Health monitoring method and device applied to intelligent running shoes and electronic equipment

Abstract

The invention provides a health monitoring method and device applied to intelligent running shoes and electronic equipment, and relates to the field of data processing. In the method, multi-modal acquisition data for a wearer of the intelligent running shoes is acquired, and the multi-modal acquisition data comprises biochemical data, acoustic data, vibration data and electrode data; performing feature processing on the biochemical data and the acoustic data by adopting an LSTM + CNN hybrid model to generate a chemical + acoustic fusion fatigue index; performing feature processing on the vibration data by adopting an LSTM + CNN hybrid model to obtain a skeleton stress early warning model; performing feature processing on the biochemical data and the electrode data by adopting an LSTM + CNN hybrid model to generate a biochemical + myoelectricity interaction health situation; and performing health monitoring on the wearer based on the chemical + acoustics fusion fatigue index, the skeleton stress early warning model and the biochemical + myoelectricity interaction health situation. By implementing the technical scheme provided by the invention, health monitoring on the wearer is facilitated.

Description

Technical Field

[0001] This application relates to the technical field of data processing, and particularly relates to a health monitoring method, device, and electronic device for intelligent running shoes. Background Art

[0002] With the continuous improvement of people's demands for health management and sports performance optimization, sports monitoring technology is undergoing a profound transformation from single physical quantity collection to multi-dimensional physiological index integration. From the earliest pedometers and simple heart rate belts to the current pressure sensors and IMUs built into smart watches and running shoes, the industry has accumulated a large amount of data on exercise intensity, gait patterns, and heart rate fluctuations.

[0003] Currently, traditional running shoes or wearable devices mostly focus on physical mechanics signals (such as pressure and acceleration) and single physiological indexes (such as heart rate), ignoring biochemical and electrophysiological information such as human metabolic level, endocrine stress, and muscle electrical activity, and it is difficult to accurately evaluate fatigue accumulation, stress response, and potential injury risks. Therefore, it is not conducive to health monitoring of the wearers of running shoes.

[0004] Therefore, there is an urgent need for a health monitoring method, device, and electronic device for intelligent running shoes. Summary of the Invention

[0005] This application provides a health monitoring method, device, and electronic device for intelligent running shoes, which is convenient for health monitoring of the wearers of running shoes.

[0006] In the first aspect of this application, a health monitoring method for intelligent running shoes is provided. The method includes: obtaining multi-modal acquisition data for the wearer of the intelligent running shoes, where the multi-modal acquisition data includes biochemical data, acoustic data, vibration data, and electrode data; using an LSTM+CNN hybrid model to perform feature processing on the biochemical data and the acoustic data to generate a chemical+acoustic fusion fatigue index; using the LSTM+CNN hybrid model to perform feature processing on the vibration data to obtain a bone stress warning model; using the LSTM+CNN hybrid model to perform feature processing on the biochemical data and the electrode data to generate a biochemical+electromyogram interaction health situation; and performing health monitoring on the wearer based on the chemical+acoustic fusion fatigue index, the bone stress warning model, and the biochemical+electromyogram interaction health situation.

[0007] By adopting the above technical solution, by introducing multi-modal data acquisition and deep fusion modeling in the intelligent running shoes, the limitation of traditional methods that only rely on physical signals is broken through, and the physical state of the wearer can be comprehensively sensed from the chemical, physiological, electrical signal and mechanical dimensions. The LSTM+CNN hybrid model is used to extract collaborative features from biochemical, acoustic, vibration and electrode signals, and three core indicators, namely fatigue index, bone stress warning and myoelectric interaction health status, are respectively constructed, realizing real-time and accurate assessment of the fatigue accumulation, bone load and muscle activity status of the exerciser, thus providing strong data support for personalized training suggestions, injury warning and health intervention, and having higher health monitoring accuracy and warning response ability. Therefore, it is convenient to monitor the health of the wearer of the running shoes.

[0008] Optionally, the obtaining of the multi-modal acquisition data for the wearer of the intelligent running shoes, where the multi-modal acquisition data includes biochemical data, acoustic data, vibration data and electrode data, specifically includes: obtaining lactic acid, glucose, electrolytes and cortisol sent by the microfluidic channel and electrochemical electrode pre-installed in the intelligent running shoes; performing differential amplification processing on the lactic acid, glucose, electrolytes and cortisol to obtain the biochemical data; obtaining the ultrasonic echo signal of the sole hitting the ground sent by the MEMS microphone array pre-installed in the intelligent running shoes; performing band-pass filtering processing on the ultrasonic echo signal to obtain the acoustic data; obtaining the foot bone conduction vibration signal sent by the bone conduction accelerometer pre-installed in the intelligent running shoes; performing low-noise differential processing on the foot bone conduction vibration signal to obtain the vibration data; obtaining the foot electromyogram signal sent by the dry sEMG electrode array pre-installed in the intelligent running shoes; performing differential amplification processing on the foot electromyogram signal to obtain the electrode data.

[0009] By adopting the above technical solution, by integrating microfluidic sensing, electrochemical sensing, MEMS microphone, bone conduction accelerometer and dry sEMG electrode array in the intelligent running shoes, the synchronous acquisition and preprocessing of metabolites such as lactic acid, glucose, electrolytes and cortisol of the wearer, sole ultrasonic echo, bone conduction vibration and electromyogram signal are realized. Compared with the traditional solution, its advantage is that it significantly improves the dimension and accuracy of health monitoring, can not only reflect the metabolic state and endocrine stress level during exercise in real time, but also deeply sense the muscle activation mode and bone conduction stress, so as to realize the comprehensive dynamic assessment of fatigue accumulation, injury risk and body load status, and provide highly reliable data support for personalized training optimization and injury prevention.

[0010] Optionally, the LSTM+CNN hybrid model is used to process the biochemical data and the acoustic data to generate a chemical+acoustic fusion fatigue index, which specifically includes: using the LSTM+CNN hybrid model to extract features from the biochemical data to obtain biochemical features, where the biochemical features include the average lactic acid slope and the glucose drop rate; using the LSTM+CNN hybrid model to extract features from the acoustic data to obtain acoustic features, where the acoustic features include the center of gravity of the energy spectrum per step and the peak value of the foot impact in the time domain; fusing the data acquisition times of the biochemical data and the acoustic data to obtain biochemical time-series features and acoustic time-series features; using the CNN unit in the LSTM+CNN hybrid model to perform convolution on the acoustic time-series features to extract local impact patterns; using the LSTM unit in the LSTM+CNN hybrid model to perform time-series splicing of the biochemical time-series features and the local impact patterns, and outputting the chemical+acoustic fusion fatigue index.

