Intelligent orthosis adaptive pressure control method based on subcutaneous pressure prediction model

Abstract

The application relates to an intelligent brace adaptive pressure control method and device based on a subcutaneous pressure prediction model, wherein the method comprises the following steps: collecting surface pressure distribution data between a brace and a preset key area of skin in real time; extracting a pressure peak value, a pressure gradient, a pressure duration and a pressure change rate, and fusing the collected physiological parameters of a user to construct a multi-modal feature vector; inputting the multi-modal feature vector into a pre-trained subcutaneous pressure prediction model to predict a subcutaneous pressure experience value of the user; and adjusting the inflation amount of an air bag inside the body of the intelligent brace if the subcutaneous pressure experience value exceeds a preset safety pressure threshold. The application can predict the subcutaneous pressure experience value by collecting surface pressure data in real time and combining physiological parameters, can give an early warning before actual subcutaneous tissue damage occurs, and can cope with pressure changes caused by user activities and posture changes by dynamically adjusting the inflation amount of the air bag.

Description

Technical Field

[0001] This invention relates to the field of medical device auxiliary technology, specifically to an intelligent brace adaptive pressure control method and device, and computer equipment based on a subcutaneous pressure prediction model. Background Technology

[0002] Braces are widely used in orthopedic rehabilitation, sports injury prevention, and posture correction. However, traditional braces often have a fixed design, which makes it difficult to adapt to individual differences and dynamic physiological changes, resulting in uneven pressure distribution and easily causing problems such as pressure sores and nerve compression, affecting patient comfort and treatment outcomes.

[0003] Currently, most existing braces employ a multi-layered structure, with an outer layer providing support and an inner layer using soft padding material to distribute pressure. However, this method is inherently static and cannot be adjusted according to the user's real-time physiological state and activity level. While some braces integrate pressure sensors to monitor the pressure between the brace and the skin, intervention only occurs when the pressure exceeds a preset value, such as through alarms or manual adjustment of the airbag inflation. For example, patent CN112535862A discloses a smart brace for scoliosis correction that uses pressure sensors to monitor skin pressure and adjusts airbag pressure via a controller. While achieving some pressure control, the lack of prediction and active adjustment of subcutaneous pressure easily leads to over-inflation or under-inflation, affecting the control effect. The pressure control strategy is also simplistic: existing smart brace pressure control systems typically employ simple threshold control strategies, lacking prediction and active adjustment of pressure changes, which easily results in poor control performance.

[0004] To address the aforementioned issues, this invention proposes an intelligent brace adaptive pressure control method based on a subcutaneous pressure prediction model. This method predicts the subcutaneous pressure experience value by collecting surface pressure data in real time and combining it with physiological parameters, providing early warning before actual subcutaneous tissue damage occurs. Furthermore, it dynamically adjusts the airbag inflation volume to cope with pressure changes caused by user activities and posture changes. Summary of the Invention

[0005] In view of the above problems, the present invention provides an intelligent brace adaptive pressure control method and device, and computer equipment based on a subcutaneous pressure prediction model.

[0006] According to one aspect of the present invention, an adaptive pressure control method for intelligent braces based on a subcutaneous pressure prediction model is provided, comprising:

[0007] A flexible pressure sensor array strategically positioned between the smart brace and the user's skin in a pre-defined key area is used to collect surface pressure distribution data between the brace and the skin in real time.

[0008] The pressure peak, pressure gradient, pressure duration, and pressure change rate are extracted from the surface pressure distribution data and fused with the collected physiological parameters of the user to construct a multimodal feature vector; wherein the physiological parameters include age, BMI, skin moisture, and local blood flow.

[0009] The multimodal feature vectors are input into a pre-trained subcutaneous pressure prediction model to predict the user's subcutaneous pressure experience value; wherein, the subcutaneous pressure prediction model is interpreted using the SHAP method to determine the contribution of multimodal features to subcutaneous pressure;

[0010] The subcutaneous pressure experience value is compared with a preset safe pressure threshold. If the subcutaneous pressure experience value exceeds the preset safe pressure threshold, the inflation volume of the airbag inside the smart brace is adjusted to change the pressure distribution in the preset key area, thereby reducing the subcutaneous pressure experience value.

[0011] In one alternative embodiment, the smart brace body includes:

[0012] Support structure, used to adjust the shape and size of the support using molding materials;

[0013] A flexible pressure sensor array is strategically arranged on the inner surface of the smart brace body that contacts a preset key area of ​​the user's skin, wherein the preset key area includes bony prominences and areas of concentrated force.

