level exceeds the safety threshold, a preset intervention program to relieve muscle fatigue is automatically triggered and executed. The program includes periodic support tension and relaxation cycles and pulsed heat therapy.

[0102] The periodic support tension-relaxation cycle specifically includes: controlling graphene electrothermal drive elements in different areas of the waist and abdomen to alternately deform and relax according to a preset time sequence and waveform, forming a dynamic wave-like pressure massage around the waist; and the cycle and amplitude of this cycle are adjusted linearly or non-linearly according to the value of the fatigue accumulation degree.

[0103] The waveform includes an axial traction waveform: controlling the drive elements located in the symmetrical partitions on both sides of the lumbar spine to produce synchronous deformation perpendicular to the body surface.

[0104] In the intelligent state analysis step, the comprehensive evaluation result includes the net fatigue load based on the physiological adaptation model;

[0105] When calculating the static load accumulation level in real time, the original load generated by the current activity mode is subtracted from the real-time physiological adaptation gain corresponding to that mode to obtain the net fatigue load, and intervention is triggered or adjusted based on this net load value.

[0106] Physiological Adaptation Modeling: Based on statistical user history motion curves, the intelligent control model incorporates a fatigue-adaptation dynamic model. This model analyzes the user's repeated exposure history to a specific activity pattern and quantifies the physiological adaptive gain of the user's lumbar muscle groups to that pattern. Specifically, the system maintains an adaptive gain variable for each user. When a user repeatedly engages in a certain activity mode m, The number of new cases will increase slowly based on the recovery rate after each activity. The model update formula is as follows: .in, R is the forgetting factor (value 0.95), and R is the adaptive gain for this activity, calculated based on the rate of decrease in muscle tension M(t) after the activity.

[0107] At the same time, the model assesses the user's tolerance gain to traction. The values ​​are updated based on the amplitude changes of the high-frequency components (>150Hz) in the electromyographic signal during traction application. A faster decrease in amplitude indicates better tolerance. Increase.

[0108] Personalized fatigue assessment: When calculating lumbar load in real time, the original load L(t) generated by the current activity mode is subtracted from the real-time physiological adaptation gain corresponding to that mode to obtain the net fatigue load. Where λ is the conversion factor. The intervention trigger condition is changed to... Upper limit of traction displacement also with Positive correlation: .

[0109] Based on the above net fatigue load, a fatigue risk assessment is performed: Calculate the time window over the past period. Static load accumulation within (e.g., 30 minutes) : Where I(·) is an indicator function, when (The load threshold, a critical value used to determine whether the condition is met, is 1 when the load is met and 0 otherwise; this formula quantifies the time a user is in a high “net load” state. The net load threshold (valued at 0.5) is chosen to account for the offsetting effect of physiological adaptation gains.

[0110] Proactive intervention: Personalized intervention is triggered in two ways:

[0111] Real-time intervention: When the real-time net fatigue load Exceeding its safety threshold During this time, the system will fine-tune the support and heat therapy parameters in real time.

[0112] Periodic intervention: when net fatigue load accumulates Exceeding the safety threshold When the timer reaches 15 minutes (equivalent to 15 minutes of high net load), a preset intervention program to relieve fatigue is automatically triggered and executed. This program includes cyclical support tension and relaxation, and pulsed heat therapy.

[0113] The cyclical support tension and relaxation specifically includes: controlling graphene electrothermal drive elements in different areas of the waist and abdomen to alternately deform and relax according to a preset timing and waveform, forming a dynamic wave-like pressure massage around the waist. Furthermore, the cycle and amplitude of this movement are determined according to… The values ​​are adjusted linearly or non-linearly. The waveform includes an axial traction waveform: controlling drive elements located in symmetrical zones on both sides of the lumbar spine to produce synchronous deformation perpendicular to the body surface. Upper limit of traction displacement. With traction tolerance gain Positive correlation: .