[0011] By adopting the above technical solution, by synchronously fusing biochemical data and acoustic data in the time dimension and using an LSTM+CNN hybrid model for deep feature extraction and dynamic modeling, more accurate quantification of the fatigue state of the exerciser is achieved. The system not only extracts the change rates of metabolites such as lactic acid and glucose, but also identifies the acoustic energy characteristics and time-domain peak values during the plantar impact. Then, the CNN is used to extract local impact patterns, and the LSTM captures the time-series evolution of metabolic and impact signals, and finally a fatigue index that fuses these two types of information is generated. The advantage of this method is that it breaks through the limitations of traditional single-signal fatigue determination, can more sensitively identify the hidden fatigue accumulation jointly driven by metabolic abnormalities and gait impacts, and improves the sensitivity, accuracy, and individual judgment ability of fatigue detection.

[0012] Optionally, the LSTM+CNN hybrid model is used to process the vibration data to obtain a bone stress warning model, which specifically includes: using the LSTM+CNN hybrid model to extract features from the vibration data to obtain vibration features, where the vibration features include the high-frequency energy attenuation rate, the resonance peak shift, and the peak value of the time-domain envelope wave; using the CNN subnet in the LSTM+CNN hybrid model to determine the high-frequency resonance peak features in the vibration features; using the LSTM subnet in the LSTM+CNN hybrid model to determine the stress accumulation trend of the vibration features; combining the high-frequency resonance peak features and the stress accumulation trend to construct the bone stress warning model.

[0013] By adopting the above technical solution, through deep modeling of the vibration data collected by the bone conduction accelerometer in the intelligent running shoes, and combining the collaborative feature extraction capabilities of LSTM and CNN, an early warning model capable of predicting the bone stress state is effectively constructed. Specifically, the CNN subnet accurately identifies the changes in high-frequency resonance peaks in the vibration signal and captures minute structural stress responses; the LSTM subnet dynamically tracks the temporal evolution of stress accumulation and forms the ability to perceive long-term bone load trends. The advantage of this method is that it can identify abnormal vibration patterns and stress aggregation trends in advance before the user perceives obvious discomfort, providing an early warning basis for bone injuries such as tibial fatigue and plantar fasciitis, and greatly improving the scientific nature of sports safety and rehabilitation guidance.

[0014] Optionally, using the LSTM+CNN hybrid model to perform feature processing on the biochemical data and the electrode data to generate a biochemical+electromyogram interaction health situation specifically includes: determining target biochemical features according to the LSTM+CNN hybrid model and the biochemical data, where the target biochemical features include the dynamic increment of cortisol and the sodium ion concentration; determining EMG features according to the LSTM+CNN hybrid model and the electrode data, where the EMG features include the average power spectrum and the intermediate frequency power ratio; using the LSTM+CNN hybrid model to aggregate the EMG features to obtain electromyogram time-frequency map features; and interacting the electromyogram time-frequency map features and the target biochemical features through the LSTM+CNN hybrid model to obtain the biochemical+electromyogram interaction health situation.

[0015] By adopting the above technical solution, by fusing the biochemical data collected by the intelligent running shoes and the surface electromyogram (sEMG) signals, and using the LSTM+CNN hybrid model for feature extraction and time-frequency interaction modeling, a "biochemical+electromyogram interaction health situation" reflecting the current physiological stress state and the synergistic relationship of muscle functions of the exerciser is effectively constructed. Among them, the system can extract key biochemical indicators such as cortisol increment and sodium ion concentration to reflect endocrine and electrolyte changes, and at the same time analyze the power spectrum features and muscle fatigue features in the electromyogram, and then fuse and model the dynamic relationship between the two through the model to quantify whether the muscle stress is in an overloaded or non-physiological activation state. The advantage of this method is that it can not only judge in real time whether the muscle output matches the body's metabolic state, but also identify potential metabolic-neural-muscular coordination imbalances, providing intelligent and dynamic monitoring means for overtraining, unbalanced force application or chronic fatigue, and significantly improving the scientific nature and timeliness of health intervention.

[0016] Optionally, health monitoring of the wearer is performed based on the chemical + acoustic fusion fatigue index, the bone stress warning model, and the biochemical + myoelectric interaction health status, specifically including: outputting a fatigue score, a stress risk score, and a health status score according to the chemical + acoustic fusion fatigue index, the bone stress warning model, and the biochemical + myoelectric interaction health status; comparing the fatigue score, the stress risk score, and the health status score with their respective preset thresholds to obtain comparison results; and generating a health feedback strategy according to the comparison results.

[0017] By adopting the above technical solution, through the joint evaluation of three types of multi-dimensional health indicators, namely the chemical + acoustic fusion fatigue index, the bone stress warning model, and the biochemical + myoelectric interaction health status, a fatigue score, a stress risk score, and a health status score are output and compared with preset personalized thresholds, and then a targeted health feedback strategy is automatically generated. The advantage of this method is that the high-order evaluation results after multi-source data fusion are quantified into intuitive health scores, which improves the efficiency of user understanding and system response. At the same time, through comparison with thresholds, a personalized health status recognition and early intervention trigger mechanism is realized, providing users with intelligent, dynamic, and operable health management suggestions, effectively supporting scientific training and injury prevention.

[0018] Optionally, the method further includes: obtaining the real-time blood oxygen data, real-time blood pressure data, and real-time heart rate data of the wearer; and fusing and displaying the health feedback strategy with the real-time blood oxygen data, the real-time blood pressure data, and the real-time heart rate data, and the fusion display methods include LED prompts and BLE notifications.

[0019] By adopting the above technical solution, by integrating the core vital sign data of the wearer such as real-time blood oxygen, blood pressure, and heart rate, and fusing and displaying it with the previously generated health feedback strategy, not only the timeliness and physiological relevance of the health monitoring results are improved, but also multi-channel and low-latency visualization and remote push are realized through LED prompts and BLE (low-power Bluetooth) notifications. The advantage is that the intuitive linkage of multi-dimensional health information is realized, which is convenient for users to perceive their own state changes in real time, quickly respond to potential health risks, and at the same time support remote synchronization of intelligent terminals such as mobile phones or watches, enhancing the user's interaction experience and the initiative and coverage of health management.

[0020] In the second aspect of the present application, a health monitoring device for intelligent running shoes is provided. The device is a controller of the intelligent running shoes. The controller of the intelligent running shoes includes an acquisition module and a processing module. Among them, the acquisition module is used to acquire multimodal acquisition data of the wearer of the intelligent running shoes. The multimodal acquisition data includes biochemical data, acoustic data, vibration data, and electrode data. The processing module is used to perform feature processing on the biochemical data and the acoustic data by using an LSTM+CNN hybrid model to generate a chemical+acoustic fusion fatigue index. The processing module is also used to perform feature processing on the vibration data by using the LSTM+CNN hybrid model to obtain a bone stress warning model. The processing module is further used to perform feature processing on the biochemical data and the electrode data by using the LSTM+CNN hybrid model to generate a biochemical+electromyogram interaction health situation. The processing module is also used to perform health monitoring on the wearer based on the chemical+acoustic fusion fatigue index, the bone stress warning model, and the biochemical+electromyogram interaction health situation.

