Breathing training assessment methods and devices

Abstract

This application discloses a method and apparatus for assessing breathing training. The method includes: acquiring physiological signals of a target user before, during, and after breathing training, and performing feature extraction and quantification to form feature arrays for pre-training, during-training, and post-training; obtaining normalized resting heart rate and normalized resting respiratory rate of the target user in the current state based on the target user's historical resting heart rate and historical resting respiratory rate; normalizing the feature arrays before and after breathing training based on the target user's current resting heart rate and resting respiratory rate; and performing a breathing training assessment based on the feature arrays during and after breathing training, as well as the normalized feature arrays before and after breathing training. By incorporating resting heart rate and resting respiratory rate into the three parts of the breathing training process (pre-training, during, and post-training), a comprehensive assessment of the entire breathing training cycle is achieved.

Description

Technical Field

[0001] This application relates to the field of smart home technology, and in particular to a breathing training assessment method and device. Background Technology

[0002] With the rapid development of smart home technology, smart mattresses, smart base beds, and other furniture products are gradually upgrading towards diversified functions and intelligent experiences, and their application potential in areas such as health monitoring and physiological intervention is becoming increasingly prominent. Therefore, scientifically evaluating the effects of breathing training in everyday use scenarios such as smart mattresses and smart base beds is of great significance in helping users improve their breathing habits and enhance the efficiency of health management.

[0003] Currently, breathing training assessment methods for consumer products are mainly divided into two categories: subjective assessment and objective examination. Subjective assessment judges the training effect by observing physical performance during breathing (such as the order of abdominal and chest expansion during inhalation, the extent of chest expansion, and the ratio of exhalation duration). However, the assessment results are easily influenced by subjective experience, resulting in insufficient accuracy and objectivity. Objective examination relies on the monitoring of physiological parameters for assessment. For example, wearable devices such as watches are used to monitor indicators such as heart rate, heart rate variability (HRV), and blood oxygen saturation (SpO2) before and after training. If the heart rate decreases, HRV improves, or blood oxygen saturation remains within the ideal range after training, the breathing training is considered effective.

[0004] However, existing assessment methods still have significant limitations: on the one hand, the assessment only compares physiological parameters before and after training, without incorporating the dynamic changes of physiological parameters during training (such as the real-time correlation between respiratory rhythm and heart rate fluctuations), making it difficult to fully reflect the impact of training quality on the results; on the other hand, the assessment is mostly limited to a single training session, without combining the long-term trends of users' physiological parameters and historical training data, making it impossible to achieve personalized and continuous effect tracking and feedback. Summary of the Invention

[0005] Therefore, it is necessary to provide a breathing training assessment method and apparatus to address the above-mentioned technical problems and solve at least one of the problems existing in the prior art.

[0006] Firstly, a breathing training assessment method is provided, including:

[0007] Physiological signals of the target user before, during and after breathing training are acquired, and feature extraction and quantification are performed to form feature arrays before, during and after breathing training.

[0008] Based on the target user's historical resting heart rate and historical resting respiratory rate, the normalized resting heart rate and normalized resting respiratory rate of the target user in the current state are obtained.

[0009] Based on the target user's resting heart rate and resting respiratory rate in the current state, the feature arrays before and after the breathing training are normalized respectively.

[0010] Breathing training is evaluated based on the feature arrays during breathing training, as well as the normalized feature arrays before and after breathing training.

[0011] In one possible implementation, the physiological signals include heart rate signals and respiratory signals. The step of acquiring the target user's physiological signals before, during, and after breathing training, and performing feature extraction and quantization to form feature arrays for pre-training, during-training, and post-training periods, includes:

[0012] After reconstructing and correcting the original physiological sensor signals, the heart rate signal is extracted, and a heart rate feature array is calculated based on the heart rate signal.

[0013] After optimizing the original physiological sensing signal, the respiratory signal envelope is extracted. The respiratory signal is obtained based on the respiratory signal envelope and its polarity is adaptively adjusted. The respiratory feature array is calculated based on the adjusted respiratory signal.

[0014] In one possible implementation, the step of extracting the heart rate signal after signal reconstruction and correction of the original physiological sensor signal, and calculating the heart rate feature array based on the heart rate signal, includes:

[0015] The original physiological sensing signals are subjected to differential processing;

[0016] Perform bandpass filtering on the differentially processed signal;

[0017] Phase compensation is performed on the signal after bandpass filtering by full-pass filtering;

[0018] Polarity correction is performed on the phase-compensated signal;

[0019] The heart rate feature array is calculated based on the feature points of the corrected signal.

[0020] In one possible implementation, the polarity correction of the phase-compensated signal includes:

[0021] Extract a first signal segment with a preset window time length from the phase-compensated signal and calculate the signal quality of the first signal segment;

[0022] If the signal quality is greater than a preset quality threshold, determine the first correlation coefficient array between the first signal segment and the preset signal template;

[0023] The phase-compensated signal is inverted to extract a second signal segment of the preset window time length, and a second correlation coefficient array between the second signal segment and the preset signal template is determined.

[0024] If the mean of the first correlation coefficient array is less than the mean of the second correlation coefficient array, then the polarity is negative, and the phase-compensated signal is inverted.

[0025] In one possible implementation, before determining the first correlation coefficient array between the first signal segment and the preset signal template, the following steps are included:

[0026] Construct the initial signal template;

[0027] The initial signal template is normalized to obtain the preset signal template;

[0028] If the signal quality persists for a preset time longer than the preset quality threshold, the update time of the signal template is calculated based on the average heart rate value within the preset time window before the current moment.

[0029] Update the preset signal template according to the specified update time length.

[0030] In one possible implementation, the calculation of the heart rate feature array based on feature points of the corrected signal includes:

[0031] Based on the corrected signal, all heart rate feature points were determined;

[0032] Based on the heart rate feature points, a heart rate interval sequence is obtained;

[0033] Based on the heart rate interval sequence, the average heart rate, the standard deviation of the heart rate interval, and the trend feature of the heart rate interval are obtained. The trend feature of the heart rate interval is the slope of the linear fit of the heart rate sequence, and the heart rate sequence is obtained based on the heart rate interval sequence.

[0034] In one possible implementation, the step of optimizing the original physiological sensing signal to extract the respiratory signal envelope, obtaining the respiratory signal based on the respiratory signal envelope and adaptively adjusting its polarity, and calculating the respiratory feature array based on the adjusted respiratory signal includes:

[0035] The original physiological sensing signal is filtered.

[0036] Phase compensation is performed on the filtered signal;

[0037] Based on the phase-compensated signal, the respiratory signal envelope is extracted;

[0038] A respiratory signal is obtained based on the respiratory signal envelope, and the polarity of the respiratory signal is adjusted.

[0039] Based on the characteristic points of the polarity-adjusted respiratory signal, a respiratory feature array is calculated.

[0040] In one possible implementation, the feature points include the inspiratory start point and the inspiratory end point. The feature points, based on the polarity-adjusted respiratory signal, are used to calculate a respiratory feature array, including:

[0041] All respiratory feature points were determined based on the polarity-adjusted respiratory signal.

[0042] Based on the respiratory feature points, the inter-respiratory phase sequence is obtained;

[0043] Based on the respiratory interval sequence, the mean respiratory rate, standard deviation of respiratory intensity, mean ratio of inspiratory duration to expiratory duration, standard deviation of ratio of inspiratory duration to expiratory duration, and respiratory interval variation trend characteristics are calculated. The respiratory interval variation trend characteristics are the slope obtained by linearly fitting the respiratory rate sequence, which is based on the respiratory interval sequence.

[0044] In one possible implementation, the breathing training evaluation based on the feature array during breathing training and the normalized feature arrays before and after breathing training includes:

[0045] Based on the normalized feature arrays before and after breathing training, a score for the breathing training result is obtained.

[0046] A score for the breathing training process is obtained based on the feature array in the breathing training.

[0047] Based on the breathing training result score, breathing training process score, historical breathing training result score, and historical breathing training process score, a breathing training improvement score is calculated.

[0048] Based on the breathing training result score, breathing training process score, and breathing training improvement score, a breathing training score is calculated.

[0049] Based on the breathing training score, the breathing training assessment result is determined.

