Biofeedback equipment database detection operation method based on augmented reality

Abstract

The invention discloses a biofeedback equipment database detection operation method based on augmented reality, and relates to the technical field of biofeedback, real-time adjustment is carried out through gait anomaly detection and AR visual feedback, a patient can see gait data in the training process, gait errors can be visually perceived through color-coded anomaly marks, and the training accuracy is improved. Wrong gait solidification is avoided; in addition, personalized gait analysis is carried out by combining Transform and GNN with transfer learning, compared with traditional anomaly judgment of a fixed threshold, the detection standard can be dynamically adjusted, self-adaptive optimization is carried out in combination with individual features of a patient, misjudgment is reduced, gait guidance is provided by using augmented reality (AR) equipment, and the detection accuracy is improved. Comprising the steps of virtual gait reference marking, gait alignment identification, dynamic footprint projection and target point setting, so that a patient obtains gait adjustment guidance visually, and in combination with vibration feedback and voice prompt of the intelligent insole, an instant adjustment signal is provided when an abnormal gait occurs.

Description

Technical Field

[0001] This invention relates to the field of biofeedback technology, and in particular to a database detection and calculation method for biofeedback devices based on augmented reality. Background Technology

[0002] Biofeedback technology is widely used in gait rehabilitation training for patients with spinal cord injury (SCI). By collecting electromyographic signals, gait data, and plantar pressure information, and combining them with database detection and calculation methods for analysis, biofeedback devices can help patients optimize gait patterns and improve neural control.

[0003] Most current biofeedback systems employ a post-training analysis model, generating data reports and adjustment suggestions after a patient completes a round of gait training. This results in a feedback lag. Patients with gait impairment due to SCI already struggle to perceive their own gait patterns. If they cannot receive real-time guidance and immediate adjustments during training, incorrect gait patterns may be reinforced, hindering recovery progress. To optimize feedback, some systems use fixed thresholds to detect abnormal gait, such as judging an abnormality as a stride length less than a set value or indicating instability as insufficient support time. However, gait patterns vary significantly among patients, and a uniform standard often leads to misjudgments or omissions. Some patients, despite deviating from healthy gait standards, exhibit good gait stability and are still judged as abnormal by the system. This mechanical judgment method may cause unnecessary anxiety for patients during rehabilitation and even affect their confidence in training.

[0004] It can be seen that current biofeedback gait detection methods are hampered by feedback lag and lack of personalization, making it difficult to provide real-time personalized guidance during gait training. Therefore, there is an urgent need for an augmented reality-based biofeedback device database detection and computation scheme to improve the real-time performance, adaptability, and interactivity of biofeedback devices in gait training. Summary of the Invention

[0005] In view of the aforementioned existing problems, the present invention is proposed.

[0006] This invention provides a database detection and calculation method for biofeedback devices based on augmented reality, which solves the problems of feedback lag, fixed thresholds, unintuitive interaction methods, high computational resource requirements, and impact on the real-time performance and personalized adaptability of gait training in traditional biofeedback systems.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0008] This invention provides a database detection and calculation method for biofeedback devices based on augmented reality, comprising:

[0009] Step S1: Collect patient gait training data to obtain gait parameters;

[0010] Step S2: Perform gait detection analysis based on gait parameters, predict gait abnormalities, and adjust gait detection standards;

[0011] Step S3: Based on the gait detection and analysis results of step S2, predict the gait training guidance target, adjust the gait training guidance method, and provide gait adjustment feedback in real time;

[0012] Step S4: Based on the gait adjustment feedback from step S3, further predict the gait optimization trend and dynamically adjust the gait training intervention strategy.

[0013] As a preferred embodiment of the augmented reality-based biofeedback device database detection and calculation method described in this invention, the gait training data includes electromyographic signals, inertial measurement unit (IMU) data, and plantar pressure data.

[0014] The gait training data is filtered and feature extracted to obtain gait parameters, including stride length, stride speed, support phase time, and swing phase time.