[0021] In the third aspect of the present application, an electronic device is provided. The electronic device includes a processor, a memory, a user interface, and a network interface. The memory is used to store instructions. The user interface and the network interface are both used to communicate with other devices. The processor is used to execute the instructions stored in the memory so that the electronic device executes the method described above.

[0022] In the fourth aspect of the present application, a computer-readable storage medium is provided. The computer-readable storage medium stores instructions. When the instructions are executed, the method described above is executed.

[0023] In summary, one or more technical solutions provided in the present application have at least the following technical effects or advantages: By introducing multimodal data acquisition and deep fusion modeling in intelligent running shoes, the limitation of traditional reliance only on physical signals is broken through, and the physical state of the wearer can be comprehensively perceived from chemical, physiological, electrical signal, and mechanical dimensions. The LSTM+CNN hybrid model is used to perform collaborative feature extraction on biochemical, acoustic, vibration, and electrode signals, and three core indicators, namely, a fatigue index, a bone stress warning, and an electromyogram interaction health situation, are respectively constructed, realizing real-time and accurate assessment of the fatigue accumulation, bone load, and muscle activity state of the exerciser. Therefore, strong data support is provided for personalized training suggestions, injury warning, and health intervention, and it has higher health monitoring accuracy and warning response ability. Therefore, it is convenient to perform health monitoring on the wearer of the running shoes. Description of the Drawings

[0024] Figure 1Schematic flowchart of a health monitoring method applied to intelligent running shoes provided by an embodiment of the present application; Figure 2 Another schematic flowchart of a health monitoring method applied to intelligent running shoes provided by an embodiment of the present application; Figure 3 Schematic module diagram of a health monitoring device applied to intelligent running shoes provided by an embodiment of the present application; Figure 4 Schematic structural diagram of an electronic device provided by an embodiment of the present application.

[0025] Description of reference numerals: 31, acquisition module; 32, processing module; 41, processor; 42, communication bus; 43, user interface; 44, network interface; 45, memory. Detailed implementation manners

[0026] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings in the embodiments of this specification. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments.

[0027] In the description of the embodiments of the present application, words such as "for example" or "for illustration" are used to represent examples, illustrations or explanations. Any embodiment or design solution described as "for example" or "for illustration" in the embodiments of the present application should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Exactly speaking, the use of words such as "for example" or "for illustration" is intended to present related concepts in a specific manner.

[0028] In the description of the embodiments of the present application, the meaning of the term "a plurality" refers to two or more. For example, a plurality of systems refers to two or more systems, and a plurality of screen terminals refers to two or more screen terminals. In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the technical features indicated. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features. The terms "include", "comprise", "have" and their variants all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0029] With the increasing emphasis on health management and sports performance optimization, sports monitoring technology is gradually evolving from the previous single physical parameter collection to an intelligent perception system that integrates multi-source physiological, biochemical, and electrophysiological information. From the initial pedometers and basic heart rate monitors to today's smartwatches and running shoes integrated with modules such as pressure sensors and inertial measurement units (IMUs), the industry has made significant progress in gait analysis, exercise intensity assessment, and heart rate variability monitoring.

[0030] However, most current traditional smart running shoes and wearable devices still mainly rely on physical mechanics signals such as pressure and acceleration and limited physiological indicators (such as heart rate) for analysis, lacking in-depth perception of key dimensions such as the body's metabolic state, endocrine stress response, and electromyographic activity. It is difficult to comprehensively and accurately judge the user's fatigue accumulation, stress level, and potential injury risks, limiting their practical utility in health monitoring.

[0031] To solve the above technical problems, this application provides a health monitoring method applied to smart running shoes, referring to Figure 1 , Figure 1 is a schematic flowchart of a health monitoring method applied to smart running shoes provided by an embodiment of this application. This method is applied to the controller of smart running shoes and includes steps S110 to S150, and the above steps are as follows: S110. Obtain multi-modal acquisition data for the wearer of the smart running shoes, where the multi-modal acquisition data includes biochemical data, acoustic data, vibration data, and electrode data.

[0032] Specifically, a central controller (usually a low-power microprocessor or system-on-chip) is embedded inside the smart running shoes, which can read data from multiple different types of sensor modules simultaneously. These sensors not only include common mechanical signal acquisition devices (such as pressure sensors and accelerometers), but also include microfluidic chemical sensors that can measure the internal biochemical changes of the human body, used to collect indicators such as lactic acid, glucose, electrolytes, and cortisol in sweat; a MEMS microphone array that can capture the sound of the sole hitting the ground and ultrasonic echoes; an accelerometer that can sense high-frequency vibration signals transmitted from the bones to the shoe body; and a dry electromyographic electrode array installed on the inner side of the shoe upper and in direct contact with the skin, used to collect the electrophysiological signals of foot muscle activities. The controller converges these data streams from different sources but complementary to each other, laying a foundation for subsequent signal preprocessing, feature extraction, and health assessment.

[0033] In a possible implementation, multi-modal acquisition data for the wearer of the intelligent running shoes is obtained. The multi-modal acquisition data includes biochemical data, acoustic data, vibration data, and electrode data. Specifically, it includes: obtaining lactic acid, glucose, electrolytes, and cortisol sent by the microfluidic channels and electrochemical electrodes pre-installed in the intelligent running shoes; performing differential amplification processing on lactic acid, glucose, electrolytes, and cortisol to obtain biochemical data; obtaining the ultrasonic echo signals of the sole hitting the ground sent by the MEMS microphone array pre-installed in the intelligent running shoes; performing band-pass filtering processing on the ultrasonic echo signals to obtain acoustic data; obtaining the foot bone conduction vibration signals sent by the bone conduction accelerometer pre-installed in the intelligent running shoes; performing low-noise differential processing on the foot bone conduction vibration signals to obtain vibration data; obtaining the foot electromyogram signals sent by the dry sEMG electrode array pre-installed in the intelligent running shoes; performing differential amplification processing on the foot electromyogram signals to obtain electrode data.

[0034] Specifically, in addition to the traditional sensors that can detect blood oxygen, heart rate, pressure, and blood pressure, the intelligent running shoes are also built-in with a variety of sensor modules. Through the central controller, four types of data highly related to the human body state are collected and preliminarily processed simultaneously: biochemical data, acoustic data, vibration data, and electrode data. Specifically, microfluidic channels and electrochemical electrodes are installed in the running shoes, which can measure biochemical indicators such as lactic acid, glucose, electrolytes, and cortisol in real time from the sweat of runners, and perform differential amplification on these signals to improve the detection sensitivity; the MEMS microphone array mounted on the sole captures the ultrasonic pulse echoes generated each time the ground is stepped on, and the sole acoustic signal is obtained after removing environmental noise through band-pass filtering; the bone conduction accelerometer at the tongue or heel of the shoe is responsible for recording the bone vibration waveform and extracting the bone conduction vibration characteristics through low-noise differential processing; at the same time, the dry surface electromyography electrode array on the inner side of the shoe upper fits the foot or calf muscles, collects the electromyogram signal and obtains clear electrode data through differential amplification. The controller synchronously converges these multi-modal data, providing rich and complementary raw materials for subsequent feature extraction and health status assessment.

