Control method for cardiac rehabilitation training based on real-time monitoring of muscle strength and electrocardiosignal

Abstract

The application relates to the technical field of medical health information, and particularly discloses a heart rehabilitation training control method based on real-time monitoring of muscle strength and electrocardiosignal. The method first calculates real-time electromyographic intensity to quantify muscle load, synchronously analyzes electrocardiosignal and respiratory frequency to obtain cardiopulmonary state, and corrects physiological indexes in combination with trunk movement state. Then, a weighted fusion algorithm is used to generate a multi-parameter fusion result, which comprehensively reflects the correlation between muscle load and heart response. Next, the system judges training intensity through a preset threshold model, generates a voice feedback instruction in real time to adjust the action of a patient, and adaptively updates training parameters to match individual rehabilitation progress. Meanwhile, the system continuously monitors multi-parameter changes, automatically triggers an early warning mechanism when detecting abnormal risks, and ensures training safety. The method overcomes the limitation of traditional schemes relying on a single index, and realizes personalized and dynamic closed-loop control of heart rehabilitation training.

Description

Technical Field

[0001] This invention relates to the field of medical and health information technology, and more specifically, to a method for controlling cardiac rehabilitation training based on real-time monitoring of muscle strength and electrocardiogram signals. Background Technology

[0002] Cardiac rehabilitation is a comprehensive, multi-dimensional medical intervention, with its core component being an individualized exercise training plan developed for heart disease patients. With the advancement of information and communication technologies, modern cardiac rehabilitation increasingly utilizes wearable sensing devices and health information systems to monitor and manage the patient's physiological state during training, ensuring the safety and effectiveness of the training. These systems process healthcare data to provide decision support for both patients and physicians.

[0003] In existing technologies, systems for monitoring cardiac rehabilitation typically use wearable devices to collect one or more basic physiological parameters of the patient. For example, commonly used techniques involve collecting the patient's electrocardiogram (ECG) signals using heart rate straps or ECG patches to monitor heart rate changes, or using accelerometers in devices such as wristbands or watches to estimate exercise intensity. The system then compares this collected data with preset, universally accepted safety thresholds, such as controlling the patient's heart rate within a target range, to guide training.

[0004] However, the aforementioned existing technical solutions have significant limitations. First, relying solely on heart rate as a feedback indicator of training intensity is unreliable, as heart rate is not only affected by physical load but also easily influenced by various non-motor factors such as patient emotions, postural changes, and drug effects. This makes it difficult for the system to accurately assess the true cardiac load caused by the training itself. Second, existing systems typically lack direct monitoring of the patient's actual muscle work, making it impossible to distinguish the different stresses exerted on the cardiovascular system by different training movements. Finally, their control logic is mostly based on fixed threshold judgments, lacking the ability to adaptively adjust training goals according to the patient's long-term rehabilitation progress, thus hindering the achievement of truly in-depth individualized treatment. Summary of the Invention

[0005] In view of this, in order to solve the problems mentioned in the background art, a cardiac rehabilitation training control method based on real-time monitoring of muscle strength and electrocardiogram signals is proposed.

[0006] The objective of this invention can be achieved through the following technical solution: This invention provides a cardiac rehabilitation training control method based on real-time monitoring of muscle strength and electrocardiogram signals, including the following steps: S1, real-time electromyographic intensity calculation: acquire the electromyographic signals and motion signals of the patient's limbs, and calculate the real-time electromyographic intensity based on the electromyographic signals and motion signals.

[0007] S2. Respiratory Rate Analysis: Acquire the patient's chest electrocardiogram (ECG) and respiratory signals, obtain real-time ECG indicators based on ECG signal analysis, and obtain the respiratory rate based on respiratory signal analysis.

[0008] S3. Trunk motion state determination: Acquire the patient's trunk acceleration signal and determine the trunk motion state based on the trunk acceleration signal.

[0009] S4. Generation of multi-parameter fusion results: The real-time electromyography intensity, real-time electrocardiogram indicators, respiratory rate and trunk movement status are fused to generate multi-parameter fusion results.

[0010] S5. Feedback control command generation: Based on the multi-parameter fusion results, generate feedback control commands for adjusting cardiac rehabilitation training.

[0011] S6. Adaptive Adjustment of Training Parameters: Stores historical training data and adaptively adjusts training parameters.

[0012] S7. Abnormal Risk Detection and Early Warning: Real-time monitoring of multi-parameter fusion results to detect abnormal risks and automatically issue early warnings.

[0013] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects: (1) The present invention quantifies the external training load by fusing muscle strength signals and motion signals, and at the same time combines multi-dimensional physiological signals such as electrocardiogram and respiration to evaluate the internal physiological response, and uses trunk posture information for correction, thereby achieving an accurate and comprehensive assessment of the patient's true physiological load state. This multi-parameter fusion analysis method overcomes the one-sidedness and susceptibility to interference of relying solely on indicators such as heart rate, and provides a more reliable basis for subsequent control decisions.