[0015] The gait parameters are visualized in an augmented reality (AR) device, allowing patients to view gait information in real time during training.

[0016] As a preferred embodiment of the augmented reality-based biofeedback device database detection and calculation method described in this invention, the gait detection and analysis adopts the Transformer and Graph Neural Network (GNN) models, and dynamically adjusts the detection threshold of abnormal gait by combining transfer learning.

[0017] The abnormal gait analysis results are visualized in the augmented reality (AR) device, and abnormal areas are highlighted; for example, when the stride is insufficient, the problem area is marked with color coding.

[0018] As a preferred embodiment of the augmented reality-based biofeedback device database detection and calculation method described in this invention, wherein: in the step of using Transformer and Graph Neural Network (GNN) models, combined with transfer learning to dynamically adjust the detection threshold of abnormal gait:

[0019] Gait training data from patients was collected, including electromyography (EMG) signals, inertial measurement unit (IMU) data, and plantar pressure data. After signal filtering and feature extraction, a personalized gait baseline was established.

[0020] The gait baseline consists of stride length, stride speed, stance phase time, and swing phase time. A Transformer encoder is used to extract long-term gait features. The extraction process is described as follows:

[0021] ,

[0022] in, Indicates the current gait state characteristics, Indicates the current input gait parameters. This indicates the gait characteristics at the previous moment;

[0023] A graph neural network (GNN) is used to construct a gait correlation graph, mapping gait temporal information into a graph structure and defining gait nodes. and its adjacency relationship The gait abnormality score is calculated using the following formula:

[0024] ,

[0025] in, Indicates the first Abnormal scoring of steps Represents the set of neighboring gait nodes. This represents the adjacency weights in the gait association graph. This represents the weight matrix of a graph neural network. Indicates the first State characteristics of each gait node Indicates the activation function;

[0026] Gait abnormalities are further refined as follows:

[0027] Asymmetric anomaly: Calculate the ratio of left to right step size. If: If the gait is asymmetrical, then it is determined to be gait asymmetry. It indicates that the left foot is longer. It indicates that the right foot is longer. This indicates the threshold for determining step size asymmetry.

[0028] Gait rhythm abnormalities: Analysis of gait cycle standard deviation ,like If so, it is determined to be an abnormal gait rhythm, among which, The standard deviation of gait period, This indicates the threshold for determining gait rhythm abnormalities.

[0029] In a preferred embodiment of the augmented reality-based biofeedback device database detection and calculation method described in this invention, the method for dynamically adjusting the detection threshold of abnormal gait is as follows:

[0030] A transfer learning strategy is employed to adapt the detection model to individual differences, and a gait anomaly detection threshold is defined. Based on reinforcement learning and dynamic adjustment, the adjustment formula is as follows:

[0031] ,

[0032] in, This indicates the updated gait anomaly detection threshold. This indicates the current gait anomaly detection threshold. Indicates the learning rate. This indicates a gait abnormality feedback signal.

[0033] Visualized gait anomalies in augmented reality (AR) devices include:

[0034] Gait trajectories are displayed in three-dimensional space, with abnormal gait trajectories highlighted in darker color.

[0035] Based on gait time-series data, abnormal gait patterns of patients are replayed and comparative analysis is provided.

[0036] As a preferred embodiment of the augmented reality-based biofeedback device database detection and calculation method described in this invention, the gait training guidance uses an augmented reality (AR) device to generate virtual gait reference markers, including stride standard lines and gait alignment markers, to guide the patient to adjust their gait pattern;

[0037] The gait feedback uses a multimodal interaction method, including voice prompts and smart insole vibration feedback, to provide immediate adjustment guidance when the patient's gait is abnormal.

[0038] As a preferred embodiment of the augmented reality-based biofeedback device database detection and calculation method of the present invention, wherein: in step S3, the step of generating virtual gait reference markers using an augmented reality (AR) device is as follows:

[0039] Generate gait guidance markers in augmented reality (AR) devices, including:

[0040] Stride length standard: Calculate stride length based on the patient's gait baseline. Project stride lines onto the ground to guide patients to walk with a standard stride.