[0014] At least one airbag, used to adjust local pressure by inflation or deflation, is located on the inner surface of the smart brace body and corresponds to the preset key area;

[0015] At least one air pump is connected to the airbag via an air tube and controls the inlet and outlet of gas via a solenoid valve to provide inflating and deflating power for the airbag;

[0016] A microcontroller unit is connected to a flexible pressure sensor array to receive pressure data and to the air pump and the solenoid valve to control air pressure regulation.

[0017] The wireless communication unit communicates with external devices via Bluetooth or Wi-Fi;

[0018] The power supply is connected to the microcontroller unit.

[0019] In one alternative approach, the multimodal coupling control equation for dynamically adjusting the inflation volume of the airbag is:

[0020]

[0021] in, The volume change during time period t; Preset safety threshold; This represents the real-time subcutaneous pressure reading. ; ; ; ; The rate of change in skin moisture; This is a local blood flow velocity index; The pressure gradient modulus; This refers to the joint's range of motion. It is a dynamic decay index; is a respiratory rate-related constant; BMI is a correction factor.

[0022] In one alternative approach, the signal acquisition frequency of the flexible pressure sensor array is:

[0023]

[0024] in, The reference sampling frequency; This is the pressure response coefficient; For pressure gradient; The modulation frequency; For phase shift; The average pressure attenuation coefficient; This represents the average pressure value over the time period t.

[0025] In one alternative approach, after adjusting the inflation level of the airbag, the method further includes:

[0026] Based on historical pressure change data, airbag inflation volume, and user physiological parameters, the predicted pressure distribution is obtained;

[0027] Assess whether the predicted pressure distribution causes new high-pressure or low-pressure areas. If it does, fine-tune the airbag inflation volume.

[0028] In one alternative approach, the effectiveness evaluation function for the multimodal feature vectors is constructed as follows:

[0029]

[0030] in, This is the effectiveness evaluation value of the i-th feature; For characteristic duration; Weights for pressure features; Weights for physiological characteristics; Confidence level for pressure characteristics; Confidence level for physiological characteristics; ; ; The signal-to-noise ratio of the pressure signal; Local blood flow index; Let i be the i-th independent feature component.

[0031] In one alternative approach, the subcutaneous pressure prediction model includes an inception-v4 network layer, an averagereduce network layer, a graph attention module, an augmented image layer, a simulated image layer, an adaptation module, and a prediction module.

[0032] The Inception-v4 network layer uses multiple convolutional kernels of different sizes to extract deep and multi-scale features from the input surface pressure distribution data, so as to capture the pressure change patterns of different receptive fields and output multi-scale pressure feature maps.

[0033] The Average Reduce network layer is used to compress the dimension of the feature map extracted by Inception-v4 and output a pressure feature vector.

[0034] The graph attention module is used to establish the relationship between each sensor node in the pressure sensor array and to highlight the influence of key nodes on subcutaneous pressure using an attention mechanism. The input is a compressed pressure feature vector and the spatial location information of the pressure sensor array, and the output is an enhanced pressure feature vector.

[0035] The expanded image layer is used to expand the enhanced pressure feature vector and output the expanded pressure feature vector.

[0036] The simulated image layer is used to simulate the expanded stress feature vector through a generative adversarial network (GAN) and output the simulated stress feature vector.

[0037] The adaptation module is used to fuse physiological parameters with simulated pressure feature vectors and output a fused feature vector.

[0038] The prediction module is used to predict the user's subcutaneous pressure experience value.

[0039] In one alternative embodiment, the adapter module includes a pressure flow encoder and a physiological flow encoder;

[0040] The pressure flow encoder uses a Transformer network to perform self-attention encoding on the simulated pressure feature vector to generate a pressure context vector.

[0041] The physiological flow encoder performs time-series encoding of physiological parameter sequences through a gated loop unit to generate a physiological context vector.

[0042] According to another aspect of the present invention, an intelligent brace adaptive pressure control device based on a subcutaneous pressure prediction model is provided, comprising:

[0043] The pressure data acquisition module is used to collect surface pressure distribution data between the smart brace and the user's skin in real time through a flexible pressure sensor array strategically arranged in the smart brace body and the user's skin in the preset key areas of the skin.

[0044] A multimodal feature construction module is used to extract pressure peak, pressure gradient, pressure duration and pressure change rate from the surface pressure distribution data, and fuse them with the collected physiological parameters of the user to construct a multimodal feature vector; wherein, the physiological parameters include age, BMI, skin moisture and local blood flow;

[0045] The subcutaneous pressure prediction module is used to input the multimodal feature vector into a pre-trained subcutaneous pressure prediction model to predict the user's subcutaneous pressure experience value; wherein, the subcutaneous pressure prediction model uses the SHAP method to interpret the model to determine the contribution of multimodal features to subcutaneous pressure;

[0046] An adaptive pressure adjustment module is used to compare the subcutaneous pressure experience value with a preset safe pressure threshold. If the subcutaneous pressure experience value exceeds the preset safe pressure threshold, the inflation volume of the airbag inside the smart brace body is adjusted to change the pressure distribution of the preset key area, thereby reducing the subcutaneous pressure experience value.