[0014] (2) By storing and analyzing patients' training data over a long period of time, the system can identify the improvement trend of their cardiopulmonary function and automatically adjust the training parameters and feedback logic accordingly. This makes the rehabilitation plan no longer static, but intelligently evolves in close accordance with the patient's rehabilitation progress, ensuring that the training intensity is always kept within the personalized range with the most therapeutic benefits, thereby maximizing the rehabilitation effect.

[0015] (3) This invention constructs a complete closed-loop feedback control and proactive safety early warning system. The system can transform complex physiological assessment results into simple and intuitive voice commands to guide patients to adjust their training in real time. At the same time, through close monitoring of key indicators and their rate of change, it can identify and warn of potential cardiovascular risks in advance. This automated real-time intervention and remote alarm mechanism greatly improves the safety of cardiac rehabilitation training, and is especially suitable for home rehabilitation scenarios without supervision, enhancing patients' training compliance and safety. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. 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.

[0017] Figure 1 This is a schematic diagram of the method steps of the present invention.

[0018] Figure 2 This is a diagram of the cardiac rehabilitation training and control system architecture of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Please see Figure 1 and Figure 2 The present invention provides a cardiac rehabilitation training control method based on real-time monitoring of muscle strength and electrocardiogram signals, including: S1, real-time electromyography intensity calculation: acquiring electromyography signals and motion signals of the patient's limbs, and calculating real-time electromyography intensity based on the electromyography signals and motion signals.

[0021] In a specific embodiment of the present invention, the specific steps for calculating real-time electromyographic intensity based on electromyographic signals and motion signals include: collecting electromyographic signals of the patient's limbs through an electromyographic monitoring device.

[0022] The motion signals of the patient's limbs are collected by an accelerometer in the electromyography (EMG) monitoring device.

[0023] It should be noted that the process of calculating real-time electromyography (EMG) intensity in this method involves first wearing an EMG monitoring device, which integrates surface electrodes and a triaxial accelerometer, on the target limb for rehabilitation training, such as the biceps brachii of the upper arm or the quadriceps femoris of the thigh. When the patient begins the training movement, the EMG monitoring device simultaneously starts data acquisition. The surface electrodes, in close contact with the skin, capture the weak bioelectrical changes generated by the target muscle group during contraction and relaxation, forming the raw EMG signal. Simultaneously, the device's built-in accelerometer measures the limb's acceleration in three-dimensional space in real time, which is the motion signal.

[0024] By combining electromyographic (EMG) signals with motion signals, the real-time EMG intensity is calculated.

[0025] It should be noted that after acquiring the two raw signals, the system processes them to calculate the real-time muscle load level. For the electromyography (EMG) signal, the system first performs bandpass filtering to remove noise such as power frequency interference and motion artifacts, then performs full-wave rectification, converting the signal into its root mean square (RMS) value within a specific time window, which effectively characterizes the degree of muscle activity. For the motion signal, the system uses high-pass filtering to separate the dynamic acceleration component generated by active limb movement and calculates its vector magnitude to quantify the intensity of the movement. Finally, the amplitudes of the processed EMG signal and the motion signal are normalized, and then a weighted fusion algorithm is used to generate a composite index that comprehensively reflects muscle physiological activation and actual physical work, namely, real-time EMG intensity. The calculation of real-time EMG intensity can be expressed as: ,in, represent Real-time electromyographic intensity at any given moment; yes The effective value of the time-normalized electromyography signal is obtained by dividing the real-time RMS value of the electromyography signal by the peak value of the electromyography signal measured in the patient under maximal voluntary contraction. yes The time-normalized limb dynamic acceleration is obtained by dividing the real-time dynamic acceleration value by a preset reference maximum acceleration under the training action. and These are preset weighting coefficients used to adjust the relative importance of muscle activation and range of motion in the final evaluation. .

[0026] In one specific embodiment of the present invention, when calculating real-time electromyography intensity, a preset weighting coefficient is used. and These are used to adjust the relative importance of muscle activation and range of motion in the final evaluation, among which... It can be set to 0.7. It can be set to 0.3. This value is based on the fact that muscle activation, as a key indicator reflecting the actual physiological work done by muscles, plays a central role in assessing muscle load during rehabilitation training, and therefore is given a high weight; while range of motion, although reflecting the range of limb activity, cannot comprehensively assess physiological exertion with a single motion signal, so it is given a lower weight. Through this weighting, it can be ensured that real-time electromyographic intensity more accurately reflects both muscle physiological activation and actual physical work.

[0027] This invention achieves precise quantification of limb muscle load by fusing electromyographic (EMG) and motion signals. It overcomes the limitations of relying solely on EMG signals, which cannot distinguish between isometric and isotonic contractions, or relying solely on motion signals, which cannot assess physiological exertion. By combining the electrophysiological activation level of muscles with the actual dynamic movement of the limbs, the calculated real-time EMG intensity more comprehensively and accurately reflects the patient's true muscle output and physiological load during rehabilitation training. This provides a reliable input basis for closed-loop control of subsequent cardiac rehabilitation training intensity, thereby improving the safety and effectiveness of training.