[0041] Gait alignment marker: Displays the gait alignment path in the AR interface, providing a visual reference for gait symmetry;

[0042] The formula for calculating the stride standard line is:

[0043] ,

[0044] in, Indicates the length of the stride standard line. Indicates the gait scaling factor. This represents the average normal stride length within the gait baseline;

[0045] The gait guidance mode is extended by setting a virtual gait target point in front of the patient to encourage them to step towards the target area. The formula for setting the target point position is as follows:

[0046] ,

[0047] in, Indicates the current target point location. Indicates the position of the previous target point. Indicates walking speed. Indicates the time step.

[0048] Real-time dynamic footprints are projected onto the ground to indicate the patient's foot placement, enhancing gait correction. The formula for setting the dynamic footprint position is as follows:

[0049] ,

[0050] in, Indicates the first Footprint location Indicates the position of the previous footprint. This indicates the step size adjustment amount.

[0051] In a preferred embodiment of the augmented reality-based biofeedback device database detection and calculation method of the present invention, wherein: in step S3, the method of guiding the patient to adjust the gait pattern includes:

[0052] Initially, high-frequency gait reference markers are provided, and auxiliary information is gradually reduced as training progresses to promote autonomous gait control;

[0053] Enhanced guidance feedback when gait abnormalities are detected:

[0054] Play real-time audio guidance to help patients adjust their gait.

[0055] The smart insole provides vibration alerts.

[0056] As a preferred embodiment of the augmented reality-based biofeedback device database detection and calculation method described in this invention, the process of dynamically adjusting the gait training intervention strategy adopts an artificial intelligence model to optimize the training plan based on the patient's real-time gait data and historical training data.

[0057] When a patient’s gait improves, reduce the frequency or intensity of gait training and guidance interventions to promote the patient’s autonomous gait adjustment.

[0058] When gait abnormalities persist, enhance gait training guidance feedback, such as increasing the visual intensity of virtual gait reference markers or increasing the frequency of vibration feedback.

[0059] In dynamic gait adjustment training, gait training data is stored and training reports are generated, including gait trends and adjustment trajectories;

[0060] The training report uses augmented reality (AR) devices to visualize and replay the training process, and employs 3D animation to present the gait adjustment process, enabling patients to intuitively understand gait changes.

[0061] The report combines AI to predict gait recovery progress and dynamically optimizes subsequent training programs based on the recovery status.

[0062] As a preferred embodiment of the augmented reality-based biofeedback device database detection and calculation method described in this invention, in step S4, the step of optimizing the training scheme based on the patient's real-time gait data and historical training data is as follows:

[0063] Based on historical gait data and current gait performance, a gait improvement index is defined to predict the patient's gait development trend.

[0064] ,

[0065] in, This indicates the current gait improvement index. This represents the gait improvement index at the previous moment. Indicates gait improvement score, Indicates the gait smoothing coefficient;

[0066] Adaptive adjustment of training difficulty, including:

[0067] Gait stabilization phase: Reduce gait guidance interventions and gradually decrease the brightness and number of virtual reference markers.

[0068] During the persistent gait abnormality phase: increase the intensity of gait feedback;

[0069] The formula for adjusting training difficulty is:

[0070] ,

[0071] in, This indicates the updated training difficulty level. Indicates the current training difficulty. Adjusting the coefficient to increase training difficulty. The gait improvement index, To improve the target threshold for gait,

[0072] A reinforcement learning framework is used to adjust the intensity of training intervention based on the patient's real-time gait data, and the training reward function is defined as follows. :

[0073] ,

[0074] in, This represents the training reward value. This is the step size error penalty coefficient. Indicates the current stride length. For the target stride, This is the penalty coefficient for gait rhythm error. Indicates gait period, Indicates the target gait period;

[0075] When the trend of gait optimization is developing positively, reduce the frequency of gait training;

[0076] When gait abnormalities persist, increase the visual intensity of the virtual gait reference markers and increase the vibration feedback frequency.