[0047] According to another aspect of the present invention, a computer device is provided, comprising: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other through the communication bus;

[0048] The memory is used to store at least one executable instruction, which causes the processor to perform the operation corresponding to the above-described intelligent brace adaptive pressure control method based on the subcutaneous pressure prediction model.

[0049] According to the solution provided by the present invention, a flexible pressure sensor array strategically positioned between the smart brace body and a preset key area of ​​the user's skin collects surface pressure distribution data between the brace and the preset key area of ​​the skin in real time. Pressure peak value, pressure gradient, pressure duration, and pressure change rate are extracted from the surface pressure distribution data and fused with the collected physiological parameters of the user to construct a multimodal feature vector. The physiological parameters include age, BMI, skin moisture, and local blood flow. The multimodal feature vector is input into a pre-trained subcutaneous pressure prediction model to predict the user's subcutaneous pressure experience value. The subcutaneous pressure prediction model is interpreted using the SHAP method to determine the contribution of the multimodal features to the subcutaneous pressure. The subcutaneous pressure experience value is compared with a preset safe pressure threshold. If the subcutaneous pressure experience value exceeds the preset safe pressure threshold, the inflation volume of the airbag inside the smart brace body is adjusted to change the pressure distribution in the preset key area, thereby reducing the subcutaneous pressure experience value. This invention predicts subcutaneous pressure experience values ​​by collecting surface pressure data in real time and combining it with physiological parameters. It can provide early warning before actual subcutaneous tissue damage occurs and respond to pressure changes caused by user activities and posture changes by dynamically adjusting the airbag inflation volume.

[0050] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and in order to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0051] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0052] Figure 1 A flowchart illustrating the adaptive pressure control method for intelligent braces based on a subcutaneous pressure prediction model according to an embodiment of the present invention is shown.

[0053] Figure 2 A schematic diagram of the framework of an intelligent brace adaptive pressure control device based on a subcutaneous pressure prediction model according to an embodiment of the present invention is shown.

[0054] Figure 3 A schematic diagram of the structure of a computer device according to an embodiment of the present invention is shown. Detailed Implementation

[0055] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this invention will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

[0056] Figure 1 A flowchart illustrating the adaptive pressure control method for intelligent braces based on a subcutaneous pressure prediction model, according to an embodiment of the present invention, is shown. Specifically, as... Figure 1 As shown, it includes the following steps:

[0057] Step S101: A flexible pressure sensor array strategically positioned between the smart brace body and the user's skin in a preset key area is used to collect surface pressure distribution data between the brace and the preset key area of ​​the skin in real time.

[0058] In this embodiment, a flexible pressure sensor array is strategically positioned in key areas between the smart brace body and the user's skin to collect surface pressure distribution data between the brace and skin in real time. This ensures timely response to pressure changes and dynamically adjusts the inflation volume of the airbag inside the brace to optimize pressure distribution and reduce the risk of pressure injuries. Specifically, the flexible pressure sensor array, strategically positioned within the smart brace body, covers key areas of contact between the brace and skin, such as bony prominences and areas of concentrated stress.

[0059] Strategically arranging the brace allows for effective monitoring of pressure distribution between the brace and skin, enabling timely detection of pressure concentration areas and changes. This allows for dynamic adjustment of the brace, reducing the risk of skin injury. For example, a strategically arranged ankle brace focuses on key areas such as the medial and lateral malleoli, the Achilles tendon, and the contact point between the brace and the dorsum of the foot. High-density circular or elliptical sensors are used at the medial and lateral malleoli, closely surrounding them to ensure coverage of the entire bony prominence area. Strip sensors are used at the Achilles tendon, extending along the tendon to monitor pressure on it. Mesh or matrix sensors are used on the dorsum of the foot to cover the area in contact with the brace, monitoring the overall pressure distribution.

[0060] In one alternative embodiment, the smart brace body includes:

[0061] Support structure, used to adjust the shape and size of the support using molding materials;

[0062] A flexible pressure sensor array is strategically arranged on the inner surface of the smart brace body that contacts a preset key area of ​​the user's skin, wherein the preset key area includes bony prominences and areas of concentrated force.