[0028] S2. Respiratory Rate Analysis: Acquire the patient's chest electrocardiogram (ECG) and respiratory signals, obtain real-time ECG indicators based on ECG signal analysis, and obtain the respiratory rate based on respiratory signal analysis.

[0029] In a specific embodiment of the present invention, the steps of obtaining real-time electrocardiogram indicators based on electrocardiogram signal analysis and obtaining respiratory rate based on respiratory signal analysis include: acquiring electrocardiogram signals and respiratory signals from the patient's chest through an electrocardiogram acquisition device.

[0030] It should be noted that the real-time assessment of the patient's cardiopulmonary function in this method begins with the simultaneous acquisition of electrocardiogram (ECG) and respiratory signals through an integrated ECG acquisition device worn on the patient's chest. This device is typically a patch or chest strap to ensure stable acquisition of high-quality physiological data during exercise.

[0031] By analyzing the waveform and rhythm characteristics of electrocardiogram (ECG) signals, real-time ECG indicators including real-time heart rate, heart rate variability, and the risk of abnormal ECGs are obtained.

[0032] It should be noted that for ECG signal processing, the system first performs a series of preprocessing operations, including applying digital filters to remove baseline drift, power line interference, and electromyographic noise to purify the ECG waveform. Subsequently, the system employs high-precision QRS complex detection algorithms, such as wavelet transform-based or adaptive thresholding methods, to accurately locate the R-wave peaks in each cardiac cycle. Based on the continuously detected R-wave peak positions, the system can calculate the time interval between adjacent R-wave peaks, i.e., the RR interval. Real-time heart rate is calculated from the RR intervals, typically expressed as heartbeats per minute. Heart rate variability analysis is based on a series of continuous RR interval data. The system calculates HRV parameters in the time or frequency domain within a sliding short time window. For example, in the time domain, it calculates the root mean square of the difference between adjacent RR intervals, an indicator that reflects parasympathetic regulatory activity. The assessment of the risk of abnormal ECGs is a more complex pattern recognition process. The system analyzes the waveform characteristics of individual heartbeats, such as whether ST segment deviation occurs and whether the T wave morphology is abnormal. It also analyzes heart rhythm characteristics, such as the presence of premature beats and extreme irregularity of the RR interval. By comparing these real-time extracted features with a pre-set library of normal and various typical abnormal ECG patterns, the system can output a quantified risk level, indicating whether there is a potential risk of myocardial ischemia, arrhythmia, or other conditions. These three factors—real-time heart rate, heart rate variability parameters, and abnormal ECG risk level—together constitute a comprehensive real-time ECG indicator.

[0033] Analyze the waveform period of the respiratory signal to obtain the respiratory rate.

[0034] It should be noted that for respiratory rate analysis, the system acquires respiratory signals from an ECG acquisition device, which can be obtained through chest impedance analysis or by extracting respiratory modulation information from the ECG signal. The system filters the raw respiratory signal to enhance the characteristics of the respiratory cycle and suppress motion artifacts. Then, a peak detection algorithm is used to identify the inspiratory or expiratory peaks in the respiratory waveform. The respiratory rate is obtained by calculating the number of consecutive peaks occurring within a specific time window, and the calculation formula is as follows: ,in, This indicates respiratory rate, measured in breaths per minute. In the time window The number of complete respiratory cycles detected internally; This represents the duration of the time window, in seconds.

[0035] This invention, through simultaneous analysis of the waveform and rhythm characteristics of electrocardiogram (ECG) signals and the periodicity of respiratory signals, enables real-time, multi-dimensional monitoring of a patient's cardiopulmonary status during rehabilitation training. Its technical advantages lie in providing not only basic heart rate and respiratory rate data, but also in-depth assessment of the autonomic nervous system's balance through heart rate variability and crucial safety warnings through abnormal ECG risk analysis. This comprehensive physiological indicator monitoring system provides a complete and profound insight into the physiological response to the patient's training load, offering core physiological evidence for truly individualized and safe cardiac rehabilitation training, and significantly enhancing the safety and scientific rigor of the training process.

[0036] S3. Trunk motion state determination: Acquire the patient's trunk acceleration signal and determine the trunk motion state based on the trunk acceleration signal.

[0037] In a specific embodiment of the present invention, the specific steps for determining the trunk motion state based on the trunk acceleration signal include: acquiring the patient's trunk acceleration signal through an acceleration sensor in an electrocardiogram acquisition device.

[0038] It should be noted that the process of determining the trunk motion state in this method involves using a triaxial accelerometer integrated into the electrocardiogram (ECG) acquisition device worn on the patient's chest to collect raw triaxial acceleration data reflecting trunk posture and motion in real time; this is the trunk acceleration signal. This signal simultaneously includes dynamic acceleration components generated by human movement and static gravitational acceleration components generated by Earth's gravity.

[0039] The patient's trunk acceleration signal was filtered to obtain a preprocessed trunk acceleration signal.

[0040] It should be noted that, to accurately extract posture information, the system first applies a low-pass filter to the acquired torso acceleration signal. The purpose of this filtering is to remove high-frequency dynamic acceleration noise caused by limb training movements, breathing, or minor body swaying, thereby separating and retaining the stable gravitational acceleration component in the signal. After filtering, a smooth, pre-processed torso acceleration signal that accurately reflects the torso's position relative to gravity is obtained. This signal has three components. It mainly characterizes the projection of gravitational acceleration onto the three orthogonal axes of the sensor.