[0035] For example, when a runner starts a warm-up jog, the microfluidic channels extract a small amount of sweat when the feet start to perspire slightly. The electrochemical electrodes measure an increase in lactic acid concentration to 3 mmol / L and a slight rise in cortisol level. At the same time, the MEMS microphone detects the ultrasonic echoes during running, which, after filtering, reflects the tissue elasticity changes in the arch area of the foot. The bone conduction accelerometer captures that the high-frequency vibration energy is concentrated at the calcaneus, indicating that this part bears a large impact. The dry EMG electrodes record the amplitude and frequency changes of the electrical activity of the tibialis anterior muscle, showing a gradual increase in muscle activation intensity. Through the parallel acquisition and preprocessing of these four types of signals, the intelligent running shoes can complete a preliminary analysis of physiological metabolism, gait impact, bone load, and muscle status locally, thus supporting higher-level fatigue assessment, injury warning, and sports guidance.

[0036] S120. Use an LSTM+CNN hybrid model to process the features of biochemical data and acoustic data to generate a chemical+acoustic fusion fatigue index.

[0037] Specifically, the main control chip inside the intelligent running shoes does not simply read the metabolic indicators in the sweat and the acoustic signals on the sole, but inputs this data into a hybrid model integrating a convolutional neural network (CNN) and a long short-term memory network (LSTM), and uses their respective advantages to complete deep feature extraction and temporal modeling. Specifically, the CNN subnet will automatically learn and identify local patterns such as the impact energy distribution and rebound characteristics for acoustic signals, such as the ultrasonic echo waveform generated by each sole strike. At the same time, the LSTM subnet is responsible for processing the biochemical indicators arranged in chronological order, such as the dynamic changes of lactic acid concentration and glucose level over running time, so as to capture the cumulative trend of metabolic fatigue. After the outputs of the two are concatenated or fused by attention, a unified "chemical+acoustic fusion fatigue index" is finally formed, which not only reflects the fatigue accumulation at the metabolic level but also integrates the mechanical fatigue information brought by gait impact.

[0038] For example, when a runner is doing a continuous 10-kilometer run, the system reads the concentrations of lactic acid and glucose in the sweat at regular intervals, and at the same time continuously collects the echo signals captured by the sole microphone array. As the running time progresses, the CNN will find that the impact peak in the acoustic signal gradually increases and the acoustic wave rebound delay lengthens, while the LSTM captures the continuous rise of the lactic acid concentration curve and the accelerated glucose consumption rate. When these two types of information are fused in the hybrid model, if the fatigue index exceeds a preset threshold (such as index 0.7), the system will determine that the runner is in a "mild fatigue" or higher-risk state and promptly alert the user to slow down or adjust breathing through in-shoe vibration or a mobile phone App. This fatigue assessment method based on the deep fusion of multi-source signals can detect latent fatigue earlier and more accurately than traditional methods that simply rely on heart rate or step frequency, helping athletes avoid overtraining and potential injuries.

[0039] In a possible implementation, an LSTM+CNN hybrid model is used to process the features of biochemical data and acoustic data to generate a chemical+acoustic fusion fatigue index, which specifically includes: using the LSTM+CNN hybrid model to extract features from the biochemical data to obtain biochemical features, where the biochemical features include the average lactate slope and the glucose drop rate; using the LSTM+CNN hybrid model to extract features from the acoustic data to obtain acoustic features, where the acoustic features include the energy spectral centroid per step and the peak value of the foot impact in the time domain; fusing the data acquisition times of the biochemical data and the acoustic data to obtain biochemical temporal features and acoustic temporal features; using the CNN unit in the LSTM+CNN hybrid model to perform convolution on the acoustic temporal features to extract local impact patterns; using the LSTM unit in the LSTM+CNN hybrid model to perform temporal splicing of the biochemical temporal features and the local impact patterns, and outputting the chemical+acoustic fusion fatigue index.

[0040] Specifically, first, for the lactate and glucose concentrations measured in sweat, the model calculates two biochemical features: the "average lactate slope" (i.e., the rising rate of lactate concentration per unit time) and the "glucose drop rate" (i.e., the decreasing amplitude of glucose concentration per unit time). At the same time, for the sound waves captured by the microphone at each step of landing, the model extracts two acoustic features: the "energy spectral centroid per step" (reflecting the distribution center of the sound wave energy in the frequency spectrum) and the "peak value of the foot impact in the time domain" (reflecting the maximum amplitude of a single landing impact). Then, the system arranges these two types of features into biochemical temporal features and acoustic temporal features according to their respective sampling times, preparing the input data for the subsequent deep learning module.

[0041] In the LSTM+CNN hybrid structure, first, the CNN unit performs one-dimensional convolution on the acoustic temporal features to automatically capture local patterns such as the impact waveform - such as recurring high-frequency oscillations or rebound delay details in the impact waveform. Then, the LSTM unit takes over and performs dynamic temporal modeling on the concatenated biochemical temporal features and the local impact patterns refined by the CNN, enabling the model to understand the fatigue evolution trend of "when lactate accumulates rapidly while the sole impact energy increases". Finally, the entire hybrid model outputs a "chemical+acoustic fusion fatigue index" between 0 and 1, which reflects both the metabolic stress level of the runner and comprehensively considers the mechanical load on the body caused by gait impacts, thus being able to more sensitively and accurately evaluate the fatigue state during exercise.

[0042] Example illustration: Suppose a runner is doing interval training. Every 3 minutes, the system reads the sweat concentration and calculates the current lactate slope and glucose drop rate; meanwhile, the acoustic data of each step is also recorded in real time. As the training progresses, at the 5th minute, the rising rate of lactate suddenly becomes faster, and the biochemical features extracted by the system significantly increase; acoustically, the CNN finds that the center of gravity of the acoustic energy spectrum shifts towards higher frequencies and the peak amplitude in the time domain increases, indicating that the impact of each step of the runner becomes stronger. When the LSTM sees these two types of features "rising" simultaneously in time, it raises the fatigue index to 0.75, exceeding the preset threshold. The system then reminds the runner through in-shoe vibration or the mobile phone App that "you are currently in mild fatigue, please relax appropriately or adjust your stride frequency". This not only avoids the lag of single-signal fatigue judgment but also effectively improves the sensitivity of fatigue monitoring.

[0043] S130. Use the LSTM+CNN hybrid model to process the vibration data to obtain the bone stress warning model.