[0063] At least one airbag, used to adjust local pressure by inflation or deflation, is located on the inner surface of the smart brace body and corresponds to the preset key area;

[0064] At least one air pump is connected to the airbag via an air tube and controls the inlet and outlet of gas via a solenoid valve to provide inflating and deflating power for the airbag;

[0065] A microcontroller unit is connected to a flexible pressure sensor array to receive pressure data and to the air pump and the solenoid valve to control air pressure regulation.

[0066] The wireless communication unit communicates with external devices via Bluetooth or Wi-Fi;

[0067] The power supply is connected to the microcontroller unit.

[0068] In this embodiment, the combination of an airbag and an air pump for inflation and deflation regulates local pressure, enabling precise control of the pressure distribution between the brace and the skin, thus preventing discomfort or injury caused by excessive or insufficient pressure. The integration of a microcontroller unit and a wireless communication unit allows the brace to communicate with external devices for remote monitoring and adjustment. The support structure is made of moldable material, allowing the shape and size of the brace to be adjusted according to the user's specific needs, providing more personalized comfort.

[0069] For example, a user needs to wear a smart brace for rehabilitation training. The support structure of the smart brace is adjusted according to the user's body shape, collects data from a flexible pressure sensor array, and the microcontroller unit collects pressure data in real time and fuses it with the user's physiological parameters. The multimodal feature vector is then input into a pre-trained subcutaneous pressure prediction model to predict the subcutaneous pressure experience value. If the subcutaneous pressure experience value exceeds a preset safe pressure threshold, the microcontroller unit controls the air pump and solenoid valve to adjust the inflation volume of the airbag. By changing the inflation volume of the airbag, the pressure distribution between the brace and the skin is adjusted, reducing the subcutaneous pressure experience value. The user or medical personnel can communicate with the smart brace through external devices (such as mobile phones or tablets) to view pressure data and adjustment status in real time, and remotely adjust the pressure distribution of the brace as needed to ensure user comfort and rehabilitation effectiveness.

[0070] In one alternative approach, the multimodal coupling control equation for dynamically adjusting the inflation volume of the airbag is:

[0071]

[0072] in, The volume change during time period t; Preset safety threshold; This represents the real-time subcutaneous pressure reading. ; ; ; ; The rate of change in skin moisture; This is a local blood flow velocity index; The pressure gradient modulus; This refers to the joint's range of motion. It is a dynamic decay index; is a respiratory rate-related constant; BMI is a correction factor.

[0073] In this embodiment, not only is the deviation between the target pressure and the current pressure considered, but also physiological parameters such as skin moisture change rate, local blood flow velocity index, pressure gradient modulus, joint range of motion, respiratory rate, and BMI, as well as environmental factors, are integrated to achieve multimodal information fusion and improve the adaptability of pressure regulation. The proportional-integral control proportional term reflects the immediate impact of the current pressure deviation; among which, The integral term considers the cumulative effect of pressure deviation, allowing the actual pressure to approach the target pressure more quickly and stably. BMI is used as a correction factor, combined with other physiological parameters, to personalize adjustments based on individual user differences, thus dynamically adjusting the airbag inflation volume according to the user's BMI. The exponential decay term considers the effect of time on inflation volume. Over time, the inflation volume is gradually reduced to avoid excessive pressure or to allow the body to gradually adapt.

[0074] In one alternative approach, the signal acquisition frequency of the flexible pressure sensor array is:

[0075]

[0076] in, The reference sampling frequency; This is the pressure response coefficient; For pressure gradient; The modulation frequency; For phase shift; The average pressure attenuation coefficient; This represents the average pressure value over the time period t.

[0077] In this embodiment, the sampling frequency is dynamically and adaptively adjusted using the pressure gradient, its rate of change, and the average pressure value. When pressure changes drastically, the sampling frequency is increased to capture more detailed pressure information. The pressure gradient and its rate of change allow for more sensitive detection of rapid pressure changes, enabling timely adjustment of the airbag inflation volume and preventing skin damage. The average pressure value allows for a reduction in the sampling frequency in pressure-stable regions, thereby reducing the overall power consumption of the smart brace during prolonged wear and minimizing data storage requirements.

[0078] In one alternative approach, after adjusting the inflation level of the airbag, the method further includes:

[0079] Based on historical pressure change data, airbag inflation volume, and user physiological parameters, the predicted pressure distribution is obtained;

[0080] Assess whether the predicted pressure distribution causes new high-pressure or low-pressure areas. If it does, fine-tune the airbag inflation volume.

[0081] In this embodiment, after actually adjusting the airbag inflation volume, instead of directly waiting for new pressure data feedback, the impact of the adjustment is evaluated through prediction. This proactively prevents the emergence of new high-pressure or low-pressure areas due to improper airbag adjustment, thereby avoiding the risk of secondary damage.