[0041] Based on the gravitational acceleration component in the preprocessed trunk acceleration signal, trunk motion states including sitting or lying positions are classified.

[0042] It should be noted that the system subsequently classifies the torso motion state based on the gravitational acceleration component in the preprocessed torso acceleration signal. The core basis for classification is calculating the torso's tilt angle, i.e., the angle between the torso coordinate system and the vertical coordinate system of the earth. A certain axis of the sensor (such as the Y-axis) can be predefined to be approximately parallel to the direction of gravity when the human body is upright. Therefore, the torso's tilt angle relative to the vertical direction... It can be calculated using the following formula: ,in, It is the torso angle; It is the component of the preprocessed torso acceleration signal on the predetermined vertical axis (Y-axis); It is the vector magnitude of the gravitational acceleration component, i.e. Its value is approximately equal to a gravitational acceleration constant under static or quasi-static conditions; , , These are the triaxial components obtained from the preprocessed torso acceleration signal; by setting an angle threshold of 30 degrees, the system can perform state judgment, and when the calculated tilt angle... When the angle is less than the threshold, it indicates that the torso is nearly upright, and the system classifies the current torso movement as a sitting state; when the tilt .... When the angle exceeds the threshold, it indicates that the torso has significantly deviated from the vertical posture, and the system determines that the torso movement state is a lying position.

[0043] This invention achieves reliable identification of a patient's basic body posture by analyzing the static gravity component in the trunk acceleration signal. Its technical advantage lies in providing crucial background state information for the entire monitoring and control system. In cardiac rehabilitation training, the body's baseline heart rate, blood pressure, and other physiological parameters inherently differ between sitting and lying positions. Accurately acquiring trunk motion status enables the system to effectively distinguish between physiological changes caused by training load and those caused by posture changes in subsequent multi-parameter fusion analysis, thereby avoiding misjudgments of the patient's condition and significantly improving the accuracy and reliability of subsequent physiological indicator correction and training status assessment.

[0044] S4. Generation of multi-parameter fusion results: The real-time electromyography intensity, real-time electrocardiogram indicators, respiratory rate and trunk movement status are fused to generate multi-parameter fusion results.

[0045] In a specific embodiment of the present invention, the real-time electromyography intensity and real-time electrocardiogram indicators are timestamped to obtain a synchronized data stream.

[0046] It should be noted that the core of generating multi-parameter fusion results in this method lies in the deep integration and intelligent analysis of multi-source heterogeneous physiological signals. First, the system processes two independent data streams from the electromyography (EMG) monitoring device and the electrocardiogram (ECG) acquisition device. Since these devices may have different sampling rates and internal clocks, the system performs a timestamp alignment operation. This operation adds a high-precision timestamp to each acquired real-time EMG intensity data point and real-time ECG index data point, and through interpolation or resampling algorithms, maps the two data streams onto a unified time axis, forming a time-synchronized data stream. This ensures that each muscle load accurately corresponds to the cardiac response at the same moment.

[0047] Based on trunk movement status and respiratory rate, real-time ECG indicators in the synchronous data stream are corrected to generate corrected ECG indicators.

[0048] It should be noted that the system then corrects the real-time ECG indicators in the synchronous data stream to eliminate interference from non-training load factors. Based on the established trunk motion state and the real-time calculated respiratory rate, the system invokes a physiological baseline correction model. When the trunk motion state changes from a lying to a sitting position, the body's baseline heart rate naturally increases; this change is not caused by training. Similarly, changes in respiratory rate affect heart rate variability through respiratory sinus arrhythmia. The correction process quantifies and isolates these effects. The corrected ECG indicators, taking the corrected heart rate as an example, can be expressed by the following formula: ,in, yes Corrected electrocardiogram parameters at any given time, such as corrected heart rate; It was obtained from the synchronous data stream. The raw, real-time heart rate at any given moment; It is a postural heart rate compensation value determined based on the current trunk movement state. This value is derived from the resting heart rate difference under different postures pre-calibrated for the patient. This is a respiratory effect compensation value determined based on the current respiratory rate, used to counteract the periodic effect of respiration on heart rate rhythm. Through this step, the generated corrected electrocardiogram indicators can more purely reflect the cardiac physiological response caused by limb muscle load.

[0049] It should also be noted that the postural heart rate compensation value is determined by the difference in resting heart rate between different postures, such as sitting and lying down, which are pre-calibrated for the patient. The specific process is as follows: when the patient is in a quiet, non-training state, their resting heart rate values ​​in both sitting and lying down postures are measured and recorded; then, the difference in resting heart rate between the two postures is calculated, and this difference is used as the postural heart rate compensation value.