[0114] Based on Example 3, this embodiment introduces a user physiological adaptability model to achieve personalized load assessment. This embodiment enables the system to learn individual differences among users, providing more accurate personalized interventions and avoiding overprotection or underprotection.

[0115] Example 5: The difference between this example and Example 4 is that the comprehensive adaptive control step also includes multi-objective collaborative optimization: a multi-objective optimization model with the constructed support effect index, thermotherapy comfort index, user tolerance index and traction safety index as the core.

[0116] The target value of the support effect index is positively correlated with the net fatigue load; the threshold of the user tolerance index is positively correlated with the physiological adaptability gain; the preferred range of the thermotherapy comfort index is dynamically calculated based on the body surface temperature signal and the metabolic improvement requirements of the net fatigue load; the traction safety index is jointly determined by the user's tolerance gain to traction intervention, real-time spinal curvature status, and electromyographic tension, and is used to constrain the range of axial traction parameters.

[0117] The method further includes, based on the value and trend of the net fatigue load, utilizing the reversible switching characteristic of the elastic modulus of the graphene-based phase change elastomer film in the graphene electrothermal drive element near 28.5°C, dynamically adjusting the rigidity or flexibility of the film when performing the wave-like pressure massage, so as to match the pressing texture required for different fatigue stages.

[0118] This embodiment, based on embodiment 4, introduces a multi-objective optimization framework into the control decision-making process.

[0119] Multi-objective collaborative optimization: After generating the initial regulatory intent, an optimization layer is introduced; four optimization objective indices are defined:

[0120] Support effect index The objective is to maximize the expected decrease in net fatigue load. ;

[0121] Heat therapy comfort index The goal is to raise the body surface temperature. Stable within a personalized comfort zone Inside.

[0122] User Tolerance Index The objective is to minimize the discomfort caused by the intervention, with its threshold and adaptive gain. Positive correlation, meaning that the higher the tolerance, the higher the upper limit of acceptable intervention intensity.

[0123] Traction safety index The objective is to apply traction safely. When muscle tension M(t) is too high or When too low, Reduce, strictly constrain .

[0124] The optimization problem can be formalized as: finding the optimal set of control parameters. This maximizes the weighted comprehensive objective function: ,in Let be the dynamic weights of each index, and satisfy . (Minimum safety line) and other constraints. Weights Dynamic allocation based on the current primary status (e.g., whether the focus is on recovery or comfort).

[0125] Refined execution involving material properties: When performing wave-like pressure massage, based on optimized parameters... and the current net fatigue load The system will precisely control the operating temperature of the graphene-based phase change elastomer film. When a firm, deep press is required, control... Slightly above the phase transition point (e.g., 29.5°C) keeps the membrane in a low-modulus elastic state; when gentle, soothing pressure is required, control... Slightly below the phase transition point (e.g., 27.5°C), the film is in a high-modulus rigid state, thus dynamically matching the pressing texture.

[0126] This embodiment systematically solves the conflict between multiple objectives such as support, thermotherapy, traction, comfort, and safety through multi-objective optimization, and achieves the global optimal or satisfactory decision within the safety boundary. By utilizing the phase change characteristics of materials, the optimization decision is accurately mapped to the physical interactive texture.

[0127] Example 6: Refer to Figure 2 The difference between this embodiment and embodiment 5 is that it also includes a weight adaptation step, specifically including:

[0128] During the intervention period, the rate of decrease of the net fatigue load is calculated in real time as the support effect feedback value, the stability of the body surface temperature signal in the thermotherapy area is monitored as the thermotherapy comfort feedback value, and the proportion of high-frequency components representing discomfort in the electromyographic signal is used as the tolerance feedback value.

[0129] The above feedback values ​​are compared with their respective dynamic benchmark values, and the magnitude and rate of weight adjustment are individually constrained based on the user's individual recovery characteristics revealed by the fatigue-adaptation dynamic model.

[0130] This embodiment, based on embodiment 5, adds online adaptive weighting capabilities based on real-time feedback to the optimization model.