[0050] Secondly, a breathing training assessment device is provided, comprising:

[0051] The feature array generation unit is used to acquire the physiological signals of the target user before, during and after breathing training, and to perform feature extraction and quantification to form feature arrays before, during and after breathing training.

[0052] The resting data acquisition unit is used to obtain the normalized resting heart rate and normalized resting respiratory rate of the target user in the current state based on the target user's historical resting heart rate and historical resting respiratory rate.

[0053] The normalization processing unit is used to normalize the feature array before breathing training and the feature array after breathing training based on the resting heart rate and resting respiratory rate of the target user in the current state.

[0054] The breathing training assessment unit is used to assess breathing training based on the feature array during breathing training and the normalized feature arrays before and after breathing training.

[0055] The above-mentioned breathing training assessment method and device include the following steps: acquiring physiological signals of the target user before, during, and after breathing training, and performing feature extraction and quantification to form a feature array before, during, and after breathing training; obtaining the normalized resting heart rate and normalized resting respiratory rate of the target user in the current state based on the target user's historical resting heart rate and historical resting respiratory rate; normalizing the feature array before and after breathing training based on the target user's current resting heart rate and resting respiratory rate; and performing a breathing training assessment based on the feature array during and after breathing training, as well as the normalized feature arrays before and after breathing training. In this embodiment, physiological signals of the target user before, during, and after breathing training are acquired, and corresponding feature arrays are formed through feature processing. Based on these arrays, breathing training evaluation is performed, achieving full-cycle coverage evaluation of breathing training. This overcomes the limitations of existing technologies that only evaluate before and after training or a single training session. It not only reflects changes in physiological state before and after training but also incorporates dynamic features during the training process, making the evaluation more comprehensive and complete in reflecting the actual effect of breathing training. Furthermore, by introducing resting heart rate and resting respiratory rate, the reference values ​​for evaluating the effect of breathing training are made more reasonable, providing more reliable technical support for the scientific evaluation of the effectiveness of breathing training. Attached Figure Description

[0056] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0057] Figure 1 This is a schematic diagram of an application environment for a breathing training assessment method according to one embodiment of this application;

[0058] Figure 2 This is a flowchart illustrating a breathing training assessment method in one embodiment of this application;

[0059] Figure 3 This is a schematic diagram of an application environment in which a user lies in a specified posture in a specified area of ​​the device (semi-sitting position) according to one embodiment of this application.

[0060] Figure 4 This is a schematic diagram of the original physiological sensor signals in the three stages of breathing training before, during and after the breathing training in one embodiment of this application.

[0061] Figure 5 This is a schematic diagram of a central impact signal (BCG) template in one embodiment of this application;

[0062] Figure 6 This is an example diagram of the selection of J points before, during, and after breathing training in one embodiment of this application;

[0063] Figure 7 This is an example diagram of respiratory feature points before, during, and after breathing training in one embodiment of this application;

[0064] Figure 8 This is a schematic diagram of the structure of a breathing training assessment device in one embodiment of this application;

[0065] Figure 9 This is a schematic diagram of a computer device according to one embodiment of this application. Detailed Implementation

[0066] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0067] The breathing training assessment method provided in this embodiment can be applied to, for example... Figure 1In the application environment, its architecture includes a server S1, a client S2, and a device S3, which can communicate with each other. The server S1 is used for user information management, including user information storage and information distribution. This information includes, but is not limited to, the user's personal information, historical physiological parameters, and breathing training history. The historical physiological parameters include resting heart rate and resting respiratory rate, which refer to the average heart rate and average respiratory rate at rest. The resting state refers to the time segment during which the user remains motionless (e.g., the time during which the amplitude of body movement detected by the device sensor is less than a preset threshold) for more than a first preset time threshold TH1, and has not entered sleep. TH1 is, for example, 5 minutes. The resting heart rate and resting respiratory rate are updated once daily at a fixed time. The breathing training history includes: changes in heart rate before and after historical breathing training, the duration of inhalation and exhalation after historical breathing training, and the ratio of inhalation to exhalation duration.

[0068] Client S2 is used for launching the breathing training function, guiding breathing training, displaying breathing training results, receiving breathing training records from device S3 and uploading them to server S1, etc.

[0069] The device-side S3 is used to acquire user physiological parameters and evaluate training effects. It mainly consists of four parts: signal acquisition module, signal quality judgment module, feature extraction module, and breathing training effect evaluation module.

[0070] It should be noted that the client includes, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices, which may have pre-installed applications or applets for breathing training. The server can be implemented using a standalone server or a server cluster consisting of multiple servers.

[0071] In one embodiment, such as Figure 2 As shown, a breathing training assessment method is provided, which is then applied to... Figure 1 Taking device S3 as an example, the explanation includes the following steps:

[0072] In step S110, physiological signals of the target user before, during and after breathing training are acquired, and feature extraction and quantification are performed to form feature arrays before, during, and after breathing training.

[0073] Optionally, the device can be a smart bed, smart sofa, or other smart furniture for user rest, with at least one signal acquisition device installed in a designated area. When a breathing training request is received from the client, the device can be adjusted to the breathing training posture. For example, taking a smart bed as an example, this smart bed has headboard and footboard tilt adjustment functions. Upon a breathing training request, the headboard and footboard can be adjusted to a specified tilt angle so that the user can lie on the device in a specified posture. For example, if the specified posture is a semi-reclining position, the upper body bed frame can be raised at a certain angle, and the leg bed frame can also be raised at a certain angle, with the angle being relatively smaller than the angle raised by the upper body bed frame, and both the upper body bed frame and the leg bed frame are raised in the same direction. Figure 3 As shown, the device allows the user to be positioned in a designated area in a specified posture, thereby collecting the user's physiological signals before, during, and after breathing training through a pre-set signal acquisition device within the designated area. If the device does not have the function of adjusting the head and foot of the bed, it can guide the user to lie supine with knees slightly bent and hands placed on the abdomen on the device.

[0074] It should be noted that physiological signals can include heart rate signals, respiratory signals, and other physiological signals. When the target user is in a specified posture within a designated area of ​​the device, a pre-set physiological acquisition device within that area can collect raw physiological sensor signals in real time before, during, and after breathing training. For example, taking the period before breathing training as an example, the physiological acquisition device could be a piezoelectric sensor, which could collect the piezoelectric signal generated before the training. Then, it can be divided into two branches, each processing the piezoelectric signal to obtain the heart rate signal and the respiratory signal. Finally, heart rate features are extracted based on the heart rate signal, and respiratory features are extracted based on the respiratory signal. The heart rate features and respiratory features are combined in a preset order to form a feature array for the period before breathing training. It is understood that this signal acquisition device can also select other sensors, such as pressure sensors, depending on the actual situation. The collected raw physiological sensor signals can be ballistocardiogram (BCG), photoplethysmogram (PPG), or electrocardiogram (ECG).

[0075] Furthermore, the feature arrays during and after breathing training can also be obtained using the methods described above. Specifically, during guided breathing training, the acquired raw physiological sensor signals are processed to form a feature array during breathing training. The composition of this feature array is consistent with that before breathing training, except that the time period is from the start of the APP's inhalation and exhalation guidance to its end. After breathing training, physiological signals are acquired and processed to form a feature array after breathing training. The structure of this post-training feature array is consistent with that before training (only the time period differs), except that the time period is from the end of the APP's inhalation and exhalation guidance until the user is notified that the breathing training is complete. During this time period, the APP will prompt the user to maintain quiet and steady breathing.

[0076] The heart rate feature may include the average heart rate, the standard deviation of the heart rate interval, and the trend feature of the heart rate interval. The trend feature of the heart rate interval is the slope of the linear fit of the heart rate sequence, which is obtained based on the heart rate interval sequence.

[0077] Among them, the average respiratory rate, the standard deviation of respiratory intensity, the average ratio of inspiratory duration to expiratory duration, the standard deviation of the ratio of inspiratory duration to expiratory duration, and the trend characteristics of the respiratory interval change, wherein the trend characteristics of the respiratory interval change are the slope obtained by linear fitting of the respiratory rate sequence, and the respiratory rate sequence is obtained based on the respiratory interval sequence.