[0050] It should also be noted that the specific process of obtaining the respiratory effect compensation value is as follows: The system synchronously acquires the patient's respiratory signal through the electrocardiogram acquisition device and analyzes it to obtain the real-time respiratory rate; in specific operation, the peak detection algorithm is first used to identify the inspiratory / expiratory peak points in the respiratory waveform and calculate the respiratory rate within a specific time window; then, based on the model established by clinical research or historical data, the amplitude of heart rate rhythm fluctuation corresponding to different respiratory rate ranges is determined, and this amplitude is used as the respiratory effect compensation value; finally, during the training process, the system calls the corresponding respiratory effect compensation value according to the real-time respiratory rate to correct the electrocardiogram indicators such as heart rate variability in the synchronous data stream in order to eliminate the periodic interference caused by breathing.

[0051] Real-time electromyographic intensity and corrected electrocardiographic indices are input into a correlation model describing the relationship between muscle load and cardiac response to generate multi-parameter fusion results.

[0052] It should be noted that, finally, the system inputs two key parameters—real-time electromyographic intensity representing training input and corrected electrocardiogram indicators representing purified physiological output—into a pre-trained correlation model. After the input data is weighted and fused by the parameters within the correlation model, a comprehensive quantitative value is output, reflecting the patient's heart's physiological adaptation to muscle load under the current training intensity. This quantitative value, as a multi-parameter fusion result, is used for subsequent training status assessment and feedback control.

[0053] In a specific embodiment of the present invention, the correlation model describing the relationship between muscle load and cardiac response is obtained through the following steps: collecting physiological data and corresponding expert evaluation labels generated synchronously by multiple patients during training under professional guidance, and constructing multi-patient training data.

[0054] It should be noted that the construction of the correlation model describing the relationship between muscle load and cardiac response in this method is a data-driven process based on supervised machine learning. This process begins with the construction of large-scale, high-quality training data. Specifically, multiple cardiac patients with different physical conditions and rehabilitation stages are recruited and undergo a series of standardized rehabilitation training under the professional guidance and close monitoring of rehabilitation physicians or therapists. Throughout the training process, the system synchronously and continuously records the real-time electromyographic intensity and corrected electrocardiogram indicators generated in the aforementioned steps, forming paired input data sequences. Simultaneously, the monitoring clinical experts assess and label the training appropriateness of the patients at any given time based on their professional knowledge and comprehensive observation of the patients' conditions. These expert assessment labels are discrete classifications, such as "moderate intensity," "low intensity," "mild overload," and "dangerous overload," which constitute the target output required for model training. Thus, a multi-patient training dataset containing a large number of triples (real-time electromyographic intensity, corrected electrocardiogram indicators, and expert assessment labels) is constructed.

[0055] By using training data from multiple patients, iterative training is performed through machine learning algorithms to obtain optimized parameters for the correlation model.

[0056] It's important to note that after acquiring training data from multiple patients, the model enters the iterative training phase. First, the dataset is divided into training, validation, and test sets. Then, a suitable machine learning algorithm, such as a deep neural network, is selected as the basic framework for the association model. The training process involves using the training set to repeatedly adjust the model's internal parameters through an optimization algorithm. In each iteration, the model makes predictions based on the input real-time electromyography (EMG) intensity and corrected electrocardiogram (ECG) indicators, comparing the prediction results with the corresponding expert evaluation labels in the training set and calculating the error between the two. This error is fed back to the optimization algorithm to guide the model parameters in fine-tuning towards minimizing the error. This process is repeated multiple times on the entire training set until the model's performance on the validation set no longer shows significant improvement; the model parameters obtained at this point are the optimized association model parameters.

[0057] The association model is constructed using optimized association model parameters, and its accuracy is verified using test data.

[0058] It should be noted that, finally, these optimized association model parameters are solidified and used to construct the final association model. To verify the model's generalization ability and reliability, a final performance evaluation of the association model must be performed using test data that was not involved in the training process. By automatically calculating the model's accuracy, precision, recall, and other performance metrics on the test data, the consistency between the evaluation results and expert judgment is verified. Only after ensuring that the model's accuracy meets clinical application standards can it be deployed in a real-world system.

[0059] This invention, through three consecutive steps—timestamp alignment, physiological indicator correction, and correlation model evaluation—transforms multi-source raw data into a single, high-order, and interpretable fusion result. Its technical advantage lies in transcending isolated monitoring of various physiological indicators, establishing a dynamic assessment system that reflects the intrinsic correlation between muscle load and cardiac response. By actively eliminating interference from confounding factors such as posture and respiration, this method significantly improves the accuracy of the assessment, enabling the system to accurately determine whether the patient's heart has achieved adequate physiological adaptation to the current rehabilitation training or whether there is a potential risk of overload. This provides a scientific and reliable basis for subsequent precise feedback control.

[0060] S5. Feedback control command generation: Based on the multi-parameter fusion results, generate feedback control commands for adjusting cardiac rehabilitation training.

[0061] In a specific embodiment of the present invention, the specific steps of generating feedback control instructions for adjusting cardiac rehabilitation training include: comparing the multi-parameter fusion result with a threshold model that defines different training intensities to determine the training state.