[0044] Specifically, the controller inside the smart running shoes inputs the vibration signals collected from the bone conduction accelerometer into a deep model that integrates a convolutional neural network (CNN) and a long short-term memory network (LSTM). Through multi-level feature extraction and time series analysis, a "bone stress warning model" that can real-time warn about the stress on the foot bones is constructed. Specifically, the CNN subnet will automatically identify the high-frequency resonance peak features related to bone stress in the vibration signals, such as the resonance frequency shift caused by bone microcracks or stress concentration; while the LSTM subnet is good at capturing the trend changes of these features over time, such as the continuous attenuation of high-frequency energy or the gradual enhancement of the envelope wave peak under repeated impacts. After the combination of the two, the warning model can not only detect potential high-stress signals in a single impact but also pay attention to the process of small stress accumulation during long-term running, thus realizing the early detection of bone fatigue or potential injuries.

[0045] In a possible implementation manner, using the LSTM+CNN hybrid model to process the vibration data to obtain the bone stress warning model specifically includes: using the LSTM+CNN hybrid model to extract features from the vibration data to obtain vibration features, where the vibration features include the high-frequency energy attenuation rate, resonance peak shift, and time-domain envelope wave peak; using the CNN subnet in the LSTM+CNN hybrid model to determine the high-frequency resonance peak features in the vibration features; using the LSTM subnet in the LSTM+CNN hybrid model to determine the stress accumulation trend of the vibration features; combining the high-frequency resonance peak features and the stress accumulation trend to construct the bone stress warning model.

[0046] Specifically, first, the system will perform preliminary processing on the foot vibration waveforms collected by the accelerometer, calculating the "high-frequency energy attenuation rate" (i.e., the rate of decline of the high-frequency band energy in the frequency spectrum over time or steps), the "resonance peak shift" (i.e., the frequency peak drift caused by the change in microstructural stress), and the "time-domain envelope wave peak" (i.e., the extreme points of the envelope line of the vibration signal in the time domain). These vibration characteristics can reflect the instantaneous response and microscopic changes of the bone under continuous impacts.

[0047] Next, the CNN subnet in the hybrid model specifically performs convolution operations on the high-frequency resonance peaks in the vibration characteristics, automatically identifying and quantifying the fine changes in the peaks, such as the slight shift in the peak frequency band or abnormal shape; while the LSTM subnet models the extracted vibration characteristic sequences in the time dimension, capturing the trend of gradually accumulating stress during continuous running, such as the increasing high-frequency energy attenuation or the continuous increase in the envelope wave peak. When the CNN detects that the resonance peak characteristics deviate significantly from the normal state, and the LSTM reflects that the stress shows an upward trend over multiple gait cycles, the system can comprehensively judge that the bone is in a high-risk load state, thereby triggering an alarm and providing timely posture adjustment or rest suggestions for the runner.

[0048] S140. Use the LSTM+CNN hybrid model to process the biochemical data and electrode data to generate a biochemical+electromyogram interaction health situation.

[0049] Specifically, the controller in the intelligent running shoes inputs the biochemical indicators detected from sweat (such as cortisol increment, sodium ion concentration) and the electromyogram signals collected by the dry sEMG electrode array (such as average power spectrum and intermediate frequency power ratio) into a hybrid model that combines a convolutional neural network (CNN) and a long short-term memory network (LSTM). Specifically, first, the CNN module automatically learns the local characteristics of the muscle activation pattern on the time-frequency diagram of the electromyogram signal, such as the change in the energy distribution of the electromyogram frequency band or the typical spectral shift during fatigue; at the same time, the LSTM module focuses on modeling the dynamic evolution of these biochemical indicators and electromyogram characteristics over time, such as the sudden increase in cortisol level or the continuous decrease in electromyogram intermediate frequency energy. By performing cross-channel splicing and temporal correlation on the deep features of the two types of signals, the model finally outputs a "biochemical+electromyogram interaction health situation" score, reflecting whether the current metabolic state and muscle function of the runner are coordinately imbalanced.

[0050] In a possible implementation, an LSTM+CNN hybrid model is used to process the features of biochemical data and electrode data to generate a biochemical+EMG interaction health situation, which specifically includes: determining target biochemical features according to the LSTM+CNN hybrid model and biochemical data, where the target biochemical features include the dynamic increment of cortisol and sodium ion concentration; determining EMG features according to the LSTM+CNN hybrid model and electrode data, where the EMG features include the average power spectrum and the intermediate frequency power ratio; aggregating the EMG features by using the LSTM+CNN hybrid model to obtain EMG time-frequency map features; and interacting the EMG time-frequency map features and the target biochemical features through the LSTM+CNN hybrid model to obtain the biochemical+EMG interaction health situation.

[0051] Specifically, first, the system analyzes the biochemical data measured in sweat according to the LSTM+CNN hybrid model, and screens out the "target biochemical features" that can best represent endocrine stress and electrolyte balance, namely the dynamic increment of cortisol (reflecting the short-term changes in stress hormone levels) and sodium ion concentration (representing the body fluid electrolyte state). Then, the model processes the EMG signals collected by the dry EMG electrode array, and extracts the "EMG features", the average power spectrum (showing the overall energy distribution of muscles in the frequency domain) and the intermediate frequency power ratio (commonly used to indicate the shift of the frequency band energy to the low frequency when the muscle is fatigued). Subsequently, the CNN subnet further performs an aggregation operation on these EMG features, and maps them to the EMG time-frequency map features, enabling the model to capture the subtle changes in muscle activity in both the time and frequency dimensions.

[0052] After completing the above extraction of biochemical and EMG features, the hybrid model performs interactive modeling on the two types of time series features through the LSTM unit: it jointly analyzes the dynamic patterns in the EMG time-frequency map and the time series changes in cortisol and sodium ion levels, and uses the time series memory ability to capture the complex relationship of "how muscle fatigue manifests when endocrine stress rises and electrolyte balance shifts". Finally, the model outputs a "biochemical+EMG interaction health situation" score, and the higher this score is, the greater the disconnection or imbalance between the runner's metabolic state and muscle function.

[0053] Taking an example, imagine an athlete who, after completing an interval sprint, the running shoes detect that his cortisol level has increased by 20% within just a few minutes, and the sodium ion concentration has decreased by 5%; at the same time, the EMG signal shows that the intermediate frequency power ratio of his gastrocnemius muscle has decreased from 0.6 to 0.4. The CNN subnet extracts the obvious low-frequency trend in the EMG spectrum as the time-frequency map feature, and the LSTM notices the drastic fluctuations in the biochemical features at this time. After the model fusion, a relatively high health situation score is given, indicating that "the current metabolic stress is high and muscle fatigue is relatively high, and there is a metabolic-muscle coordination imbalance". Then the running shoes will remind the runner to slow down appropriately and replenish electrolyte drinks through vibration and LED to prevent over-fatigue or cramps.

[0054] S150 monitors the wearer's health based on the chemical + acoustic fusion fatigue index, the bone stress warning model, and the biochemical + electromyography interaction health status.