[0078] In step S120, based on the target user's historical resting heart rate and historical resting respiratory rate, the normalized resting heart rate and normalized resting respiratory rate of the target user in the current state are obtained.

[0079] The resting heart rate and resting respiratory rate refer to the average heart rate and average respiratory rate in a resting state; the resting state refers to the time segment during which the user remains inactive (body movement amplitude < preset threshold) for more than a first preset time threshold TH1 (e.g., 5 minutes) and has not entered sleep. The resting heart rate and resting respiratory rate are updated once a day at a fixed time (e.g., after waking up in the morning when there is no activity).

[0080] Optionally, historical resting heart rate and historical resting respiratory rate can be low-pass filtered. Long-term stable trends in historical resting heart rate / respiratory rate (such as the user's normal resting level) are considered low-frequency signals and will be retained. Short-term fluctuations (high-frequency signals) caused by temporary factors (such as slightly higher values ​​due to mild stress during a measurement) will be filtered out. For example, if a user's resting heart rate over the past 7 days is 70, 75, 72, 80, 73, 71, and 72 beats per minute, the "80" might be a temporary fluctuation (high-frequency interference). After low-pass filtering, it will be smoothed to a value closer to the normal range of around 72, rather than directly taking the average (which might be higher due to the influence of the "80"), thus more closely reflecting the current true resting state. The final result is a resting heart rate and resting respiratory rate that more closely reflects the user's current physiological state, providing a reliable benchmark for subsequent assessments.

[0081] It should be noted that if this is the first time using the device, the historical resting heart rate and historical resting respiratory rate can be based on the population baseline values ​​(e.g., resting heart rate of 60-80 beats / min).

[0082] In step S130, based on the resting heart rate and resting respiratory rate of the target user in the current state, the feature array before breathing training and the feature array after breathing training are normalized respectively.

[0083] It should be noted that heart rate-related features, such as average heart rate and heart rate trend features, can be directly divided by the current resting heart rate to convert the "absolute heart rate value" into a "proportion relative to one's own resting state." For example, a user with a resting heart rate of 60 beats / min will have an average heart rate of 54 beats / min after training, while a user with a resting heart rate of 80 beats / min will have an average heart rate of 72 beats / min after training. Although the original values ​​are different, they are both normalized to 0.9, which can fairly reflect the same improvement of "a 10% decrease in heart rate." On the other hand, dividing the standard deviation of heart rate intervals by "60 × current resting heart rate" utilizes the physiological relationship that "heart rate is inversely proportional to intervals" (the faster the heart rate, the shorter the heartbeat interval) to standardize interval fluctuations and avoid misjudgment of the intensity of fluctuations caused by different baseline heart rates.

[0084] Where 60 is the conversion factor between minutes and seconds, the current resting heart rate is in beats / minute, and 60 ÷ resting heart rate gives the average heartbeat interval (seconds).

[0085] For respiratory-related characteristics, dividing the average respiratory rate and the trend of respiratory rate changes by the current resting respiratory rate can also eliminate individual differences in baseline respiratory rate. For example, a user with a resting respiratory rate of 16 breaths / min who drops to 12 breaths / min after training, and a user with a resting respiratory rate of 20 breaths / min who drops to 15 breaths / min after training, both have a normalized value of 0.75, clearly demonstrating the consistent effect of "25% decrease in respiratory rate". This processing transforms the original characteristics, which are originally affected by individual innate physiological differences (such as differences in baseline heart rate due to age and physical condition) and daily fluctuations (such as a slight increase in resting heart rate due to fatigue), into standardized indicators that can be compared horizontally among different users and tracked vertically for the same user at different times. This ensures that when calculating the changes in characteristics before and after training, it can accurately reflect the actual impact of training on the user, rather than being interfered with by individual baseline differences, providing solid data support for the scientific and personalized evaluation of the final respiratory training effect.

[0086] In step S140, a breathing training evaluation is performed based on the feature array during breathing training and the normalized feature arrays before and after breathing training.

[0087] Optionally, the result score is obtained based on the normalized feature array before and after breathing training, the process score is obtained by combining the feature array during training, and the improvement score is calculated by fusing the two and the corresponding historical scores. Finally, the training score is calculated based on the result score, the process score and the improvement score, and then the breathing training evaluation result is determined.

[0088] Specifically, the breathing training results are scored by first using the normalized feature arrays before and after breathing training. Specifically, four core change features are calculated: heart rate change before and after breathing training, standard deviation of heart rate interval change, respiratory rate change, and ratio of inhalation to exhalation duration change. Then, these four features are weighted and summed to obtain the breathing training result score.

[0089] Then, based on the feature array in the breathing training, the score of the breathing training process is calculated. This requires first statistically analyzing two key process indicators: one is the absolute difference between the actual number of breaths during training and the guided number of breaths; the other is the number of cycles in which the deviation between the ratio of inhalation duration to exhalation duration and the guided ratio exceeds a preset threshold, such as 30%. Then, based on empirically set weights, these two indicators are weighted and summed.

[0090] Finally, historical breathing training scores can be obtained, which may include the score for the result of the previous breathing training session and the score for the process of the previous breathing training session. A comprehensive breathing training score is calculated using the scores for the result of the previous breathing training session, the process of the previous breathing training session, the result of the current breathing training session, and the process of the current breathing training session. If the comprehensive breathing training score... ≥80 indicates an excellent training evaluation result; otherwise, if the overall breathing training score is 80 or higher... A score of ≥60 indicates a good training assessment result; otherwise, a comprehensive breathing training score of 60 or higher indicates a poor result. If the score is ≥40, the training evaluation result is considered average; otherwise, the training evaluation result is considered poor.

[0091] In addition, the breathing training status can be obtained using feature arrays before, during, and after breathing training, so that results can be displayed and status prompts can be provided on the target user's client. Specifically, the quality of physiological signals during the training process can be detected. If the percentage of durations with a signal-to-noise ratio (SNR) greater than a preset threshold TH3 is less than a preset threshold TH11 (e.g., 70%), the signal quality is considered poor. In this case, the training evaluation results will not be displayed, and a message will be displayed indicating poor signal quality during the training process, requesting the user to remain as quiet as possible during breathing training. If the signal quality meets the requirements, the number of abnormal breathing intervals (a breathing interval refers to the time interval between the start points of two adjacent breaths) can be further detected. If the number of abnormal breathing intervals is... If the respiratory interval exceeds the preset threshold TH12 (e.g., 10), the breathing interval during training is considered abnormal. The training assessment results will be displayed, suggesting that breathing should be as even and consistent in intensity as possible during training. If the respiratory interval is normal, the system will continue to check for abnormal breathing rhythms during training. Specifically, it will check if any of the following four conditions are met: the absolute value of the difference between the number of breaths and the guided breathing count exceeds the preset threshold TH13 (e.g., 3); the absolute value of the difference between the average ratio of inspiratory to expiratory duration and the average ratio of guided inspiratory to expiratory duration exceeds the preset threshold TH14 (e.g., 0.3); the standard deviation of the ratio of inspiratory to expiratory duration exceeds the preset threshold TH15 (e.g., 0.2); or the standard deviation of breathing intensity exceeds the preset threshold TH16 (e.g., 0.05). If any of these conditions are met, the breathing rhythm is considered abnormal, and the training assessment results will be displayed, suggesting that the breathing rhythm be controlled according to the guided breathing. If none of the above abnormal conditions are met, the breathing training is considered normal, and only the training assessment results will be displayed.

[0092] This application provides a breathing training assessment method, comprising: acquiring physiological signals of a target user before, during, and after breathing training, and performing feature extraction and quantification to form a feature array before breathing training, a feature array during breathing training, and a feature array after breathing training; obtaining the normalized resting heart rate and normalized resting respiratory rate of the target user in the current state based on the target user's historical resting heart rate and historical resting respiratory rate; normalizing the feature array before and after breathing training based on the target user's current resting heart rate and resting respiratory rate; and performing a breathing training assessment based on the feature array during breathing training and the normalized feature arrays before and after breathing training. In this embodiment, physiological signals of the target user before, during, and after breathing training are acquired, and corresponding feature arrays are formed through feature processing. Based on these arrays, breathing training evaluation is performed, achieving full-cycle coverage evaluation of breathing training. This overcomes the limitations of existing technologies that only evaluate before and after training or a single training session. It not only reflects changes in physiological state before and after training but also incorporates dynamic features during the training process, making the evaluation more comprehensive and complete in reflecting the actual effect of breathing training. Furthermore, by introducing resting heart rate and resting respiratory rate, the reference values ​​for evaluating the effect of breathing training are made more reasonable, providing more reliable technical support for the scientific evaluation of the effectiveness of breathing training.