[0131] At the end of each intervention cycle (e.g., a 10-minute intervention procedure), the system calculates the actual feedback value:

[0132] Support effect feedback value Calculate the net fatigue load before and after the intervention. The rate of change.

[0133] Thermotherapy Comfort Feedback Value Calculate body surface temperature during intervention. In the target range The percentage of time spent inside.

[0134] Tolerance Feedback Value : Calculate the average power percentage of high-frequency components (e.g., 150-400Hz) that characterize discomfort in electromyographic signals during the intervention period.

[0135] Feedback vector The target value predicted by the optimization model The comparison generates an error vector ΔF. Based on ΔF and the user's individual recovery characteristics (such as the recovery rate factor η) revealed by the fatigue-adaptation dynamic model, the target weights for the next intervention cycle are determined. Adjustments are made. The adjustment rules tend to strengthen the weights of those that have not met expectations and are constrained by η: for users with slow recovery, the weight adjustment magnitude Δω is smaller and the rate is slower to ensure the gentleness of the strategy change. That is: ,in δ is the learning rate, clip is the clipping function, and δ is the maximum adjustment step size.

[0136] This embodiment makes the system's multi-objective optimization strategy no longer static, but capable of closed-loop feedback and self-adjustment based on the actual effect of each intervention, and respects the user's personal physiological rhythm, thus realizing the continuous personalized evolution of the control strategy.

[0137] This application also discloses a rehabilitation device.

[0138] A rehabilitation device includes: a processor, and a memory communicatively connected to the processor;

[0139] The memory is provided with a computer-readable storage medium, and a computer program is stored on the computer-readable storage medium.

[0140] When the processor processes the computer program stored on the computer-readable storage medium, it implements a graphene film-based intelligent lumbar support control method.

[0141] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A rehabilitation device, characterized in that: include: A processor, and a memory communicatively connected to the processor; The memory is provided with a computer-readable storage medium, and a computer program is stored on the computer-readable storage medium. When the processor processes the computer program, it performs the following steps: Multimodal data acquisition: Collect multimodal physiological and kinematic signals of the user's waist and abdomen. The multimodal physiological and kinematic signals include at least the circumferential strain signal of the waist and abdomen, the body surface temperature signal, and the electromyographic signal. Intelligent state analysis: The multimodal physiological and motion signals are input into a pre-trained intelligent control model for analysis and processing, and output a comprehensive assessment result of the user's lumbar load, muscle tension and thermal comfort. Based on the output result, the user's current lumbar physiological state, specific activity mode and need for implementing heat therapy are analyzed. Comprehensive Adaptive Regulation: Based on the analysis of the lumbar physiological state, specific activity patterns, spinal traction requirements, and thermotherapy requirements, a coordinated control command is generated to synchronously adjust the mechanical deformation parameters, axial traction parameters, and thermotherapy parameters of the graphene electrothermal drive element integrated in the flexible substrate. The intelligent state analysis step also includes physiological adaptive modeling: A fatigue-adaptation dynamic model is built into the intelligent control model based on the user's historical motion curves. The fatigue-adaptation dynamic model quantifies the physiological adaptation gain of the user's lumbar muscle group to a specific activity pattern by analyzing the user's repeated exposure history to that pattern. The fatigue-adaptation dynamic model also quantifies the user's tolerance and effectiveness gain to traction intervention by analyzing the user's immediate response and relaxation rate of electromyographic signals when axial traction intervention is applied. Fatigue risk assessment: Based on the long-term continuous electromyographic and postural signals, the static load accumulation of the lumbar muscles is analyzed; Active intervention: When the accumulated static load exceeds the safety threshold, a preset intervention program to relieve muscle fatigue is automatically triggered and executed. The intervention program includes periodic support tension and relaxation cycles and pulsed heat therapy.

2. The rehabilitation device according to claim 1, characterized in that: The intelligent state analysis step also includes: Action pattern recognition: Based on the temporal variation characteristics of strain signals and electromyographic signals, the intelligent control model identifies the specific actions that the user is performing, including standing still, walking, sitting, bending over, or crunching.