[0082] Step S102: Extract the pressure peak, pressure gradient, pressure duration and pressure change rate from the surface pressure distribution data, and fuse them with the collected physiological parameters of the user to construct a multimodal feature vector; wherein, the physiological parameters include age, BMI, skin humidity and local blood flow.

[0083] In this embodiment, pressure features such as peak pressure, pressure gradient, pressure duration, and pressure change rate are extracted to capture the static and dynamic characteristics of pressure distribution, thus more effectively reflecting the impact of pressure on the skin. Physiological parameters include age, BMI, skin moisture, and local blood flow. Personalized modeling can be performed using these physiological parameters. For example, older people have thinner skin and are more sensitive to pressure; people with higher BMIs may experience different local stress conditions; skin moisture and local blood flow directly affect the skin's pressure resistance and repair capabilities.

[0084] In one alternative approach, the effectiveness evaluation function for the multimodal feature vectors is constructed as follows:

[0085]

[0086] in, This is the effectiveness evaluation value of the i-th feature; For characteristic duration; Weights for pressure features; Weights for physiological characteristics; Confidence level for pressure characteristics; Confidence level for physiological characteristics; ; ; The signal-to-noise ratio of the pressure signal; Local blood flow index; Let i be the i-th independent feature component.

[0087] In this embodiment, This study considers the signal-to-noise ratio of the pressure signal. When the signal quality is poor, the confidence of the pressure feature decreases, thereby reducing the impact of noise on the evaluation of the feature's effectiveness. This factor considers the local blood flow index, which reflects the blood circulation status of the skin. Blood flow status is closely related to the skin's sensitivity to pressure, thus helping to improve the accuracy of predictions.

[0088] Step S103: Input the multimodal feature vector into a pre-trained subcutaneous pressure prediction model to predict the user's subcutaneous pressure experience value; wherein, the subcutaneous pressure prediction model uses the SHAP method for model interpretation to determine the contribution of multimodal features to subcutaneous pressure.

[0089] In this embodiment, the SHAP (Shapley Additive Explanations) method is a game theory-based model interpretation method that can provide the contribution of each feature to the model output. The SHAP method can clearly reveal the specific impact of each multimodal feature (such as pressure peak, pressure gradient, physiological parameters, etc.) on subcutaneous pressure prediction, thereby improving the interpretability of the model.

[0090] In one alternative approach, the subcutaneous pressure prediction model includes an inception-v4 network layer, an averagereduce network layer, a graph attention module, an augmented image layer, a simulated image layer, an adaptation module, and a prediction module.

[0091] The Inception-v4 network layer uses multiple convolutional kernels of different sizes to extract deep and multi-scale features from the input surface pressure distribution data, so as to capture the pressure change patterns of different receptive fields and output multi-scale pressure feature maps.

[0092] The Average Reduce network layer is used to compress the dimension of the feature map extracted by Inception-v4 and output a pressure feature vector.

[0093] The graph attention module is used to establish the relationship between each sensor node in the pressure sensor array and to highlight the influence of key nodes on subcutaneous pressure using an attention mechanism. The input is a compressed pressure feature vector and the spatial location information of the pressure sensor array, and the output is an enhanced pressure feature vector.

[0094] The expanded image layer is used to expand the enhanced pressure feature vector and output the expanded pressure feature vector.

[0095] The simulated image layer is used to simulate the expanded stress feature vector through a generative adversarial network (GAN) and output the simulated stress feature vector.

[0096] The adaptation module is used to fuse physiological parameters with simulated pressure feature vectors and output a fused feature vector.

[0097] The prediction module is used to predict the user's subcutaneous pressure experience value.

[0098] In this embodiment, the Inception-v4 network layer utilizes convolutional kernels of different sizes to capture pressure change patterns in different receptive fields, enabling a better understanding of the complexity of pressure distribution. For example, small convolutional kernels capture localized pressure concentration areas, while large kernels capture the overall pressure distribution trend. The graph attention module identifies the influence of key nodes (e.g., pressure peak locations) on subcutaneous pressure by establishing relationships between pressure sensor nodes. The extended image layer and simulated image layer utilize generative adversarial networks for data augmentation, generating more diverse simulated pressure distribution data that can adapt to the physiological characteristics and usage scenarios of different users. The Average Reduce network layer performs feature vector compression, reducing computational cost and preventing overfitting. The graph attention module enhances the compressed feature vectors, highlighting key pressure features.

[0099] In one alternative embodiment, the adapter module includes a pressure flow encoder and a physiological flow encoder;

[0100] The pressure flow encoder uses a Transformer network to perform self-attention encoding on the simulated pressure feature vector to generate a pressure context vector.