[0093] In one embodiment of this application, the physiological signals include heart rate signals and respiratory signals. The step of acquiring the physiological signals of the target user before, during, and after breathing training, and performing feature extraction and quantization to form feature arrays for pre-training, during-training, and post-training periods, includes:

[0094] After reconstructing and correcting the original physiological sensor signals, the heart rate signal is extracted, and a heart rate feature array is calculated based on the heart rate signal.

[0095] After optimizing the original physiological sensing signal, the respiratory signal envelope is extracted. The respiratory signal is obtained based on the respiratory signal envelope and its polarity is adaptively adjusted. The respiratory feature array is calculated based on the adjusted respiratory signal.

[0096] It should be noted that the raw physiological sensing signal refers to the physiological sensing signal collected by a signal acquisition device set in the designated area of ​​the device when the target user is in a specified posture on the designated area before, during, and after breathing training. For example, if the signal acquisition device is a piezoelectric sensor, then the physiological sensing signal can be a piezoelectric signal. It is understandable that ballistocardiogram (BCG), photoplethysmography (PPG), and electrocardiogram (ECG) signals can also be used as raw physiological sensing signals, all of which can separate heart rate and respiratory signals.

[0097] Optionally, when the target user is in a designated area on the device in a designated posture, physiological sensor signals before, during, and after breathing training are collected by a signal acquisition device set in the designated area, and heart rate and respiratory signals are processed separately through two parallel paths:

[0098] Heart rate signal path: First, the original physiological sensor signal is reconstructed and corrected (e.g., filtering and denoising, phase compensation, polarity correction) to eliminate interference components (e.g., motion artifacts, environmental noise) and correct signal deviations, laying the foundation for subsequent extraction of pure heart rate signal; then, the heart rate signal is directly extracted from the processed signal (focusing on heartbeat-related characteristic fluctuations); finally, based on the extracted heart rate signal, key indicators such as average heart rate, standard deviation of heart rate interval, and heart rate change trend are calculated to form a structured heart rate feature array.

[0099] Respiratory signal path: First, the original physiological sensor signal is optimized (e.g., filtered to remove high-frequency interference); then, by extracting the respiratory signal envelope, low-frequency fluctuation information related to breathing is extracted from the mixed signal (filtering out non-respiratory components); next, after obtaining the respiratory signal based on the signal envelope, adaptive polarity adjustment is performed (unifying the signal polarity of the inspiratory and expiratory phases to ensure waveform consistency); finally, based on the adjusted respiratory signal, features such as average respiratory rate, ratio of inspiratory to expiratory duration, and changes in respiratory intensity are extracted to form a respiratory feature array.

[0100] It should be noted that the breathing feature arrays before, during, and after breathing training are generated in the same way, and can all be obtained through the above method.

[0101] In one embodiment of this application, the step of extracting the heart rate signal after signal reconstruction and correction of the original physiological sensor signal, and calculating the heart rate feature array based on the heart rate signal, includes:

[0102] The original physiological sensing signals are subjected to differential processing;

[0103] Perform bandpass filtering on the differentially processed signal;

[0104] Phase compensation is performed on the signal after bandpass filtering by full-pass filtering;

[0105] Polarity correction is performed on the phase-compensated signal;

[0106] The heart rate feature array is calculated based on the feature points of the corrected signal.

[0107] Optionally, the original physiological sensor signal f(t) can be differentially processed, such as by first-order difference, to obtain the differential signal f1(t). The original physiological sensor signal f(t) includes, but is not limited to, the ballistocardiogram (BCG). Taking the BCG signal as an example, the original physiological sensor signals for the entire breathing training process can be obtained as follows: Figure 4 As shown, it consists of three parts: before breathing training, during breathing training, and after breathing training. The raw physiological sensor signals of each of these three parts are processed using first-order difference, as shown in the following formula:

[0108] f1(t) = f(t+1) - f(t);

[0109] Where t represents the current time point.

[0110] The differential signal f1(t) is subjected to bandpass filtering to remove low-frequency and high-frequency noise, resulting in f2(t). The bandpass filter cutoff frequency is [0.8Hz, 20Hz], and the filter is an Infinite Impulse Response (IIR) filter.

[0111] While using an IIR filter can significantly reduce the computational load, it introduces nonlinear phase distortion. To compensate for the impact of this nonlinear phase distortion on the waveform, an all-pass filter can be used to reduce the nonlinearity of the filter bank's phase response, performing phase compensation on f2(t) to obtain f3(t). The polarity of the f3(t) waveform is then detected; if the polarity is negative, it is inverted to obtain f4(t). If it remains positive, no further processing is performed. Finally, feature points are selected from the processed signal to obtain all heart rate features, which are then assembled into a heart rate feature array.

[0112] In one embodiment of this application, the polarity correction of the phase-compensated signal includes:

[0113] Extract a first signal segment with a preset window time length from the phase-compensated signal and calculate the signal quality of the first signal segment;

[0114] If the signal quality is greater than a preset quality threshold, determine the first correlation coefficient array between the first signal segment and the preset signal template;

[0115] The phase-compensated signal is inverted to extract a second signal segment of the preset window time length, and a second correlation coefficient array between the second signal segment and the preset signal template is determined.

[0116] If the mean of the first correlation coefficient array is less than the mean of the second correlation coefficient array, then the polarity is negative, and the phase-compensated signal is inverted.

[0117] Optionally, from the phase-compensated signal f3(t), a signal segment with a preset window time length TH2, such as 5 seconds, is extracted, and the signal quality (SNR) of this signal segment is calculated. If the signal quality (SNR) is greater than a preset threshold TH3, such as 2, the signal quality is considered to meet the requirements. At this time, the first correlation coefficient array between f3(t) and a preset signal template, such as a cardiac impact signal template, within the preset window time TH2 can be calculated. Then, f3(t) is inverted to obtain f4(t), i.e., f4(t) = -f3(t), and the correlation coefficient array between f4(t) and the preset signal template, such as a cardiac impact signal template, within the preset window time TH2 is calculated. Finally, the average value of the correlation coefficient array of f3(t) is compared with the average value of the correlation coefficient array of f4(t). If the former is greater than the latter, the waveform polarity is positive; otherwise, the polarity is negative.

[0118] In one embodiment of this application, calculating the signal quality of the first signal segment includes:

[0119] Based on the original physiological sensor signals, high-frequency signals and heart rate signals are extracted;

[0120] Based on the high-frequency signal, the energy of the high-frequency signal within the preset window time is determined;

[0121] Based on the original physiological sensor signals, the heart rate signal energy within the preset window time is determined;

[0122] The signal quality is determined based on the ratio between the heart rate signal energy and the high-frequency signal energy.

[0123] Optionally, the original physiological sensing signal f(t) is subjected to IIR high-pass extraction to extract high-frequency signals, with a filter cutoff frequency of 8Hz, to obtain signal f5(t);

[0124] Calculate the signal energy f5(t) within the preset window time TH2, i.e., the high-frequency signal energy. The calculation method can be:

[0125] ;

[0126] Where i is the i-th sampling point within the preset window time TH2, and n is the number of sampling points within the preset window time. For example, if the preset window time TH2 is 5s and the signal sampling rate is 200Hz, then n is 1000.

[0127] The original physiological sensing signal f(t) is bandpass filtered with an IIR filter and the cutoff frequency of the bandpass filter is [0.8Hz, 8Hz] to obtain the heart rate frequency band signal f6(t).

[0128] Calculate the signal energy within the preset window time f6(t), which is the heart rate signal energy. The calculation method is as follows:

[0129] ;

[0130] The signal quality of this segment is obtained by calculating the ratio of heart rate signal energy to high-frequency signal energy, as shown below:

[0131] .