3. The rehabilitation device according to claim 2, characterized in that: When the user is detected to be performing abdominal crunches, the comprehensive adaptive control steps include: Dynamic deformation assistance: The graphene electrothermal drive element is controlled to generate a preset curling deformation inward in the corresponding area on the front side of the user's abdomen to provide dynamic auxiliary support. Spinal extension assistance: During the recovery phase of the crunch, a graphene electrothermal drive element located in the posterior region of the lumbar spine is controlled to produce a gentle, axial extension deformation to assist the spine in extending segment by segment and reduce pressure on the posterior intervertebral discs. Synergistic thermotherapy enhancement: Simultaneously increase the far-infrared thermotherapy output power of the corresponding area to cooperate with muscle movement and promote metabolism.

4. The rehabilitation device according to claim 1, characterized in that: The multimodal data acquisition step also includes acquiring humidity signals from the skin surface of the waist; the comprehensive adaptive control step also includes: Wearable microenvironment management: When the humidity signal exceeds a preset threshold, the graphene electrothermal drive element is controlled to switch to a low-power ventilation and dehumidification mode, which promotes air circulation and moisture evaporation while maintaining basic heat sensation.

5. The rehabilitation device according to claim 1, characterized in that: The periodic support tension-relaxation cycle specifically includes: controlling graphene electrothermal drive elements in different areas of the waist and abdomen to alternately deform and relax according to a preset time sequence and waveform, forming a dynamic wave-like pressure massage around the waist; and the cycle and amplitude of this cycle are adjusted linearly or non-linearly according to the value of the degree of fatigue accumulation. The waveform includes an axial traction waveform: controlling the drive elements located in the symmetrical partitions on both sides of the lumbar spine to produce synchronous deformation perpendicular to the body surface.

6. The rehabilitation device according to claim 5, characterized in that: In the intelligent state analysis step, the comprehensive evaluation results include net fatigue load based on a physiological adaptation model; When calculating the static load accumulation level in real time, the original load generated by the current activity mode is subtracted from the real-time physiological adaptation gain corresponding to that mode to obtain the net fatigue load, and intervention is triggered or adjusted based on this net fatigue load.

7. The rehabilitation device according to claim 6, characterized in that: The comprehensive adaptive control step also includes multi-objective collaborative optimization: constructing a multi-objective optimization model with the support effect index, thermotherapy comfort index, user tolerance index and traction safety index as the core; The target value of the support effect index is positively correlated with the net fatigue load; the threshold of the user tolerance index is positively correlated with the physiological adaptability gain; the preferred range of the thermotherapy comfort index is dynamically calculated based on the body surface temperature signal and the metabolic improvement requirements of the net fatigue load; the traction safety index is jointly determined by the user's tolerance gain to traction intervention, real-time spinal curvature status, and electromyographic tension, and is used to constrain the range of axial traction parameters. Based on the value and trend of the net fatigue load, the reversible switching characteristic of the elastic modulus of the graphene-based phase change elastomer film in the graphene electrothermal drive element at around 28.5℃ is used to dynamically adjust the rigidity or flexibility of the film when performing the dynamic wave-like pressure massage, so as to match the pressing texture required for different fatigue stages.

8. The rehabilitation device according to claim 7, characterized in that: It also includes a weight adaptation step, specifically including: During the intervention period, the rate of decrease of the net fatigue load is calculated in real time as the support effect feedback value, the stability of the body surface temperature signal in the thermotherapy area is monitored as the thermotherapy comfort feedback value, and the proportion of high-frequency components representing discomfort in the electromyographic signal is used as the tolerance feedback value. The above feedback values ​​are compared with their respective dynamic benchmark values, and the magnitude and rate of weight adjustment are individually constrained based on the user's individual recovery characteristics revealed by the fatigue-adaptation dynamic model.

Images