[0077] The beneficial effects of this invention are as follows: This invention makes real-time adjustments through gait anomaly detection and AR visualization feedback. Patients can see gait data during training and intuitively perceive gait errors through color-coded anomaly markers, avoiding the solidification of incorrect gait. In addition, it uses Transformer and GNN combined with transfer learning for personalized gait analysis. Compared with the traditional fixed threshold anomaly judgment, it can dynamically adjust the detection criteria and adaptively optimize based on individual patient characteristics, thereby improving the accuracy of gait analysis and reducing misjudgments or omissions.

[0078] This invention utilizes augmented reality (AR) devices to provide gait guidance, including virtual gait reference markers, gait alignment markers, dynamic footprint projection, and target point settings. This allows patients to receive visual guidance on gait adjustment. Combined with vibration feedback from smart insoles and voice prompts, it provides immediate adjustment signals when abnormal gait occurs, thereby improving patients' awareness of gait adjustment and their participation in training.

[0079] This invention addresses the training optimization problem by dynamically predicting the patient's gait recovery trend based on the gait improvement index and reinforcement learning-based feedback optimization strategy. The training difficulty is adjusted accordingly, reducing gait guidance intervention during the gait improvement phase to promote patient self-adjustment, and increasing visual marker intensity or vibration feedback frequency during the abnormality persistence phase to enhance training effectiveness. Furthermore, the gait adjustment process can be visualized through 3D animation playback and data trend analysis, allowing patients to intuitively understand their training progress, increasing their confidence and improving rehabilitation compliance.

[0080] In summary, this invention makes gait training more intelligent, efficient, and personalized through real-time gait detection, personalized training optimization, and augmented reality interaction modes. It not only improves patients' gait correction ability but also optimizes the training feedback mechanism, accelerates the rehabilitation process, and enhances the effectiveness and long-term adaptability of gait training. Attached Figure Description

[0081] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0082] Figure 1 This is a flowchart illustrating the database detection and calculation method for the biofeedback device of the present invention. Detailed Implementation

[0083] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0084] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0085] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0086] Example 1, referring to Figure 1 This embodiment provides a database detection and computation method for augmented reality-based biofeedback devices, including:

[0087] Step S1: Collect patient gait training data to obtain gait parameters;

[0088] Gait training data includes electromyographic signals, inertial measurement unit (IMU) data, and plantar pressure data;

[0089] Gait training data is filtered and feature extracted to obtain gait parameters, including stride length, stride speed, support phase time and swing phase time;

[0090] Gait parameters are visualized in augmented reality (AR) devices, allowing patients to view gait information in real time during training.

[0091] Step S2: Perform gait detection analysis based on gait parameters, predict gait abnormalities, and adjust gait detection standards;

[0092] Gait detection and analysis employs Transformer and Graph Neural Network (GNN) models, combined with transfer learning to dynamically adjust the detection threshold for abnormal gait.

[0093] Abnormal gait analysis results are visualized in augmented reality (AR) devices, and abnormal areas are highlighted; for example, when stride length is insufficient, the problem area is identified by color coding.

[0094] In the step of dynamically adjusting the detection threshold for abnormal gait using Transformer and Graph Neural Network (GNN) models combined with transfer learning:

[0095] Gait training data from patients was collected, including electromyography (EMG) signals, inertial measurement unit (IMU) data, and plantar pressure data. After signal filtering and feature extraction, a personalized gait baseline was established.