[0132] In one embodiment of this application, before determining the first correlation coefficient array between the first signal segment and the preset signal template, the following steps are included:

[0133] Construct the initial signal template;

[0134] The initial signal template is normalized to obtain the preset signal template;

[0135] If the signal quality persists for a preset time longer than the preset quality threshold, the update time of the signal template is calculated based on the average heart rate value within the preset time window before the current moment.

[0136] Update the preset signal template according to the specified update time length.

[0137] Optionally, first, initialize the cardiac impact signal template, and the signal template duration is... If the sampling rate is 200Hz, then the number of template sampling points is 200.

[0138] Then, a typical cardiac impact signal template is constructed, which is composed of the superposition of I, J, and K, as shown below. Figure 5 As shown. The three characteristic waves I, J, and K (I wave corresponds to the early stage of ventricular systole, J wave corresponds to the rapid ejection phase of the ventricle, and K wave corresponds to the early stage of ventricular diastole) can appear sequentially and superimpose in chronological order. These I, J, and K waveforms are all composed of Gaussian-modulated sinusoidal pulses. The mathematical expression of the Gaussian-modulated sinusoidal pulse is as follows:

[0139] ;

[0140] Where A is the signal amplitude, I wave is set to 0.7, J wave to 1, and K wave to -0.5; e is the natural constant. The pulse center time is set to 0. The initial phase is set to 0. Using time vectors, the I-wave time vector advances by 0.15 seconds overall, the J-wave by 0.25 seconds, and the K-wave by 0.35 seconds overall. The center frequency of the cosine signal is in Hz. The I wave is set to 10Hz, the J wave to 15Hz, and the K wave to 12Hz. For bandwidth parameters, I wave 0.7, J wave 0.5, K-wave It is 0.6; The standard deviation of the Gaussian envelope is given by the bandwidth parameter. Decide, The calculation method is as follows:

[0141] ;

[0142] in, The initial phase is set to 0. The generated I, J, and K are superimposed to obtain the BCG signal template. Then, the BCG signal template is normalized by dividing it by the maximum absolute value to obtain the final BCG signal template, as shown in the figure.

[0143] If the signal quality SNR meets the requirements for a duration exceeding the preset threshold TH4, for example, 5 minutes, the average heart rate value within the preset time threshold TH4 time window prior to the current moment is used. Calculate the time length of the BCG signal template , If the signal template time length is updated, the signal template is also updated. By adding waveform polarity detection based on a variable template, the accuracy of feature point selection is improved to some extent.

[0144] Furthermore, taking the first correlation coefficient array as an example, the calculation process of the correlation coefficient array will be explained. First, we can start from the first sampling point of the f3(t) signal segment with a window time length of TH2. Set to the time point corresponding to the first sampling point, and extract [ For the signal segment f3(t) at time [t], calculate the correlation coefficient between this signal segment and the cardiac impact signal template, and add this correlation coefficient to the first correlation coefficient array. The step size is TH5, for example, 0.25s. Control the window movement interval and update [the relevant information]. A new time window is obtained. The signal segment within the new time window is then used to calculate the correlation coefficient again with the cardiac impact signal template. The calculated correlation coefficient is added to the first correlation coefficient array. This process is repeated until the end of the window. + The calculation stops when the last sampling point of the signal segment is exceeded.

[0145] In one embodiment of this application, the step of calculating the heart rate feature array based on feature points of the corrected signal includes:

[0146] Based on the corrected signal, all heart rate feature points were determined;

[0147] Based on the heart rate feature points, a heart rate interval sequence is obtained;

[0148] Based on the heart rate interval sequence, the average heart rate, the standard deviation of the heart rate interval, and the trend feature of the heart rate interval are obtained. The trend feature of the heart rate interval is the slope of the linear fit of the heart rate sequence, and the heart rate sequence is obtained based on the heart rate interval sequence.

[0149] Optionally, the heart rate feature point refers to the J point, which is a key feature point on the J wave. An example of J point selection is shown below. Figure 6 As shown, the selection process for the J-wave is as follows: Minimum point detection is performed on the corrected signal. If the detected minimum value is less than the preset threshold TH6 and the time interval between it and the previous valid minimum value is reasonable, it is marked as a valid minimum value, and the relevant buffer and TH6 are updated. Then, maximum value detection is performed. If the detected maximum value is greater than the preset threshold TH7, and a matching valid minimum value exists before this maximum value (the minimum value flag is 1), then this point is considered the J-point. The maximum value buffer and TH7 are updated, and the valid minimum value flag is set to 0.

[0150] It should be noted that the initial TH6 and initial TH7 can be obtained from the pre-collected data. The pre-collected data refers to the data collected in advance from different groups of people lying quietly in bed. The peak and trough amplitudes of the positive polarity heart rate signal are extracted, and the lower quartile of the peak is multiplied by a certain coefficient to obtain the initial TH7. The upper quartile of the trough is multiplied by a certain coefficient to obtain the initial TH6.

[0151] Understandably, the heart rate feature array can include the average heart rate, the standard deviation of the heart rate interval, and the trend characteristics of the heart rate interval. Once the heart rate feature point, i.e., point J, is obtained, the heart rate feature array can be obtained as follows: First, calculate the time interval between adjacent J points, i.e., the heart rate interval JJI (J-point Interval), which can be calculated using the following formula:

[0152] ;

[0153] in It refers to the time corresponding to the i-th J point.

[0154] All obtained time intervals are sorted into a heart rate interval sequence (JJI) according to time or other order. The JJI sequence is then low-pass filtered to obtain the JJI baseline sequence. If any JJI in the sequence deviates from its corresponding baseline by a preset threshold (e.g., 20%), it is considered abnormal and can be directly removed to avoid interference with the analysis. After removing abnormal JJIs, if the sequence length is insufficient (e.g., less than 5 valid points), it can be filled with nearby valid data or a signal deficiency warning can be issued. If the original JJI sequence is discrete and unequally spaced, linear interpolation (or other methods) can be used to generate an equidistant time sequence (e.g., one point every 0.25 seconds) to obtain the final JJI sequence, adapting the data format to subsequent analysis.

[0155] Among them, average heart rate It can be calculated using the following formula:

[0156] ;

[0157] Where k is the number of JJI;

[0158] The standard deviation of heart rate intervals can be calculated using the following formula:

[0159] ,in The mean of the JJI sequence;

[0160] Heart rate trend characteristics The slope of the first-order linear fit to the heart rate sequence is given by the following formula: , The heart rate sequence corresponds one-to-one with the JJI sequence, and the heart rate value is 60 / JJI, which is the heart rate cycle per minute.

[0161] In one embodiment of this application, the step of extracting the respiratory signal envelope after optimizing the original physiological sensing signal, obtaining the respiratory signal based on the respiratory signal envelope and adaptively adjusting its polarity, and calculating the respiratory feature array based on the adjusted respiratory signal includes:

[0162] The original physiological sensing signal is filtered.

[0163] Phase compensation is performed on the filtered signal;

[0164] Based on the phase-compensated signal, the respiratory signal envelope is extracted;

[0165] A respiratory signal is obtained based on the respiratory signal envelope, and the polarity of the respiratory signal is adjusted.

[0166] Based on the characteristic points of the polarity-adjusted respiratory signal, a respiratory feature array is calculated.

[0167] Optionally, the original physiological sensing signal f(t) is processed by IIR high-pass filtering (cutoff frequency 0.1Hz) to obtain f7(t) to filter out signal baseline drift. Phase compensation is then performed on f7(t) using an all-pass filter to obtain phase-corrected f8(t). Then, based on f8(t) and the number of sampling points N corresponding to a preset window time length TH8 (e.g., 1.5s), the respiratory signal envelope (which reflects the overall trend of signal amplitude change during respiration, increasing during inspiration and decreasing during expiration) is extracted by calculating the root mean square, resulting in f9(t). TH8 can be 1.5s. The calculation method for f9(t) is as follows:

[0168] ;

[0169] Where N is the number of sampling points corresponding to the preset window time length TH8, and i represents the i-th sampling point.