[0101] The physiological flow encoder performs time-series encoding of physiological parameter sequences through a gated loop unit to generate a physiological context vector.

[0102] In this embodiment, the pressure flow encoder converts the simulated pressure feature vector into a low-dimensional embedding vector through an embedding layer, learns the relationships between features from different perspectives through a multi-head self-attention layer, and performs a nonlinear transformation on the output of the self-attention layer through a feedforward neural network to increase the model's expressive power. The physiological flow encoder converts each parameter in the physiological parameter sequence into a low-dimensional embedding vector through an embedding layer, processes the physiological parameter sequence cyclically through a GRU layer to learn temporal information, and uses the GRU hidden state of the last time step as the physiological context vector in the output layer.

[0103] Step S104: Compare the subcutaneous pressure experience value with a preset safe pressure threshold. If the subcutaneous pressure experience value exceeds the preset safe pressure threshold, adjust the inflation amount of the airbag inside the smart brace body to change the pressure distribution of the preset key area, thereby reducing the subcutaneous pressure experience value.

[0104] In this embodiment, by monitoring the subcutaneous pressure experience value in real time and comparing it with a preset safe pressure threshold, the inflation volume of the airbag inside the smart brace can be adjusted in a timely manner to dynamically adjust the pressure distribution of the brace on the user's skin, avoid damage to the skin from prolonged high pressure, and improve the user's comfort and safety.

[0105] According to the solution provided by the present invention, a flexible pressure sensor array strategically positioned between the smart brace body and a preset key area of ​​the user's skin collects surface pressure distribution data between the brace and the preset key area of ​​the skin in real time. Pressure peak value, pressure gradient, pressure duration, and pressure change rate are extracted from the surface pressure distribution data and fused with the collected physiological parameters of the user to construct a multimodal feature vector. The physiological parameters include age, BMI, skin moisture, and local blood flow. The multimodal feature vector is input into a pre-trained subcutaneous pressure prediction model to predict the user's subcutaneous pressure experience value. The subcutaneous pressure prediction model is interpreted using the SHAP method to determine the contribution of the multimodal features to the subcutaneous pressure. The subcutaneous pressure experience value is compared with a preset safe pressure threshold. If the subcutaneous pressure experience value exceeds the preset safe pressure threshold, the inflation volume of the airbag inside the smart brace body is adjusted to change the pressure distribution in the preset key area, thereby reducing the subcutaneous pressure experience value. This invention predicts subcutaneous pressure experience values ​​by collecting surface pressure data in real time and combining it with physiological parameters. It can provide early warning before actual subcutaneous tissue damage occurs and respond to pressure changes caused by user activities and posture changes by dynamically adjusting the airbag inflation volume.

[0106] Figure 2A schematic diagram of the framework of an intelligent brace adaptive pressure control device based on a subcutaneous pressure prediction model according to an embodiment of the present invention is shown. The intelligent brace adaptive pressure control device based on a subcutaneous pressure prediction model includes:

[0107] The pressure data acquisition module 210 is used to acquire surface pressure distribution data between the smart brace and the preset key areas of the skin in real time through a flexible pressure sensor array strategically arranged in the smart brace body and the preset key areas of the user's skin.

[0108] The multimodal feature construction module 220 is used to extract pressure peak, pressure gradient, pressure duration and pressure change rate from the surface pressure distribution data, and fuse them with the collected physiological parameters of the user to construct a multimodal feature vector; wherein, the physiological parameters include age, BMI, skin moisture and local blood flow;

[0109] The subcutaneous pressure prediction module 230 is used to input the multimodal feature vector into a pre-trained subcutaneous pressure prediction model to predict the user's subcutaneous pressure experience value; wherein, the subcutaneous pressure prediction model uses the SHAP method to interpret the model to determine the contribution of multimodal features to subcutaneous pressure;

[0110] The adaptive pressure adjustment module 240 is used to compare the subcutaneous pressure experience value with a preset safe pressure threshold. If the subcutaneous pressure experience value exceeds the preset safe pressure threshold, the inflation volume of the airbag inside the smart brace body is adjusted to change the pressure distribution of the preset key area, thereby reducing the subcutaneous pressure experience value.

[0111] Figure 3 The diagram shows a structural schematic of an embodiment of the computer device of the present invention. The specific embodiments of the present invention do not limit the specific implementation of the computer device.

[0112] like Figure 3 As shown, the computer device may include: a processor 302, a communications interface 304, a memory 306, and a communications bus 308.

[0113] The processor 302, communication interface 304, and memory 306 communicate with each other via communication bus 508. Communication interface 304 is used to communicate with other network elements such as clients or other servers. The processor 302 executes program 310, specifically performing the relevant steps in the above embodiment of the intelligent brace adaptive pressure control method based on a subcutaneous pressure prediction model.