[0096] The gait baseline consists of stride length, stride speed, stance phase time, and swing phase time. A Transformer encoder is used to extract long-term gait features. The extraction process is described as follows:

[0097] ,

[0098] in, Indicates the current gait state characteristics, Indicates the current input gait parameters. This indicates the gait characteristics at the previous moment;

[0099] A graph neural network (GNN) is used to construct a gait correlation graph, mapping gait temporal information into a graph structure and defining gait nodes. and its adjacency relationship The gait abnormality score is calculated using the following formula:

[0100] ,

[0101] in, Indicates the first Abnormal scoring of steps Represents the set of neighboring gait nodes. This represents the adjacency weights in the gait association graph. This represents the weight matrix of a graph neural network. Indicates the first State characteristics of each gait node Indicates the activation function;

[0102] Gait abnormalities are further refined as follows:

[0103] Asymmetric anomaly: Calculate the ratio of left to right step size. If: If the gait is asymmetrical, then it is determined to be gait asymmetry. It indicates that the left foot is longer. It indicates that the right foot is longer. This indicates the threshold for determining step size asymmetry.

[0104] Gait rhythm abnormalities: Analysis of gait cycle standard deviation ,like If so, it is determined to be an abnormal gait rhythm, among which, The standard deviation of gait period, This indicates the threshold for determining gait rhythm abnormalities;

[0105] The method for dynamically adjusting the detection threshold for abnormal gait is as follows:

[0106] A transfer learning strategy is employed to adapt the detection model to individual differences, and a gait anomaly detection threshold is defined. Based on reinforcement learning and dynamic adjustment, the adjustment formula is as follows:

[0107] ,

[0108] in, This indicates the updated gait anomaly detection threshold. This indicates the current gait anomaly detection threshold. Indicates the learning rate. This indicates a gait abnormality feedback signal.

[0109] Visualized gait anomalies in augmented reality (AR) devices include:

[0110] Gait trajectories are displayed in three-dimensional space, with abnormal gait trajectories highlighted in darker color.

[0111] Based on gait time-series data, abnormal gait patterns of patients are replayed and comparative analysis is provided.

[0112] Specifically, this method utilizes Transformer to extract long-term gait features, establishes a personalized gait baseline, and constructs a gait correlation graph to detect gait abnormality types, including gait asymmetry and gait rhythm abnormalities. Subsequently, based on transfer learning, the abnormal gait detection threshold is dynamically adjusted to adapt the detection criteria to individual differences and improve the accuracy of gait analysis. Finally, in an augmented reality (AR) device, abnormal gait is visualized through gait trajectory reconstruction and animation playback, helping patients intuitively understand their own gait problems and enhancing their autonomy in gait correction.

[0113] Step S3: Based on the gait detection and analysis results of step S2, predict the gait training guidance target, adjust the gait training guidance method, and provide gait adjustment feedback in real time;

[0114] Gait training guidance uses augmented reality (AR) devices to generate virtual gait reference markers, including stride standard lines and gait alignment markers, to guide patients in adjusting their gait patterns;

[0115] Gait feedback employs a multimodal interaction method, including voice prompts and smart insole vibration feedback, providing immediate adjustment guidance when the patient's gait is abnormal;

[0116] In step S3, the step of generating virtual gait reference markers using an augmented reality (AR) device is as follows:

[0117] Generate gait guidance markers in augmented reality (AR) devices, including:

[0118] Stride length standard: Calculate stride length based on the patient's gait baseline. Project stride lines onto the ground to guide patients to walk with a standard stride.

[0119] Gait alignment marker: Displays the gait alignment path in the AR interface, providing a visual reference for gait symmetry;

[0120] The formula for calculating the stride standard line is:

[0121] ,

[0122] in, Indicates the length of the stride standard line. Indicates the gait scaling factor. This represents the average normal stride length within the gait baseline;

[0123] The gait guidance mode is extended by setting a virtual gait target point in front of the patient to encourage them to step towards the target area. The formula for setting the target point position is as follows:

[0124] ,

[0125] in, Indicates the current target point location. Indicates the position of the previous target point. Indicates walking speed. Indicates the time step.

[0126] Real-time dynamic footprints are projected onto the ground to indicate the patient's foot placement, enhancing gait correction. The formula for setting the dynamic footprint position is as follows:

[0127] ,

[0128] in, Indicates the first Footprint location Indicates the position of the previous footprint. Indicates the step size adjustment amount;

[0129] In step S3, the methods for guiding the patient to adjust their gait pattern include:

[0130] Initially, high-frequency gait reference markers are provided, and auxiliary information is gradually reduced as training progresses to promote autonomous gait control;

[0131] Enhanced guidance feedback when gait abnormalities are detected:

[0132] Play real-time audio guidance to help patients adjust their gait.