[0170] Next, bandpass filtering is applied to f9(t) to remove the signal baseline and high-frequency noise, yielding f10(t). Then, f10(t) is processed by full-pass filtering to obtain f11(t). Subsequently, the polarity of signal f3(t) is acquired. If its polarity is negative, f11(t) is inverted to obtain f12(t), i.e., f12(t) = -f11(t). If the polarity of f3(t) is positive, no operation is performed, and f12(t) = f11(t). Finally, respiratory feature points are selected, and a respiratory feature array is calculated for feature points such as the start and end of inspiration, completing the entire process from raw signal to respiratory signal extraction and feature calculation. A combination of IIR and full-pass filters is used in heart rate and respiratory signal processing, reducing computational complexity and the impact of nonlinear phase response while ensuring real-time calculation capability.

[0171] In one embodiment of this application, the feature points include the inspiratory start point and the inspiratory end point. The feature points, calculated based on the polarity-adjusted respiratory signal, form a respiratory feature array, including:

[0172] All respiratory feature points were determined based on the polarity-adjusted respiratory signal.

[0173] Based on the respiratory feature points, the inter-respiratory phase sequence is obtained;

[0174] Based on the respiratory interval sequence, the mean respiratory rate, standard deviation of respiratory intensity, mean ratio of inspiratory duration to expiratory duration, standard deviation of ratio of inspiratory duration to expiratory duration, and respiratory interval variation trend characteristics are calculated. The respiratory interval variation trend characteristics are the slope obtained by linearly fitting the respiratory rate sequence, which is based on the respiratory interval sequence.

[0175] Optionally, the respiratory feature points include the inspiratory start point and the inspiratory end point (the inspiratory start point is the moment when the signal rises from the trough, and the inspiratory end point is the moment when the signal reaches its peak). The feature point selection results are as follows: Figure 7 As shown, it consists of three parts: before breathing training, during breathing training, and after breathing training. The feature point selection method for each part is as follows: For the polarity-adjusted signal, minimum point detection is performed. If the detected minimum value is less than the preset threshold TH9 and the time interval between it and the previous valid minimum value is reasonable, it is marked as a valid minimum value, and the relevant cache and TH9 are updated. Then, maximum value detection is performed. If the detected maximum value is greater than the preset threshold TH10, and there is a matching valid minimum value point before this maximum value point (the minimum value point flag is 1), then this point is considered the end of inhalation. The maximum value cache and TH10 are updated, and the valid minimum value point flag is set to 0. Then, the inhalation start point is searched backward from the end of inhalation. The inhalation start point needs to meet the following conditions: this point is a minimum value; the amplitude of this point is less than the preset threshold TH9. After finding the first point that meets the above conditions, the search stops; that is, this point is the inhalation start point paired with the end of inhalation.

[0176] It should be noted that the respiratory feature array includes the mean respiratory rate, the standard deviation of respiratory intensity, the mean ratio of inspiratory duration to expiratory duration, the standard deviation of the ratio of inspiratory duration to expiratory duration, and the trend of respiratory interval changes. Based on the found feature points, the above features can be obtained in the following ways:

[0177] Calculate the time interval between adjacent inspiratory end points, i.e., the interspiratory period (BB). (Breath - to - BreathInterval) can be calculated using the following formula:

[0178] ;

[0179] in, This refers to the time corresponding to the end of the i-th inhalation;

[0180] All obtained time intervals are sorted into a respiratory interval sequence (BBI sequence) according to time or other order. The BBI sequence is then low-pass filtered to obtain the BBI baseline sequence. If any BBI in the sequence deviates from its corresponding baseline by a preset threshold (e.g., 50%), it is considered an anomaly (e.g., sudden cough, sensor malfunction) and can be directly removed to avoid interference with the analysis. If the original BBI sequence is discrete and unequally spaced, linear interpolation (or other methods) can be used to generate an equally spaced time series (e.g., one point per second) to obtain the final BBI sequence, adapting the data format to subsequent analysis.

[0181] Then, mean respiratory rate It can be calculated using the following formula:

[0182] ;

[0183] Where k is the number of BBIs;

[0184] The respiratory intensity array is formed by calculating the absolute value of the difference between the amplitudes at the start and end of paired inhalation, and then the standard deviation of the respiratory intensity array is calculated to obtain the respiratory intensity standard deviation.

[0185] Specifically, the inhalation duration is calculated by determining the time difference between the end of the paired inhalation and the start of the inhalation; then, the exhalation duration is calculated by determining the time difference between the start of the next inhalation and the end of the inhalation (the duration of breath-holding is not considered here); finally, all respiratory cycles are traversed, and the ratio of the inhalation duration to the exhalation duration in each cycle is calculated to form an array of the ratios of inhalation duration to exhalation duration.

[0186] The average ratio of inhalation duration to exhalation duration can be obtained by calculating the average of the ratios of inhalation duration to exhalation duration.

[0187] The standard deviation of the ratio of inhalation duration to exhalation duration can be obtained by calculating the standard deviation of the ratio of inhalation duration to exhalation duration.

[0188] Respiratory rate trend characteristics The slope is used for first-order linear fitting of the respiratory sequence. The respiratory rate sequence corresponds one-to-one with the BBI sequence, and the respiratory value is 60 / BBI, which is the respiratory cycle per minute.

[0189] The pre-breathing training feature array refers to the average heart rate, standard deviation of heart rate intervals, trend characteristics of heart rate intervals, average respiratory rate, standard deviation of respiratory intensity, average ratio of inhalation duration to exhalation duration, standard deviation of the ratio of inhalation duration to exhalation duration, and trend characteristics of respiratory interval changes during the period from when the user lies down in the designated area in the specified posture as instructed to the time before the client APP begins the inhalation and exhalation guidance. During this period, the APP will prompt the user to maintain quiet and steady breathing.

[0190] In one embodiment of this application, the step of evaluating breathing training based on the feature array during breathing training and the normalized feature arrays before and after breathing training includes:

[0191] Based on the normalized feature arrays before and after breathing training, a score for the breathing training result is obtained.

[0192] A score for the breathing training process is obtained based on the feature array in the breathing training.

[0193] Based on the breathing training result score, breathing training process score, historical breathing training result score, and historical breathing training process score, a breathing training improvement score is calculated.

[0194] Based on the breathing training result score, breathing training process score, and breathing training improvement score, a breathing training score is calculated.

[0195] Based on the breathing training score, the breathing training assessment result is determined.

[0196] Optionally, by using the normalized feature arrays before and after breathing training, the following four core change features can be calculated respectively: heart rate change features before and after breathing training, standard deviation of heart rate interval change features, respiratory rate change features, and ratio of inhalation to exhalation duration change features. Then, the four features are weighted and summed to obtain the breathing training result score.

[0197] Specifically, the characteristics of heart rate changes before and after breathing training It can be calculated using the following formula:

[0198] ;

[0199] in, The normalized mean heart rate before breathing training. The normalized heart rate variation characteristics, The preset time threshold is set to half the duration of the breathing training. This represents the normalized mean heart rate after breathing training.

[0200] Characteristics of changes in the standard deviation of heart rate intervals before and after breathing training It can be calculated using the following formula:

[0201] ;

[0202] in, The normalized standard deviation of heart rate intervals before breathing training. The normalized standard deviation of heart rate intervals after breathing training.

[0203] Characteristics of respiratory rate changes before and after breathing training It can be calculated using the following formula:

[0204] ;

[0205] in, This represents the normalized mean respiratory rate before breathing training. The normalized respiratory rate variation characteristics, The preset time threshold is set to half the duration of the breathing training. The normalized mean respiratory rate is the rate after breathing training.

[0206] Characteristics of changes in the ratio of inhalation duration to exhalation duration before and after breathing training It can be calculated using the following formula:

[0207] ;

[0208] in, This represents the average ratio of inhalation duration to exhalation duration before breathing training. This represents the average ratio of inhalation duration to exhalation duration after breathing training.

[0209] Then, the heart rate changes before and after the breathing training calculated above were analyzed. Characteristics of changes in the standard deviation of heart rate intervals before and after breathing training Characteristics of respiratory rate changes before and after breathing training And the characteristics of changes in the ratio of inhalation duration to exhalation duration before and after breathing training. Each element is assigned a corresponding weight, and the results are summed to obtain a score for the breathing training outcome. ,

[0210] ;

[0211] in, , , , The weights for each feature are derived from data fitting. If less than 0, take 0. If the value is greater than 45, take 45.