[0114] Specifically, program 310 may include program code that includes computer operation instructions.

[0115] Processor 302 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention. The computer device includes one or more processors, which may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0116] Memory 306 is used to store program 310. Memory 306 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0117] According to the solution provided by the present invention, a flexible pressure sensor array strategically positioned between the smart brace body and a preset key area of ​​the user's skin collects surface pressure distribution data between the brace and the preset key area of ​​the skin in real time. Pressure peak value, pressure gradient, pressure duration, and pressure change rate are extracted from the surface pressure distribution data and fused with the collected physiological parameters of the user to construct a multimodal feature vector. The physiological parameters include age, BMI, skin moisture, and local blood flow. The multimodal feature vector is input into a pre-trained subcutaneous pressure prediction model to predict the user's subcutaneous pressure experience value. The subcutaneous pressure prediction model is interpreted using the SHAP method to determine the contribution of the multimodal features to the subcutaneous pressure. The subcutaneous pressure experience value is compared with a preset safe pressure threshold. If the subcutaneous pressure experience value exceeds the preset safe pressure threshold, the inflation volume of the airbag inside the smart brace body is adjusted to change the pressure distribution in the preset key area, thereby reducing the subcutaneous pressure experience value. This invention predicts subcutaneous pressure experience values ​​by collecting surface pressure data in real time and combining it with physiological parameters. It can provide early warning before actual subcutaneous tissue damage occurs and respond to pressure changes caused by user activities and posture changes by dynamically adjusting the airbag inflation volume.

[0118] Those skilled in the art will understand that modules in the device of the embodiments can be adaptively changed and placed in one or more devices different from that embodiment. Modules, units, or components in the embodiments can be combined into a single module, unit, or component, and further, they can be divided into multiple sub-modules, sub-units, or sub-components. Except where at least some of such features and / or processes or units are mutually exclusive, any combination of all features disclosed in this specification (including the accompanying claims, abstract, and drawings) and all processes or units of any method or device so disclosed can be employed. Unless expressly stated otherwise, each feature disclosed in this specification (including the accompanying claims, abstract, and drawings) may be replaced by an alternative feature that serves the same, equivalent, or similar purpose. Furthermore, those skilled in the art will understand that although some embodiments herein include certain features included in other embodiments but not others, combinations of features from different embodiments are intended to be within the scope of the invention and form different embodiments. For example, in the following claims, any of the claimed embodiments can be used in any combination. The invention can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims listing several devices, several of these devices may be embodied by the same hardware item. Unless otherwise specified, the steps in the above embodiments should not be construed as limiting the order of execution.

Claims

1. An intelligent brace adaptive pressure control system based on a subcutaneous pressure prediction model, characterized in that, include: A flexible pressure sensor array strategically positioned between the smart brace and the user's skin in a pre-defined key area is used to collect surface pressure distribution data between the brace and the skin in real time. The pressure peak, pressure gradient, pressure duration, and pressure change rate are extracted from the surface pressure distribution data and fused with the collected physiological parameters of the user to construct a multimodal feature vector; wherein the physiological parameters include age, BMI, skin moisture, and local blood flow. The multimodal feature vectors are input into a pre-trained subcutaneous pressure prediction model to predict the user's subcutaneous pressure experience value; wherein, the subcutaneous pressure prediction model is interpreted using the SHAP method to determine the contribution of multimodal features to subcutaneous pressure; The subcutaneous pressure experience value is compared with a preset safe pressure threshold. If the subcutaneous pressure experience value exceeds the preset safe pressure threshold, the inflation volume of the airbag inside the smart brace is adjusted to change the pressure distribution in the preset key area, thereby reducing the subcutaneous pressure experience value. The multimodal coupling control equation for dynamically adjusting the inflation volume of the airbag is: ； in, This represents the volume change within time period t. Preset safety threshold; This represents the real-time subcutaneous pressure reading. ; ; ; ; The rate of change in skin moisture; This is a local blood flow velocity index; The pressure gradient modulus; This refers to the joint's range of motion. It is a dynamic decay index; is a respiratory rate-related constant; BMI is a correction factor.

2. The intelligent brace adaptive pressure control system based on a subcutaneous pressure prediction model according to claim 1, characterized in that, The intelligent brace body includes: Support structure, used to adjust the shape and size of the support using molding materials; A flexible pressure sensor array is strategically arranged on the inner surface of the smart brace body that contacts a preset key area of ​​the user's skin, wherein the preset key area includes bony prominences and areas of concentrated force. At least one airbag, used to adjust local pressure by inflation or deflation, is located on the inner surface of the smart brace body and corresponds to the preset key area; At least one air pump is connected to the airbag via an air tube and controls the inlet and outlet of gas via a solenoid valve to provide inflating and deflating power for the airbag; A microcontroller unit is connected to a flexible pressure sensor array to receive pressure data and to the air pump and the solenoid valve to control air pressure regulation. The wireless communication unit communicates with external devices via Bluetooth or Wi-Fi; The power supply is connected to the microcontroller unit.