[0062] It should be noted that this method generates feedback control commands to achieve closed-loop regulation of the training process. The specific implementation steps are as follows: First, the system continuously receives the multi-parameter fusion result generated in the previous step. This multi-parameter fusion result is a dynamically changing quantified value. The system has a preset threshold model that defines different training intensities. This model is essentially a set of boundary points, dividing the entire possible range of values ​​for the multi-parameter fusion result into several non-overlapping intervals. Each interval corresponds to a specific training state. For example, the threshold model can define several key training states such as "low intensity," "moderate intensity," "mild overload," and "severe overload." The system compares the real-time multi-parameter fusion result with this threshold model in real time to determine which interval it falls into, thereby determining the patient's current training state.

[0063] Based on the training state determined by the judgment, voice prompt instructions are matched from the instruction library that stores the mapping relationship between instructions and states.

[0064] It's important to note that once the training status is determined, the system queries a built-in instruction library. This library is a pre-built database that stores the mapping between training statuses and specific voice prompts. This mapping is pre-set by rehabilitation experts based on their clinical experience. For example, when the system determines the training status to be "low intensity," the library will match voice prompts such as "Please increase the range of motion appropriately" or "Please speed up the movements slightly"; if it determines it to be "slightly overloaded," it will match instructions such as "Please slow down the movements and adjust your breathing" or "Please reduce the height of the lift"; and when the status is "moderate intensity," it will output encouraging instructions such as "You're holding up well, please continue." These instructions cover suggestions for adjusting aspects such as limb range of motion, breathing rhythm, and training posture.

[0065] Voice prompts are output to audio devices to adjust the range of limb movements, breathing rhythm, or training posture.

[0066] It should be noted that, finally, after the system matches the corresponding voice prompt instruction from the instruction library, it converts it into an audio signal through digital signal processing and drives the connected audio device, such as headphones or speakers, to clearly play the instruction to the patient. After hearing the voice prompt, the patient can adjust their training behavior in real time according to the instruction content, thus forming a complete closed-loop control of monitoring-evaluation-feedback-adjustment.

[0067] This invention transforms complex physiological assessment results into simple and intuitive voice commands, enabling real-time, automated guidance of the rehabilitation training process. Its technical advantage lies in constructing a dynamic, interactive feedback control system that embeds professional rehabilitation guidance logic into the device, allowing patients to train safely and effectively even without the presence of medical personnel. This instant feedback mechanism continuously guides patients to maintain training intensity within their personalized optimal treatment window, avoiding both poor rehabilitation outcomes due to insufficient training and cardiac risks caused by overtraining, thereby significantly improving the safety, compliance, and ultimate effectiveness of cardiac rehabilitation training.

[0068] S6. Adaptive Adjustment of Training Parameters: Stores historical training data and adaptively adjusts training parameters.

[0069] In a specific embodiment of the present invention, the specific steps of storing historical training data and adaptively adjusting training parameters include: storing real-time electromyography intensity, real-time electrocardiogram indicators and respiratory rate during the training process to form historical training data.

[0070] It should be noted that the adaptive adjustment of training parameters in this method is a continuous optimization process based on longitudinal data analysis. During each cardiac rehabilitation training session, the system not only processes various physiological signals in real time but also systematically stores key data points—namely, real-time electromyography intensity, real-time electrocardiogram indicators, and respiratory rate—along with their precise timestamps and corresponding training stage information, in the patient's personal database. This process accumulates continuously, forming a detailed personal historical training data profile describing the evolution of the patient's physiological state throughout the training process.

[0071] Analyze the trends of physiological indicators over time in historical training data and adaptively generate adjusted training parameters.

[0072] It's important to note that within a set period, such as after each training session, the system automatically triggers a background analysis program to process the historical training data. The core task of this program is to analyze the trends of physiological indicators over time. Through statistical and modeling methods, it examines whether patients' real-time electromyography (EMG) indicators, such as heart rate, heart rate variability, and respiratory rate, show positive adaptive changes under similar real-time EMG intensity (i.e., similar external training loads). For example, the system analyzes whether, over the past month, when real-time EMG intensity is at a certain level, the patient's average heart rate has shown a downward trend, or whether the rate of heart rate recovery to resting level has accelerated. These trends provide objective quantitative evidence for assessing the improvement of the patient's cardiopulmonary function and rehabilitation progress. Based on the results of the above trend analysis, the system adaptively generates a set of adjusted training parameters. These training parameters mainly involve updating the core thresholds in the feedback control instruction generation logic. For example, if the analysis shows a significant improvement in the patient's cardiac function, the system will automatically raise the upper limit of the heart rate threshold defining the "moderate intensity" state, or increase the EMG intensity threshold that triggers the "low intensity" prompt. Conversely, if the data shows that the patient is not responding well or is experiencing a delayed recovery at the current intensity, the system will adjust these parameters accordingly to ensure safety.

[0073] The logic for generating feedback control instructions is updated based on the adjusted training parameters.

[0074] It should be noted that, finally, the system applies this set of adjusted training parameters to the logic for generating feedback control instructions, that is, it replaces the old model with the new threshold model. Thus, in the next training, the system will evaluate the patient's condition and provide guidance based on the updated criteria.