[0055] Specifically, the controller of the smart running shoes simultaneously receives and synthesizes the outputs from three in-depth evaluation models, namely the chemical + acoustic fusion fatigue index, the bone stress warning model, and the biochemical + electromyography interaction health status, so as to construct a real-time health monitoring system for the overall physical state of the runner. Specifically, the controller first synchronizes the three groups of indicators in time, and weights or rules-fuses the metabolic and mechanical fatigue levels reflected by the chemical + acoustic fatigue index, the bone load risks revealed by the bone stress warning model, and the metabolic-muscle coordination represented by the biochemical + electromyography interaction trend to form a unified health status evaluation result. Based on this comprehensive result, the system can make a comprehensive judgment on the runner's fatigue level, bone overload risk, and muscle function matching degree, and trigger various feedback mechanisms accordingly, including vibration prompts, LED light color warnings, and mobile app notifications, etc., to help users adjust their training plans or postures in time and avoid sports injuries.

[0056] In a possible implementation manner, monitoring the wearer's health based on the chemical + acoustic fusion fatigue index, the bone stress warning model, and the biochemical + electromyography interaction health status specifically includes: outputting a fatigue score, a stress risk score, and a health status score according to the chemical + acoustic fusion fatigue index, the bone stress warning model, and the biochemical + electromyography interaction health status; comparing the fatigue score, the stress risk score, and the health status score with their respective preset thresholds to obtain comparison results; generating a health feedback strategy according to the comparison results.

[0057] Specifically, after the independent evaluation of the three models is completed, the controller converts the three multi-dimensional health scores of "chemical + acoustic fusion fatigue index", "bone stress warning model", and "biochemical + electromyography interaction health status" into three specific scores (i.e., fatigue score, stress risk score, and health status score), and compares them with the pre-set thresholds respectively. The thresholds can be adjusted according to the user's training level or individual differences. For example, the fatigue score threshold is set to 0.7, the stress risk score threshold is set to 0.8, and the health status score threshold is set to 0.6. Through the comparison results, the controller can judge whether each risk indicator is in the normal range, the warning range, or the dangerous range, so as to provide a decision basis for the next intervention strategy.

[0058] For example, during a long-distance running training, the controller calculated that a runner had a fatigue score of 0.75 (over 0.7), a stress risk score of 0.85 (over 0.8), and a health status score of 0.55 (under 0.6). The comparison results show that the runner currently has obvious accumulation of biochemical and impact fatigue, as well as a risk of bone overload, but metabolic-electromyographic coordination is acceptable. The health feedback strategy generated by the system may be: immediately emit continuous vibration in the heel of the running shoe and light up the red LED, prompting "The bone load is too large, please slow down or pause training immediately"; at the same time, push text and voice instructions to the App via BLE, suggesting that users do targeted stretching recovery and replenish electrolyte drinks to prevent sports injuries.

[0059] In one possible implementation, refer to Figure 2 , Figure 2 Another flow chart of a health monitoring method for smart running shoes provided in an embodiment of the present application. The method includes steps S210 to S220, which are as follows: S210, obtaining the wearer's real-time blood oxygen data, real-time blood pressure data, and real-time heart rate data; S220, integrating the health feedback strategy with the real-time blood oxygen data, real-time blood pressure data, and real-time heart rate data, and the integrated display method includes LED prompts and BLE notifications.

[0060] Specifically, in addition to the fatigue, stress and health status scores generated based on the multimodal model, the controller inside the smart running shoes will continue to obtain the wearer's core vital signs, including blood oxygen saturation, blood pressure and heart rate, from additional physiological sensor modules. By monitoring these indicators in real time, the system can have a more comprehensive understanding of the user's current cardiopulmonary function and circulatory status, and integrate more dimensional information when generating health feedback strategies; for example, when the fatigue score and bone stress score are high, if a drop in blood oxygen or a sharp increase in heart rate is detected at the same time, the system will judge that the risk is more serious, thereby triggering more urgent prompts or interventions.

[0061] In specific implementation, the controller will combine the generated health feedback strategy with the latest measured blood oxygen, blood pressure, and heart rate data into a fusion message, and convey it to the user through two channels: one is to visually prompt the current status with different colors or flashing frequencies on the LED light ring on the heel or side of the running shoe; the other is to push the fusion information to the App of the smartphone or wearable terminal through low-power Bluetooth (BLE), displaying detailed vital sign values and recommended actions. For example, if the system detects that the runner's fatigue score exceeds the threshold and the blood oxygen drops to 92%, the LED switches to red and flashes rapidly, and the App receives a push: "Blood oxygen is low and fatigue is accumulated, please stop running immediately and take a deep breath or replenish oxygen." Such a multi-channel, real-time and targeted display can help runners quickly perceive their own status and make timely adjustments.

[0062] The present application also provides a health monitoring device for intelligent running shoes. Refer to Figure 3 , Figure 3 which is a schematic module diagram of a health monitoring device for intelligent running shoes provided by an embodiment of the present application. The device is a controller of the intelligent running shoes. The controller includes an acquisition module 31 and a processing module 32. Among them, the acquisition module 31 acquires multimodal acquisition data for the wearer of the intelligent running shoes. The multimodal acquisition data includes biochemical data, acoustic data, vibration data, and electrode data. The processing module 32 uses an LSTM+CNN hybrid model to perform feature processing on the biochemical data and the acoustic data to generate a chemical+acoustic fusion fatigue index. The processing module 32 uses an LSTM+CNN hybrid model to perform feature processing on the vibration data to obtain a bone stress warning model. The processing module 32 uses an LSTM+CNN hybrid model to perform feature processing on the biochemical data and the electrode data to generate a biochemical+electromyogram interaction health situation. The processing module 32 performs health monitoring on the wearer based on the chemical+acoustic fusion fatigue index, the bone stress warning model, and the biochemical+electromyogram interaction health situation.

[0063] In a possible implementation manner, the acquisition module 31 acquires multimodal acquisition data for the wearer of the intelligent running shoes. The multimodal acquisition data includes biochemical data, acoustic data, vibration data, and electrode data, specifically including: the acquisition module 31 acquires lactic acid, glucose, electrolytes, and cortisol sent by a microfluidic channel and an electrochemical electrode pre-installed in the intelligent running shoes. The processing module 32 performs differential amplification processing on lactic acid, glucose, electrolytes, and cortisol to obtain biochemical data. The acquisition module 31 acquires ultrasonic echo signals of foot strikes sent by a MEMS microphone array pre-installed in the intelligent running shoes. The processing module 32 performs band-pass filtering processing on the ultrasonic echo signals to obtain acoustic data. The acquisition module 31 acquires foot bone conduction vibration signals sent by a bone conduction accelerometer pre-installed in the intelligent running shoes. The processing module 32 performs low-noise differential processing on the foot bone conduction vibration signals to obtain vibration data. The acquisition module 31 acquires foot electromyogram signals sent by a dry sEMG electrode array pre-installed in the intelligent running shoes. The processing module 32 performs differential amplification processing on the foot electromyogram signals to obtain electrode data.