[0133] Smart insoles provide vibration alerts;

[0134] Specifically, this method provides gait training guidance based on AR devices and uses various methods such as virtual gait reference markers and gait alignment markers to help patients optimize their gait patterns. Through stride standard lines, dynamic footprints, and virtual target points, it guides patients to train according to standard gait parameters. The gait guidance mode is intelligently adjusted for different training stages. In the early stages, high-frequency visual markers are provided, and auxiliary information is reduced as training progresses to promote autonomous gait adjustment. In addition, the training feedback adopts multimodal interaction, including voice prompts and intelligent insole vibration feedback, to ensure that patients can perceive and adjust in real time when gait abnormalities occur, effectively improving the interactivity and personalization of gait training and enhancing the adaptability and corrective effect of gait training.

[0135] Step S4: Based on the gait adjustment feedback in step S3, further predict the gait optimization trend and dynamically adjust the gait training intervention strategy.

[0136] The process of dynamically adjusting gait training intervention strategies uses an artificial intelligence model to optimize training programs based on patients’ real-time gait data and historical training data.

[0137] When a patient’s gait improves, reduce the frequency or intensity of gait training and guidance interventions to promote the patient’s autonomous gait adjustment.

[0138] When gait abnormalities persist, enhance gait training guidance feedback, such as increasing the visual intensity of virtual gait reference markers or increasing the frequency of vibration feedback.

[0139] In dynamic gait adjustment training, gait training data is stored and training reports are generated, including gait trends and adjustment trajectories;

[0140] The training report uses augmented reality (AR) devices to visualize and replay the training process, and employs 3D animation to present the gait adjustment process, enabling patients to intuitively understand gait changes.

[0141] The report combines AI to predict gait recovery progress and dynamically optimizes subsequent training programs based on the recovery status;

[0142] In step S4, the step of optimizing the training plan based on the patient's real-time gait data and historical training data is as follows:

[0143] Based on historical gait data and current gait performance, a gait improvement index is defined to predict the patient's gait development trend.

[0144] ,

[0145] in, This indicates the current gait improvement index. This represents the gait improvement index at the previous moment. Indicates gait improvement score, Indicates the gait smoothing coefficient;

[0146] Adaptive adjustment of training difficulty, including:

[0147] Gait stabilization phase: Reduce gait guidance interventions and gradually decrease the brightness and number of virtual reference markers.

[0148] During the persistent gait abnormality phase: increase the intensity of gait feedback;

[0149] The formula for adjusting training difficulty is:

[0150] ,

[0151] in, This indicates the updated training difficulty level. Indicates the current training difficulty. Adjusting the coefficient to increase training difficulty. The gait improvement index, To improve the target threshold for gait,

[0152] A reinforcement learning framework is used to adjust the intensity of training intervention based on the patient's real-time gait data, and the training reward function is defined as follows. :

[0153] ,

[0154] in, This represents the training reward value. This is the step size error penalty coefficient. Indicates the current stride length. For the target stride, This is the penalty coefficient for gait rhythm error. Indicates gait period, Indicates the target gait period;

[0155] When the trend of gait optimization is developing positively, reduce the frequency of gait training;

[0156] When gait abnormalities persist, increase the visual intensity of the virtual gait reference markers and increase the vibration feedback frequency;

[0157] Specifically, this approach employs an artificial intelligence model to optimize the training program based on the patient's real-time gait data and historical training data. Firstly, it predicts the patient's gait development trend based on a gait improvement index and dynamically adjusts the training difficulty accordingly. Secondly, it utilizes a reinforcement learning framework to optimize the training feedback strategy based on the patient's real-time gait data, adjusting the frequency of gait guidance and the intensity of intervention. More specifically, when the gait improvement trend is positive, training intervention is gradually reduced to promote voluntary gait control. When gait abnormalities persist, gait feedback is enhanced by increasing the frequency of vibration cues or adding visual guidance, thereby improving the real-time nature and personalized adaptability of training feedback, accelerating the gait rehabilitation process, and enhancing the effectiveness of the patient's gait training.