[0075] This invention establishes a training-storage-analysis-adjustment cycle, enabling intelligent and personalized evolution of rehabilitation programs. Its technical advantage lies in transforming rehabilitation training from a static, pre-set process into a dynamic system capable of self-optimization based on the patient's long-term rehabilitation progress. This method automatically identifies improvements or fluctuations in the patient's physical condition and adjusts training intensity and goals accordingly, ensuring that training remains within the most effective stimulus-adaptation range. This not only maximizes the long-term effectiveness of rehabilitation but also guarantees long-term safety through continuous personalized adjustments, achieving truly individualized and progressive rehabilitation treatment.

[0076] S7. Abnormal Risk Detection and Early Warning: Real-time monitoring of multi-parameter fusion results to detect abnormal risks and automatically issue early warnings.

[0077] In a specific embodiment of the present invention, the specific steps of real-time monitoring of multi-parameter fusion results to detect abnormal risks and automatically issue early warnings include: real-time monitoring of multi-parameter fusion results, and detecting abnormal risks based on their values ​​and rates of change.

[0078] It should be noted that the implementation of abnormal risk detection and automatic early warning functions in this method is a proactive safety assurance mechanism built on continuous monitoring. The system continuously monitors the multi-parameter fusion results output by the preceding steps in a high-frequency loop. This monitoring process is not simply recording, but includes a two-dimensional risk assessment logic. The first layer is static threshold detection based on numerical values. The system compares the instantaneous multi-parameter fusion results with a preset absolute danger threshold representing severe physiological decompensation. Once the value of the multi-parameter fusion result exceeds this threshold, a risk judgment is immediately triggered. The second layer is dynamic trend detection based on the rate of change. The system calculates the rate of change of the multi-parameter fusion results within a short time window, which can be approximated as: ,in, yes The rate of change of the multi-parameter fusion result at time t; yes The multi-parameter fusion result at time step, and It is the previous extremely short time interval The previous fusion results; the system will calculate If the rate of change exceeds the preset rate of change threshold, it means that the patient's physiological condition is deteriorating rapidly. Even if the value has not yet reached the absolute danger threshold, the system will still determine it as an abnormal risk.

[0079] In one specific embodiment of the present invention, the preset absolute danger threshold representing severe physiological decompensation can be set as a fixed value when the value of the multi-parameter fusion result reaches the critical danger range. This value is determined based on the physiological index limit value when patients experience myocardial ischemia or severe arrhythmia in clinical studies. The preset rate of change threshold is obtained by analyzing the fastest safe fluctuation rate of physiological indicators in healthy people and heart disease patients during training, and is used to capture early signs of rapid deterioration of the condition.

[0080] When an abnormal risk is detected, a risk warning command is generated that includes patient braking instructions and medical staff alarm information.

[0081] The system automatically sends medical staff alarm information from risk warning instructions to the monitoring terminal of medical staff.

[0082] It should be noted that once any of the above conditions are met, the system immediately detects an abnormal risk and instantly generates a structured risk warning instruction. This instruction is a composite information package containing two core parts. The first part is a braking instruction for the patient, typically an urgent, clear, and high-priority voice command, such as "Danger, please stop moving immediately and remain still." The second part is a medical staff alarm information, a detailed dataset containing patient identification, the specific time the warning was triggered, the multi-parameter fusion result value and its rate of change at the time of the warning, and key physiological indicators such as heart rate, electromyography intensity, and abnormal electrocardiogram snapshots from the period prior to the warning. After generating the instruction, the system automatically transmits the medical staff alarm information portion of the instruction, in encrypted form, to a pre-configured medical staff monitoring terminal, such as the hospital's central monitoring station or the mobile device of the responsible physician, in real time via its communication module and wireless network.

[0083] This invention constructs a proactive and sensitive safety barrier through real-time monitoring and a dual risk assessment mechanism. Its technical effect lies in significantly improving the safety of cardiac rehabilitation training, especially in remote or home settings. The system not only promptly stops training when the patient's physiological state enters a dangerous zone to prevent further harm, but more importantly, it can capture early signs of rapid deterioration through rate of change monitoring, achieving more proactive risk warnings. Simultaneously, by automatically sending alarm information containing key situational data to medical personnel, it ensures that professionals can intervene immediately and obtain sufficient decision-making information in emergency situations, thus providing patients with a closed-loop, comprehensive safety guarantee from immediate intervention to remote professional assistance.

[0084] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications and additions should fall within the protection scope of the present invention.

Claims

1. A method for controlling cardiac rehabilitation training based on real-time monitoring of muscle strength and electrocardiogram signals, characterized in that, Includes the following steps: S1. Real-time electromyography intensity calculation: Acquire electromyography and motion signals of the patient's limbs, and calculate the real-time electromyography intensity based on the electromyography and motion signals; S2. Respiratory rate analysis: Acquire the patient's chest electrocardiogram (ECG) and respiratory signals, obtain real-time ECG indicators based on ECG signal analysis, and obtain the respiratory rate based on respiratory signal analysis; S3. Trunk motion status determination: Acquire the patient's trunk acceleration signal and determine the trunk motion status based on the trunk acceleration signal; S4. Generation of multi-parameter fusion results: Fusion of real-time electromyography intensity, real-time electrocardiogram indicators, respiratory rate and trunk motion status to generate multi-parameter fusion results; S5. Feedback control command generation: Based on the multi-parameter fusion results, generate feedback control commands for adjusting cardiac rehabilitation training; S6. Adaptive Adjustment of Training Parameters: Stores historical training data and adaptively adjusts training parameters; S7. Abnormal Risk Detection and Early Warning: Real-time monitoring of multi-parameter fusion results to detect abnormal risks and automatically issue early warnings.