[0064] In a possible implementation, the processing module 32 uses an LSTM+CNN hybrid model to process the biochemical data and acoustic data to generate a chemical+acoustic fusion fatigue index, specifically including: the processing module 32 uses the LSTM+CNN hybrid model to extract features from the biochemical data to obtain biochemical features, and the biochemical features include the average lactate slope and the glucose drop rate; the processing module 32 uses the LSTM+CNN hybrid model to extract features from the acoustic data to obtain acoustic features, and the acoustic features include the center of gravity of the energy spectrum per step and the peak value of the foot impact in the time domain; the processing module 32 fuses the data acquisition times of the biochemical data and the acoustic data to obtain biochemical time-series features and acoustic time-series features; the processing module 32 uses the CNN unit in the LSTM+CNN hybrid model to perform convolution on the acoustic time-series features to extract local impact patterns; the processing module 32 uses the LSTM unit in the LSTM+CNN hybrid model to perform time-series splicing of the biochemical time-series features and the local impact patterns, and outputs a chemical+acoustic fusion fatigue index.

[0065] In a possible implementation, the processing module 32 uses an LSTM+CNN hybrid model to process the vibration data to obtain a bone stress warning model, specifically including: the processing module 32 uses the LSTM+CNN hybrid model to extract features from the vibration data to obtain vibration features, and the vibration features include the high-frequency energy attenuation rate, the resonance peak shift, and the peak value of the time-domain envelope wave; the processing module 32 uses the CNN subnet in the LSTM+CNN hybrid model to determine the high-frequency resonance peak features in the vibration features; the processing module 32 uses the LSTM subnet in the LSTM+CNN hybrid model to determine the stress accumulation trend of the vibration features; the processing module 32 combines the high-frequency resonance peak features and the stress accumulation trend to construct a bone stress warning model.

[0066] In a possible implementation, the processing module 32 uses an LSTM+CNN hybrid model to process the biochemical data and electrode data to generate a biochemical+EMG interaction health situation, specifically including: the processing module 32 determines target biochemical features according to the LSTM+CNN hybrid model and the biochemical data, and the target biochemical features include the dynamic increment of cortisol and the sodium ion concentration; the processing module 32 determines EMG features according to the LSTM+CNN hybrid model and the electrode data, and the EMG features include the average power spectrum and the intermediate frequency power ratio; the processing module 32 uses the LSTM+CNN hybrid model to aggregate the EMG features to obtain EMG time-frequency map features; the processing module 32 interacts the EMG time-frequency map features and the target biochemical features through the LSTM+CNN hybrid model to obtain a biochemical+EMG interaction health situation.

[0067] In a possible implementation, the processing module 32 monitors the wearer's health based on the chemical + acoustic fusion fatigue index, the bone stress warning model, and the biochemical + electromyogram interaction health situation, specifically including: the processing module 32 outputs a fatigue score, a stress risk score, and a health situation score according to the chemical + acoustic fusion fatigue index, the bone stress warning model, and the biochemical + electromyogram interaction health situation; the processing module 32 compares the fatigue score, the stress risk score, and the health situation score with their respective preset thresholds to obtain a comparison result; the processing module 32 generates a health feedback strategy according to the comparison result.

[0068] In a possible implementation, the acquisition module 31 acquires the wearer's real-time blood oxygen data, real-time blood pressure data, and real-time heart rate data; the processing module 32 fuses and displays the health feedback strategy with the real-time blood oxygen data, real-time blood pressure data, and real-time heart rate data, and the fusion display methods include LED prompts and BLE notifications.

[0069] It should be noted that when the device provided in the above embodiment realizes its functions, only the above-mentioned division of each functional module is used for illustration. In actual applications, the above functions can be allocated to different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the device and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be seen in the method embodiment, which will not be repeated here.

[0070] This application also provides an electronic device. Refer to Figure 4 , Figure 4 which is a schematic structural diagram of an electronic device provided in an embodiment of this application. The electronic device may include: at least one processor 41, at least one network interface 44, a user interface 43, a memory 45, and at least one communication bus 42.

[0071] Among them, the communication bus 42 is used to realize the connection and communication between these components.

[0072] Among them, the user interface 43 may include a display screen (Display) and a camera (Camera). Optionally, the user interface 43 may further include a standard wired interface and a wireless interface.

[0073] Among them, the network interface 44 may optionally include a standard wired interface and a wireless interface (such as a Wi-Fi interface).

[0074] Among them, the processor 41 may include one or more processing cores. The processor 41 connects various parts within the entire server through various interfaces and lines. By running or executing instructions, programs, code sets, or instruction sets stored in the memory 45, and by calling the data stored in the memory 45, it performs various functions of the server and processes data. Optionally, the processor 41 may be implemented in at least one hardware form of digital signal processing (DSP), field-programmable gate array (FPGA), or programmable logic array (PLA). The processor 41 may integrate a combination of one or several of a central processing unit (CPU), a graphics processing unit (GPU), and a modem, etc. Among them, the CPU mainly processes the operating system, user interface, application programs, etc.; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communication. It can be understood that the above-mentioned modem may not be integrated into the processor 41 and may be implemented separately by a single chip.

[0075] Among them, the memory 45 may include random access memory (RAM) and may also include read-only memory. Optionally, the memory 45 includes a non-transitory computer-readable storage medium. The memory 45 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 45 may include a program storage area and a data storage area. Among them, the program storage area may store instructions for implementing the operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store the data involved in the above-mentioned various method embodiments. Optionally, the memory 45 may also be at least one storage device located far from the aforementioned processor 41. As Figure 4 shown, the memory 45, as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application program for a health monitoring method applied to intelligent running shoes.

[0076] In Figure 4In the electronic device shown, the user interface 43 is mainly used to provide an interface for the user to input and obtain the data input by the user; while the processor 41 can be used to call an application program stored in the memory 45, which is a health monitoring method applied to intelligent running shoes. When executed by one or more processors, the electronic device executes one or more of the methods in the foregoing embodiments.

[0077] It should be noted that, for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that this application is not limited by the described action sequence, because according to this application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0078] This application also provides a computer-readable storage medium, which stores instructions. When executed by one or more processors, the electronic device executes one or more of the methods described in the foregoing embodiments.

[0079] In the above embodiments, the descriptions of the various embodiments have their own emphases. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0080] In several embodiments provided by this application, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of units is only a logical function division. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some service interfaces. The indirect coupling or communication connection of the device or unit can be in an electrical or other form.

[0081] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place, or they can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0082] In addition, in each embodiment of this application, the functional units can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0083] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in various embodiments of the present application. The aforementioned memory includes various media such as USB flash drives, external hard drives, magnetic disks, or optical discs that can store program codes.