[0158] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A database detection and calculation method for biofeedback devices based on augmented reality, characterized in that: include, Step S1: Collect patient gait training data to obtain gait parameters; Step S2: Perform gait detection analysis based on gait parameters, predict gait abnormalities, and adjust gait detection standards; Step S3: Based on the gait detection and analysis results of step S2, predict the gait training guidance target, adjust the gait training guidance method, and provide gait adjustment feedback in real time; Step S4: Based on the gait adjustment feedback from step S3, further predict the gait optimization trend and dynamically adjust the gait training intervention strategy.

2. The method for database detection and computation of augmented reality-based biofeedback devices as described in claim 1, characterized in that: The gait training data includes electromyographic signals, inertial measurement unit (IMU) data, and plantar pressure data; The gait training data is filtered and feature extracted to obtain gait parameters, including stride length, stride speed, support phase time, and swing phase time. The gait parameters are visualized in an augmented reality (AR) device, allowing patients to view gait information in real time during training.

3. The method for detecting and calculating a database of biofeedback devices based on augmented reality as described in claim 2, characterized in that: The gait detection and analysis uses Transformer and Graph Neural Network (GNN) models, combined with transfer learning to dynamically adjust the detection threshold for abnormal gait. The abnormal gait analysis results are visualized in the augmented reality (AR) device, and the abnormal areas are highlighted.

4. The method for detecting and calculating a database of biofeedback devices based on augmented reality as described in claim 3, characterized in that: In the step of using Transformer and Graph Neural Network (GNN) models, combined with transfer learning to dynamically adjust the detection threshold for abnormal gait: Gait training data from patients was collected, including electromyography (EMG) signals, inertial measurement unit (IMU) data, and plantar pressure data. After signal filtering and feature extraction, a personalized gait baseline was established. The gait baseline consists of stride length, stride speed, stance phase time, and swing phase time. A Transformer encoder is used to extract long-term gait features. The extraction process is described as follows: ， in, Indicates the current gait state characteristics, Indicates the current input gait parameters. This indicates the gait characteristics at the previous moment; A graph neural network (GNN) is used to construct a gait correlation graph, mapping gait temporal information into a graph structure and defining gait nodes. and its adjacency relationship The gait abnormality score is calculated using the following formula: ， in, Indicates the first Abnormal scoring of steps Represents the set of neighboring gait nodes. This represents the adjacency weights in the gait association graph. This represents the weight matrix of a graph neural network. Indicates the first State characteristics of each gait node Indicates the activation function; Gait abnormalities are further refined as follows: Asymmetric anomaly: Calculate the ratio of left to right step size. If: If the gait is asymmetrical, then it is determined to be gait asymmetry. It indicates that the left foot is longer. It indicates that the right foot is longer. This indicates the threshold for determining step size asymmetry. Gait rhythm abnormalities: Analysis of gait cycle standard deviation ,like If so, it is determined to be an abnormal gait rhythm, among which, The standard deviation of gait period, This indicates the threshold for determining gait rhythm abnormalities.

5. The method for detecting and calculating a database of biofeedback devices based on augmented reality as described in claim 4, characterized in that: The method for dynamically adjusting the detection threshold of abnormal gait is as follows: A transfer learning strategy is employed to adapt the detection model to individual differences, and a gait anomaly detection threshold is defined. Based on reinforcement learning and dynamic adjustment, the adjustment formula is as follows: ， in, This indicates the updated gait anomaly detection threshold. This indicates the current gait anomaly detection threshold. Indicates the learning rate. This indicates a gait abnormality feedback signal. Visualized gait anomalies in augmented reality (AR) devices include: Gait trajectories are displayed in three-dimensional space, with abnormal gait trajectories highlighted in darker color. Based on gait time-series data, abnormal gait patterns of patients are replayed and comparative analysis is provided.