[0212] The breathing training process can be scored using feature arrays, as shown below: First, calculate the absolute value of the difference between the number of breaths and the number of guided breaths during the breathing training process. And during breathing training, the ratio of inhalation duration to exhalation duration compared to the guided inhalation duration to exhalation duration is calculated. If the deviation exceeds a preset threshold for the guided inhalation duration to exhalation duration ratio, such as 30% of the number of breathing cycles... .

[0213] Then, the breathing training process is scored. ;

[0214] in , The weights for each feature are set empirically, for example... Take 4. Take 2, If less than 0, take 0; if greater than 45, take 45.

[0215] Next, historical training data is obtained, namely historical breathing training result scores and historical breathing training process scores. The historical breathing training result scores refer to the breathing training result scores of the previous breathing training session. The historical breathing training process score refers to the score of the previous breathing training process. The results of this breathing training were scored. Breathing training process scoring Scoring of the previous breathing training session and the score of the previous breathing training process This yields a comprehensive assessment of the breathing training. Specifically, a score is first awarded based on the results of this breathing training session. Breathing training process scoring Scoring of the previous breathing training session and the score of the previous breathing training process The breathing training improvement score is calculated using the following formula:

[0216] ;

[0217] in, , The weights for each feature are set empirically, when... , The value is positive, such as 1, otherwise it is 0. , The value is positive, such as 1, otherwise it is 0. Greater than 10, Take 10.

[0218] Finally, the breathing training score ;

[0219] like ≥80, training evaluation result is excellent; otherwise, if ≥60, training evaluation result is good; otherwise, if A score of ≥40 indicates a fair training evaluation result; otherwise, the evaluation result is poor. The effectiveness of breathing training is assessed from multiple dimensions, including before, during, and after training, as well as historical training data, providing a more comprehensive evaluation.

[0220] In this embodiment, physiological signals of the target user before, during, and after breathing training are acquired, and corresponding feature arrays are formed through feature processing. Based on these arrays, breathing training evaluation is performed, achieving full-cycle coverage evaluation of breathing training. This overcomes the limitations of existing technologies that only evaluate before and after training or a single training session. It not only reflects changes in physiological state before and after training but also incorporates dynamic features during the training process, making the evaluation more comprehensive and complete in reflecting the actual effect of breathing training. Furthermore, by introducing resting heart rate and resting respiratory rate, the impact of differences in the user's current state on the evaluation of breathing training effect can be reduced, making the reference value for evaluating breathing training effect more reasonable and providing more reliable technical support for the scientific evaluation of breathing training effectiveness.

[0221] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0222] In one embodiment, a breathing training assessment device is provided, which corresponds one-to-one with the breathing training assessment methods described in the above embodiments. For example... Figure 8 As shown, the breathing training assessment device includes a feature array generation unit 10, a resting data acquisition unit 20, a normalization processing unit 30, and a breathing training assessment unit 40. Detailed descriptions of each functional module are as follows:

[0223] The feature array generation unit 10 is used to acquire the physiological signals of the target user before, during and after breathing training, and to perform feature extraction and quantification to form feature arrays before, during and after breathing training.

[0224] The resting data acquisition unit 20 is used to obtain the normalized resting heart rate and normalized resting respiratory rate of the target user in the current state based on the target user's historical resting heart rate and historical resting respiratory rate.

[0225] The normalization processing unit 30 is used to normalize the feature array before breathing training and the feature array after breathing training based on the resting heart rate and resting respiratory rate of the target user in the current state.

[0226] The breathing training assessment unit 40 is used to assess breathing training based on the feature array during breathing training and the normalized feature arrays before and after breathing training.

[0227] In one embodiment of this application, the physiological signals include heart rate signals and respiratory signals, and the feature array generation unit 10 is further used for:

[0228] After reconstructing and correcting the original physiological sensor signals, the heart rate signal is extracted, and a heart rate feature array is calculated based on the heart rate signal.

[0229] After optimizing the original physiological sensing signal, the respiratory signal envelope is extracted. The respiratory signal is obtained based on the respiratory signal envelope and its polarity is adaptively adjusted. The respiratory feature array is calculated based on the adjusted respiratory signal.

[0230] In one embodiment of this application, the feature array generation unit 10 is further configured to:

[0231] The original physiological sensing signals are subjected to differential processing;

[0232] Perform bandpass filtering on the differentially processed signal;

[0233] Phase compensation is performed on the signal after bandpass filtering by full-pass filtering;

[0234] Polarity correction is performed on the phase-compensated signal;

[0235] The heart rate feature array is calculated based on the feature points of the corrected signal.

[0236] In one embodiment of this application, the feature array generation unit 10 is further configured to:

[0237] Extract a first signal segment with a preset window time length from the phase-compensated signal and calculate the signal quality of the first signal segment;

[0238] If the signal quality is greater than a preset quality threshold, determine the first correlation coefficient array between the first signal segment and the preset signal template;

[0239] The phase-compensated signal is inverted to extract a second signal segment of the preset window time length, and a second correlation coefficient array between the second signal segment and the preset signal template is determined.

[0240] If the mean of the first correlation coefficient array is less than the mean of the second correlation coefficient array, then the polarity is negative, and the phase-compensated signal is inverted.

[0241] In one embodiment of this application, the feature array generation unit 10 is further configured to:

[0242] Construct the initial signal template;

[0243] The initial signal template is normalized to obtain the preset signal template;

[0244] If the signal quality persists for a preset time longer than the preset quality threshold, the update time of the signal template is calculated based on the average heart rate value within the preset time window before the current moment.

[0245] Update the preset signal template according to the specified update time length.

[0246] In one embodiment of this application, the feature array generation unit 10 is further configured to:

[0247] Based on the corrected signal, all heart rate feature points were determined;

[0248] Based on the heart rate feature points, a heart rate interval sequence is obtained;

[0249] Based on the heart rate interval sequence, the average heart rate, the standard deviation of the heart rate interval, and the trend feature of the heart rate interval are obtained. The trend feature of the heart rate interval is the slope of the linear fit of the heart rate sequence, and the heart rate sequence is obtained based on the heart rate interval sequence.

[0250] In one embodiment of this application, the feature array generation unit 10 is further configured to:

[0251] The original physiological sensing signal is filtered.

[0252] Phase compensation is performed on the filtered signal;

[0253] Based on the phase-compensated signal, the respiratory signal envelope is extracted;

[0254] A respiratory signal is obtained based on the respiratory signal envelope, and the polarity of the respiratory signal is adjusted.

[0255] Based on the characteristic points of the polarity-adjusted respiratory signal, a respiratory feature array is calculated.

[0256] In one embodiment of this application, the feature array generation unit 10 is further configured to:

[0257] All respiratory feature points were determined based on the polarity-adjusted respiratory signal.

[0258] Based on the respiratory feature points, the inter-respiratory phase sequence is obtained;

[0259] Based on the respiratory interval sequence, the mean respiratory rate, standard deviation of respiratory intensity, mean ratio of inspiratory duration to expiratory duration, standard deviation of ratio of inspiratory duration to expiratory duration, and respiratory interval variation trend characteristics are calculated. The respiratory interval variation trend characteristics are the slope obtained by linearly fitting the respiratory rate sequence, which is based on the respiratory interval sequence.

[0260] In one embodiment of this application, the breathing training assessment unit 40 is further configured to:

[0261] Based on the normalized feature arrays before and after breathing training, a score for the breathing training result is obtained.

[0262] A score for the breathing training process is obtained based on the feature array in the breathing training.

[0263] Based on the breathing training result score, breathing training process score, historical breathing training result score, and historical breathing training process score, a breathing training improvement score is calculated.

[0264] Based on the breathing training result score, breathing training process score, and breathing training improvement score, a breathing training score is calculated.

[0265] Based on the breathing training score, the breathing training assessment result is determined.

[0266] In this embodiment, physiological signals of the target user before, during, and after breathing training are acquired, and corresponding feature arrays are formed through feature processing. Based on these arrays, breathing training evaluation is performed, achieving full-cycle coverage evaluation of breathing training. This overcomes the limitations of existing technologies that only evaluate before and after training or a single training session. It not only reflects changes in physiological state before and after training but also incorporates dynamic features during the training process, making the evaluation more comprehensive and complete in reflecting the actual effect of breathing training. Furthermore, by introducing resting heart rate and resting respiratory rate, the impact of differences in the user's current state on the evaluation of breathing training effect can be reduced, making the reference value for evaluating breathing training effect more reasonable and providing more reliable technical support for the scientific evaluation of breathing training effectiveness.