[0084] The foregoing are only exemplary embodiments of the present disclosure and should not be used to limit the scope of the present disclosure. That is, all equivalent changes and modifications made in accordance with the teachings of the present disclosure still fall within the scope covered by the present disclosure. After considering the specification and the disclosure of the practical truth, those skilled in the art will readily think of other implementation manners of the present disclosure. The present application is intended to cover any variations, uses, or adaptive changes of the present disclosure, and these variations, uses, or adaptive changes follow the general principles of the present disclosure and include the common general knowledge or conventional technical means in the technical field not recorded in the present disclosure. The specification and the embodiments are only regarded as exemplary, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A health monitoring method applied to intelligent running shoes, characterized in that, A controller applied to intelligent running shoes, the method comprising: Obtaining multi-modal acquisition data for the wearer of the intelligent running shoes, the multi-modal acquisition data including biochemical data, acoustic data, vibration data, and electrode data; Using an LSTM+CNN hybrid model to perform feature processing on the biochemical data and the acoustic data to generate a chemical+acoustic fusion fatigue index; Using the LSTM+CNN hybrid model to perform feature processing on the vibration data to obtain a bone stress warning model; Using the LSTM+CNN hybrid model to perform feature processing on the biochemical data and the electrode data to generate a biochemical+electromyogram interaction health situation; Based on the chemical+acoustic fusion fatigue index, the bone stress warning model, and the biochemical+electromyogram interaction health situation, performing health monitoring on the wearer.

2. The health monitoring method applied to intelligent running shoes according to claim 1, characterized in that, The obtaining of the multi-modal acquisition data for the wearer of the intelligent running shoes, the multi-modal acquisition data including biochemical data, acoustic data, vibration data, and electrode data, specifically includes: Obtaining lactic acid, glucose, electrolytes, and cortisol sent by a microfluidic channel and an electrochemical electrode pre-installed in the intelligent running shoes; Performing differential amplification processing on the lactic acid, glucose, electrolytes, and cortisol to obtain the biochemical data; Obtaining an ultrasonic echo signal of a foot strike sent by a MEMS microphone array pre-installed in the intelligent running shoes; Performing band-pass filtering processing on the ultrasonic echo signal to obtain the acoustic data; Obtaining a foot bone conduction vibration signal sent by a bone conduction accelerometer pre-installed in the intelligent running shoes; Performing low-noise differential processing on the foot bone conduction vibration signal to obtain the vibration data; Obtaining a foot electromyogram signal sent by a dry sEMG electrode array pre-installed in the intelligent running shoes; Performing differential amplification processing on the foot electromyogram signal to obtain the electrode data.

3. The health monitoring method applied to intelligent running shoes according to claim 1, characterized in that The using of the LSTM+CNN hybrid model to perform feature processing on the biochemical data and the acoustic data to generate a chemical+acoustic fusion fatigue index, specifically includes: Using the LSTM+CNN hybrid model to perform feature extraction on the biochemical data to obtain biochemical features, the biochemical features including an average lactic acid slope and a glucose drop rate; Using the LSTM+CNN hybrid model to perform feature extraction on the acoustic data to obtain acoustic features, the acoustic features including an energy spectral centroid per step and a peak value in the foot strike time domain; Fusing the data acquisition times of the biochemical data and the acoustic data to obtain biochemical time series features and acoustic time series features; Using the CNN unit in the LSTM+CNN hybrid model to perform convolution on the acoustic time series features to extract local impact patterns; Using the LSTM unit in the LSTM+CNN hybrid model to perform time series splicing of the biochemical time series features and the local impact patterns, and outputting the chemical+acoustic fusion fatigue index.

4. The health monitoring method applied to intelligent running shoes according to claim 1, characterized in that, The using of the LSTM+CNN hybrid model to perform feature processing on the vibration data to obtain a bone stress warning model, specifically includes: Use the LSTM+CNN hybrid model to extract features from the vibration data to obtain vibration features, where the vibration features include high-frequency energy decay rate, resonance peak shift, and time-domain envelope wave peak; Use the CNN subnet in the LSTM+CNN hybrid model to determine the high-frequency resonance peak features in the vibration features; Use the LSTM subnet in the LSTM+CNN hybrid model to determine the stress accumulation trend of the vibration features; Combine the high-frequency resonance peak features and the stress accumulation trend to construct the bone stress warning model.

5. The health monitoring method applied to intelligent running shoes according to claim 1, characterized in that Use the LSTM+CNN hybrid model to perform feature processing on the biochemical data and the electrode data to generate a biochemical+EMG interaction health situation, specifically including: Determine target biochemical features according to the LSTM+CNN hybrid model and the biochemical data, where the target biochemical features include cortisol dynamic increment and sodium ion concentration; Determine EMG features according to the LSTM+CNN hybrid model and the electrode data, where the EMG features include average power spectrum and intermediate frequency power ratio; Use the LSTM+CNN hybrid model to aggregate the EMG features to obtain EMG time-frequency map features; Perform interaction on the EMG time-frequency map features and the target biochemical features through the LSTM+CNN hybrid model to obtain the biochemical+EMG interaction health situation.

6. The health monitoring method applied to intelligent running shoes according to claim 1, characterized in that, Based on the chemical+acoustic fusion fatigue index, the bone stress warning model, and the biochemical+EMG interaction health situation, perform health monitoring on the wearer, specifically including: Output a fatigue score, a stress risk score, and a health situation score according to the chemical+acoustic fusion fatigue index, the bone stress warning model, and the biochemical+EMG interaction health situation; Compare the fatigue score, the stress risk score, and the health situation score with their respective preset thresholds to obtain a comparison result; Generate a health feedback strategy according to the comparison result.

7. The health monitoring method applied to smart running shoes according to claim 6, characterized in that, The method further includes: Obtain the real-time blood oxygen data, real-time blood pressure data, and real-time heart rate data of the wearer; Fusion display the health feedback strategy with the real-time blood oxygen data, the real-time blood pressure data, and the real-time heart rate data, and the fusion display methods include LED prompts and BLE notifications.

8. A health monitoring device applied to smart running shoes, characterized in that, The device is a controller of an intelligent running shoe, and the controller of the intelligent running shoe includes an acquisition module (31) and a processing module (32), where, The acquisition module (31) is used to acquire multi-modal acquisition data of the wearer of the intelligent running shoe, and the multi-modal acquisition data includes biochemical data, acoustic data, vibration data, and electrode data; The processing module (32) is used to perform feature processing on the biochemical data and the acoustic data by using an LSTM+CNN hybrid model to generate a chemical+acoustic fusion fatigue index; The processing module (32) is further used to perform feature processing on the vibration data by using the LSTM+CNN hybrid model to obtain a bone stress warning model; The processing module (32) is further configured to perform feature processing on the biochemical data and the electrode data by using the LSTM+CNN hybrid model to generate a biochemical+electromyogram interaction health situation; The processing module (32) is further configured to perform health monitoring on the wearer based on the chemical+acoustic fusion fatigue index, the bone stress warning model, and the biochemical+electromyogram interaction health situation.

9. An electronic device, characterized in that, The electronic device includes a processor (41), a memory (45), a user interface (43), and a network interface (44). The memory (45) is used for storing instructions. Both the user interface (43) and the network interface (44) are used for communicating with other devices. The processor (41) is configured to execute the instructions stored in the memory (45) so that the electronic device executes the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions which, when executed, perform the method according to any one of claims 1 to 7.

Images