6. The method for detecting and calculating a database of biofeedback devices based on augmented reality as described in claim 5, characterized in that: The gait training guidance uses augmented reality (AR) devices to generate virtual gait reference markers, including stride standard lines and gait alignment markers, to guide patients in adjusting their gait patterns; The gait feedback uses a multimodal interaction method, including voice prompts and smart insole vibration feedback, to provide immediate adjustment guidance when the patient's gait is abnormal.

7. The method for detecting and calculating a database of biofeedback devices based on augmented reality as described in claim 6, characterized in that: In step S3, the step of generating virtual gait reference markers using an augmented reality (AR) device is as follows: Generate gait guidance markers in augmented reality (AR) devices, including: Stride length standard: Calculate stride length based on the patient's gait baseline. Project stride lines onto the ground to guide patients to walk with a standard stride. Gait alignment marker: Displays the gait alignment path in the AR interface, providing a visual reference for gait symmetry; The formula for calculating the stride standard line is: ， in, Indicates the length of the stride standard line. Indicates the gait scaling factor. This represents the average normal stride length within the gait baseline; The gait guidance mode is extended by setting a virtual gait target point in front of the patient to encourage them to step towards the target area. The formula for setting the target point position is as follows: ， in, Indicates the current target point location. Indicates the position of the previous target point. Indicates walking speed. Indicates the time step. Real-time dynamic footprints are projected onto the ground to indicate the patient's foot placement, enhancing gait correction. The formula for setting the dynamic footprint position is as follows: ， in, Indicates the first Footprint location Indicates the position of the previous footprint. This indicates the step size adjustment amount.

8. The method for detecting and calculating a database of biofeedback devices based on augmented reality as described in claim 7, characterized in that: In step S3, the method of guiding the patient to adjust their gait pattern includes: Initially, high-frequency gait reference markers are provided, and auxiliary information is gradually reduced as training progresses to promote autonomous gait control; Enhanced guidance feedback when gait abnormalities are detected: Play real-time audio guidance to help patients adjust their gait. The smart insole provides vibration feedback.

9. The method for detecting and calculating a database of biofeedback devices based on augmented reality as described in claim 8, characterized in that: The process of dynamically adjusting the gait training intervention strategy uses an artificial intelligence model to optimize the training plan based on the patient's real-time gait data and historical training data. When a patient’s gait improves, reduce the frequency or intensity of gait training and guidance interventions. When gait abnormalities persist, enhance gait training guidance and feedback.

10. The method for detecting and calculating a database of biofeedback devices based on augmented reality as described in claim 9, characterized in that: In step S4, the step of optimizing the training program based on the patient's real-time gait data and historical training data is as follows: Based on historical gait data and current gait performance, a gait improvement index is defined to predict the patient's gait development trend. ， in, This indicates the current gait improvement index. This represents the gait improvement index at the previous moment. Indicates gait improvement score, Indicates the gait smoothing coefficient; Adaptive adjustment of training difficulty, including: Gait stabilization phase: Reduce gait guidance interventions and gradually decrease the brightness and number of virtual reference markers. During the persistent gait abnormality phase: increase the intensity of gait feedback; The formula for adjusting training difficulty is: ， in, This indicates the updated training difficulty level. Indicates the current training difficulty. Adjust the coefficient to increase training difficulty. The gait improvement index, To improve the target threshold for gait, A reinforcement learning framework is used to adjust the intensity of training intervention based on the patient's real-time gait data, and the training reward function is defined as follows. : ， in, This represents the training reward value. This is the step size error penalty coefficient. Indicates the current stride length. For the target stride, This is the penalty coefficient for gait rhythm error. Indicates gait period, Indicates the target gait period; When the trend of gait optimization is developing positively, reduce the frequency of gait training; When gait abnormalities persist, increase the visual intensity of the virtual gait reference markers and increase the vibration feedback frequency.

Images