2. The cardiac rehabilitation training control method based on real-time monitoring of muscle strength and electrocardiogram signals according to claim 1, characterized in that: The specific steps for calculating real-time electromyographic intensity based on electromyographic signals and motion signals include: Electromyographic signals from the patient's limbs were collected using an electromyography (EMG) monitoring device. The patient's limb movement signals are collected using an accelerometer in an electromyography (EMG) monitoring device. By combining electromyographic (EMG) signals with motion signals, the real-time EMG intensity is calculated.

3. The cardiac rehabilitation training control method based on real-time monitoring of muscle strength and electrocardiogram signals according to claim 1, characterized in that: The specific steps for obtaining real-time electrocardiogram (ECG) indicators based on ECG signal analysis and obtaining respiratory rate based on respiratory signal analysis include: The patient's electrocardiogram (ECG) and respiratory signals were collected using an ECG acquisition device. By analyzing the waveform and rhythm characteristics of electrocardiogram (ECG) signals, real-time ECG indicators including real-time heart rate, heart rate variability, and the risk of abnormal ECGs are obtained. Analyze the waveform period of the respiratory signal to obtain the respiratory rate.

4. The cardiac rehabilitation training control method based on real-time monitoring of muscle strength and electrocardiogram signals according to claim 1, characterized in that: The specific steps for determining the trunk motion state based on trunk acceleration signals include: The patient's trunk acceleration signal is acquired using an accelerometer in the electrocardiogram acquisition device; The patient's trunk acceleration signal was filtered to obtain a preprocessed trunk acceleration signal. Based on the gravitational acceleration component in the preprocessed trunk acceleration signal, trunk motion states including sitting or lying positions are classified.

5. The cardiac rehabilitation training control method based on real-time monitoring of muscle strength and electrocardiogram signals according to claim 1, characterized in that: The specific steps for fusing real-time electromyography intensity, real-time electrocardiogram indices, respiratory rate, and trunk movement status to generate a multi-parameter fusion result include: Real-time electromyography intensity and real-time electrocardiogram indicators are timestamped to obtain a synchronized data stream; Based on trunk motion state and respiratory rate, the real-time ECG indicators in the synchronous data stream are corrected to generate corrected ECG indicators. Real-time electromyographic intensity and corrected electrocardiographic indices are input into a correlation model describing the relationship between muscle load and cardiac response to generate multi-parameter fusion results.

6. The cardiac rehabilitation training control method based on real-time monitoring of muscle strength and electrocardiogram signals according to claim 5, characterized in that: The correlation model describing the relationship between muscle load and cardiac response was obtained through the following steps: Collect physiological data and corresponding expert evaluation labels generated synchronously by multiple patients during training under professional guidance to construct multi-patient training data; By using training data from multiple patients, iterative training is performed through machine learning algorithms to obtain optimized parameters for the association model. The association model is constructed using optimized association model parameters, and its accuracy is verified using test data.

7. The cardiac rehabilitation training control method based on real-time monitoring of muscle strength and electrocardiogram signals according to claim 1, characterized in that: The specific steps for generating feedback control instructions for adjusting cardiac rehabilitation training include: The training state is determined by comparing the multi-parameter fusion result with a threshold model that defines different training intensities. Based on the training state determined by the judgment, the voice prompt command is matched from the command library that stores the mapping relationship between the command and the state; Voice prompts are output to audio devices to adjust the range of limb movements, breathing rhythm, or training posture.

8. The cardiac rehabilitation training control method based on real-time monitoring of muscle strength and electrocardiogram signals according to claim 1, characterized in that: The specific steps for storing historical training data and adaptively adjusting training parameters include: During training, real-time electromyography intensity, real-time electrocardiogram indicators, and respiratory rate are stored to form historical training data. Analyze the trends of physiological indicators over time in historical training data and adaptively generate adjusted training parameters; The logic for generating feedback control instructions is updated based on the adjusted training parameters.

9. The cardiac rehabilitation training control method based on real-time monitoring of muscle strength and electrocardiogram signals according to claim 1, characterized in that: The specific steps for real-time monitoring of multi-parameter fusion results to detect abnormal risks and automatically issue early warnings include: Real-time monitoring of multi-parameter fusion results, and detection of abnormal risks based on their values ​​and rates of change; When an abnormal risk is detected, a risk warning command is generated that includes patient braking instructions and medical staff alarm information; The system automatically sends medical staff alarm information from risk warning instructions to the monitoring terminal of medical staff.

Images