[0267] Specific limitations regarding the breathing training assessment device can be found in the limitations of the breathing training assessment method described above, and will not be repeated here. Each module in the aforementioned breathing training assessment device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0268] In one embodiment, a computer device is provided, which may be a terminal device, and its internal structure diagram may be as follows: Figure 9 As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computational and control capabilities. The memory includes a readable storage medium storing computer-readable instructions. The network interface communicates with external terminals via a network connection. When executed by the processor, the computer-readable instructions implement a breathing training assessment method. The readable storage medium provided in this embodiment includes both non-volatile and volatile readable storage media.

[0269] In this application embodiment, a computer device is provided, including a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor. When the processor executes the computer-readable instructions, it implements the steps of the breathing training assessment method described above.

[0270] In one embodiment of the application, a readable storage medium is provided, which stores computer-readable instructions. When the computer-readable instructions are executed by a processor, they implement the steps of the breathing training assessment method described above.

[0271] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by instructing related hardware with computer-readable instructions. These computer-readable instructions can be stored in a non-volatile readable storage medium or a volatile readable storage medium. When executed, these computer-readable instructions can include the processes of the embodiments of the methods described above. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0272] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0273] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A breathing training assessment method, characterized in that, The method includes: Physiological signals of the target user before, during and after breathing training are acquired, and feature extraction and quantification are performed to form feature arrays before, during and after breathing training. Based on the target user's historical resting heart rate and historical resting respiratory rate, the normalized resting heart rate and normalized resting respiratory rate of the target user in the current state are obtained, including: performing low-pass filtering on the historical resting heart rate and historical respiratory rate. In the historical resting heart rate or historical respiratory rate, the long-term stable trend is a low-frequency signal and will be retained, while short-term fluctuations caused by temporary factors will be filtered out. Based on the target user's resting heart rate and resting respiratory rate in the current state, the feature arrays before and after the breathing training are normalized respectively. A breathing training evaluation is performed based on the feature arrays during breathing training and the normalized feature arrays before and after breathing training. This includes: obtaining a breathing training result score based on the normalized feature arrays before and after breathing training; obtaining a breathing training process score based on the feature arrays during breathing training; calculating a breathing training improvement score based on the breathing training result score, the breathing training process score, historical breathing training result scores, and historical breathing training process scores; calculating a breathing training score based on the breathing training result score, the breathing training process score, and the breathing training improvement score; and determining the breathing training evaluation result based on the breathing training score.

2. The breathing training assessment method as described in claim 1, characterized in that, The physiological signals include heart rate signals and respiratory signals. The process involves acquiring physiological signals from the target user before, during, and after breathing training, and performing feature extraction and quantization to form feature arrays for pre-training, during-training, and post-training stages, including: After reconstructing and correcting the original physiological sensor signals, the heart rate signal is extracted, and a heart rate feature array is calculated based on the heart rate signal. After optimizing the original physiological sensing signal, the respiratory signal envelope is extracted. The respiratory signal is obtained based on the respiratory signal envelope and its polarity is adaptively adjusted. The respiratory feature array is calculated based on the adjusted respiratory signal.

3. The breathing training assessment method as described in claim 2, characterized in that, The process of extracting the heart rate signal after reconstructing and correcting the original physiological sensor signal, and calculating the heart rate feature array based on the heart rate signal, includes: The original physiological sensor signal is subjected to differential processing; Perform bandpass filtering on the differentially processed signal; Phase compensation is performed on the signal after bandpass filtering by full-pass filtering; Polarity correction is performed on the phase-compensated signal; Heart rate feature arrays are calculated based on feature points of the corrected signal.

4. The breathing training assessment method as described in claim 3, characterized in that, The polarity correction of the phase-compensated signal includes: Extract a first signal segment with a preset window time length from the phase-compensated signal and calculate the signal quality of the first signal segment; If the signal quality is greater than a preset quality threshold, determine the first correlation coefficient array between the first signal segment and the preset signal template; The phase-compensated signal is inverted to extract a second signal segment of the preset window time length, and a second correlation coefficient array between the second signal segment and the preset signal template is determined. If the mean of the first correlation coefficient array is less than the mean of the second correlation coefficient array, then the polarity is negative, and the phase-compensated signal is inverted.

5. The breathing training assessment method as described in claim 4, characterized in that, Before determining the first correlation coefficient array between the first signal segment and the preset signal template, the following steps are included: Construct the initial signal template; The initial signal template is normalized to obtain the preset signal template; If the signal quality persists for a preset time longer than the preset quality threshold, the update time of the signal template is calculated based on the average heart rate value within the preset time window before the current moment. Update the preset signal template according to the specified update time length.

6. The breathing training assessment method as described in claim 3, characterized in that, The heart rate feature array obtained by calculating feature points based on the corrected signal includes: Based on the corrected signal, all heart rate feature points were determined; Based on the heart rate feature points, a heart rate interval sequence is obtained; Based on the heart rate interval sequence, the average heart rate, the standard deviation of the heart rate interval, and the trend feature of the heart rate interval are obtained. The trend feature of the heart rate interval is the slope of the linear fit of the heart rate sequence, and the heart rate sequence is obtained based on the heart rate interval sequence.

7. The breathing training assessment method as described in claim 2, characterized in that, The process involves optimizing the original physiological sensor signal to extract the respiratory signal envelope, obtaining the respiratory signal based on the respiratory signal envelope and adaptively adjusting its polarity, and calculating a respiratory feature array based on the adjusted respiratory signal, including: The original physiological sensing signal is filtered. Phase compensation is performed on the filtered signal; Based on the phase-compensated signal, the respiratory signal envelope is extracted; A respiratory signal is obtained based on the respiratory signal envelope, and the polarity of the respiratory signal is adjusted. Based on the characteristic points of the polarity-adjusted respiratory signal, a respiratory feature array is calculated.

8. The breathing training assessment method as described in claim 7, characterized in that, The feature points include the inspiratory start point and the inspiratory end point. The feature points based on the polarity-adjusted respiratory signal are used to calculate a respiratory feature array, including: All respiratory feature points were determined based on the polarity-adjusted respiratory signal. Based on the respiratory feature points, the inter-respiratory phase sequence is obtained; Based on the respiratory interval sequence, the mean respiratory rate, standard deviation of respiratory intensity, mean ratio of inspiratory duration to expiratory duration, standard deviation of ratio of inspiratory duration to expiratory duration, and respiratory interval variation trend characteristics are calculated. The respiratory interval variation trend characteristics are the slope obtained by linearly fitting the respiratory rate sequence, which is based on the respiratory interval sequence.

9. A breathing training assessment device, characterized in that, The device includes: The feature array generation unit is used to acquire the physiological signals of the target user before, during and after breathing training, and to perform feature extraction and quantification to form feature arrays before, during and after breathing training. The resting data acquisition unit is used to obtain the normalized resting heart rate and normalized resting respiratory rate of the target user in the current state based on the target user's historical resting heart rate and historical resting respiratory rate, including: performing low-pass filtering on the historical resting heart rate and historical respiratory rate, in which long-term stable trends of historical resting heart rate or historical respiratory rate are low-frequency signals and are filtered out short-term fluctuations caused by temporary factors; The normalization processing unit is used to normalize the feature array before breathing training and the feature array after breathing training based on the resting heart rate and resting respiratory rate of the target user in the current state. The breathing training assessment unit is used to assess breathing training based on the feature array during breathing training and the normalized feature arrays before and after breathing training. The breathing training assessment unit is also used for: Based on the normalized feature arrays before and after breathing training, a score for the breathing training result is obtained. A score for the breathing training process is obtained based on the feature array in the breathing training. Based on the breathing training result score, breathing training process score, historical breathing training result score, and historical breathing training process score, a breathing training improvement score is calculated. Based on the breathing training result score, breathing training process score, and breathing training improvement score, a breathing training score is calculated. Based on the breathing training score, the breathing training assessment result is determined.

Images