3. The intelligent brace adaptive pressure control system based on a subcutaneous pressure prediction model according to claim 1, characterized in that, The signal acquisition frequency of the flexible pressure sensor array is: ； in, The reference sampling frequency; This is the pressure response coefficient; For pressure gradient; The modulation frequency; For phase shift; The average pressure attenuation coefficient; This represents the average pressure value over the time period t.

4. The intelligent brace adaptive pressure control system based on a subcutaneous pressure prediction model according to claim 1, characterized in that, After adjusting the airbag inflation level, it also includes: Based on historical pressure change data, airbag inflation volume, and user physiological parameters, the predicted pressure distribution is obtained; Assess whether the predicted pressure distribution causes new high-pressure or low-pressure areas. If it does, fine-tune the airbag inflation volume.

5. The intelligent brace adaptive pressure control system based on a subcutaneous pressure prediction model according to claim 1, characterized in that, The subcutaneous pressure prediction model includes an Inception-v4 network layer, an Average Reduce network layer, a graph attention module, an augmented image layer, a simulated image layer, an adaptation module, and a prediction module. The Inception-v4 network layer uses multiple convolutional kernels of different sizes to extract deep and multi-scale features from the input surface pressure distribution data, in order to capture pressure change patterns in different receptive fields and output multi-scale pressure feature maps. The Average Reduce network layer is used to compress the dimension of the feature map extracted by Inception-v4 and output a pressure feature vector. The graph attention module is used to establish the relationship between each sensor node in the pressure sensor array and to highlight the influence of key nodes on subcutaneous pressure using an attention mechanism. The input is a compressed pressure feature vector and the spatial location information of the pressure sensor array, and the output is an enhanced pressure feature vector. The expanded image layer is used to expand the enhanced pressure feature vector and output the expanded pressure feature vector. The simulated image layer is used to simulate the expanded stress feature vector through a generative adversarial network (GAN) and output the simulated stress feature vector. The adaptation module is used to fuse physiological parameters with simulated pressure feature vectors and output a fused feature vector. The prediction module is used to predict the user's subcutaneous pressure experience value.

6. The intelligent brace adaptive pressure control system based on a subcutaneous pressure prediction model according to claim 5, characterized in that, The adapter module includes a pressure flow encoder and a physiological flow encoder; The pressure flow encoder uses a Transformer network to perform self-attention encoding on the simulated pressure feature vector to generate a pressure context vector. The physiological flow encoder performs time-series encoding of physiological parameter sequences through a gated loop unit to generate a physiological context vector.

7. An intelligent brace adaptive pressure control device based on a subcutaneous pressure prediction model, comprising the intelligent brace adaptive pressure control system based on a subcutaneous pressure prediction model as described in any one of claims 1-6, characterized in that, include: The pressure data acquisition module is used to collect surface pressure distribution data between the smart brace and the user's skin in real time through a flexible pressure sensor array strategically arranged in the smart brace body and the user's skin in the preset key areas of the skin. A multimodal feature construction module is used to extract pressure peak, pressure gradient, pressure duration and pressure change rate from the surface pressure distribution data, and fuse them with the collected physiological parameters of the user to construct a multimodal feature vector; wherein, the physiological parameters include age, BMI, skin moisture and local blood flow; The subcutaneous pressure prediction module is used to input the multimodal feature vector into a pre-trained subcutaneous pressure prediction model to predict the user's subcutaneous pressure experience value; wherein, the subcutaneous pressure prediction model uses the SHAP method to interpret the model to determine the contribution of multimodal features to subcutaneous pressure; An adaptive pressure adjustment module is used to compare the subcutaneous pressure experience value with a preset safe pressure threshold. If the subcutaneous pressure experience value exceeds the preset safe pressure threshold, the inflation volume of the airbag inside the smart brace body is adjusted to change the pressure distribution of the preset key area, thereby reducing the subcutaneous pressure experience value.

8. A computer device, comprising: The processor, memory, communication interface, and communication bus are provided, wherein the processor, memory, and communication interface communicate with each other via the communication bus. The memory is used to store at least one executable instruction, which causes the processor to perform the operation corresponding to the intelligent brace adaptive pressure control system based on the subcutaneous pressure prediction model as described in any one of claims 1-6.

Images