A respiratory rehabilitation monitoring and training method and training system

Abstract

The present application relates to the field of respiratory rehabilitation medicine. The present application discloses a kind of respiratory rehabilitation monitoring and training method and training system, it includes S1 synchronization acquisition respiratory airflow velocity, chest and abdomen dynamic and blood oxygen data, integration into initial monitoring data set;S2 clean data removes interference, generates effective data set;S3 calculates respiratory characteristic parameters;S4 combines parameters, abnormal duration and blood oxygen, determines rehabilitation state;S5 state dynamic adjustment training device parameters according to.The present application realizes the individualization of respiratory rehabilitation monitoring and training, solves the one-sidedness of traditional scheme monitoring, determines subjectivity and other problems.By synchronously collecting multidimensional physiological data, interference is removed by cleaning, based on individualized baseline and quantitative indicators, the state determination is more scientific.Oxygen priority determination and heart rate verification are used to build a safety line, and training parameters are dynamically adjusted according to the rehabilitation state, taking into account effectiveness and safety.

Description

Technical Field

[0001] This invention relates to the field of respiratory rehabilitation medicine, and in particular to a respiratory rehabilitation monitoring and training method and training system. Background Technology

[0002] Traditional respiratory rehabilitation training relies heavily on manual assessment by medical staff, primarily monitoring the subject's condition through subjective observation or single physiological indicators. This approach struggles to comprehensively capture multi-dimensional physiological information such as respiratory airflow velocity, chest and abdominal movement, and blood oxygen saturation. This method not only presents incomplete data collection but is also susceptible to human factors, lacking accuracy and timeliness. It fails to reflect dynamic changes in breathing patterns in real time, resulting in a lack of scientific basis for assessing rehabilitation effectiveness and hindering targeted adjustments to training programs.

[0003] Existing respiratory rehabilitation equipment often uses fixed training parameters and uniform judgment thresholds, ignoring the differences in age, physical condition, and disease status among different subjects, and lacking personalized adaptation capabilities. Some devices lack effective data cleaning mechanisms, making them susceptible to motion artifacts, equipment interference, and other factors, resulting in signal distortion and misjudgment of status. Furthermore, the training parameter adjustments of most devices rely on manual operation, failing to dynamically adapt to the subject's real-time physiological state, making it difficult to balance training effectiveness and safety.

[0004] Existing equipment generally lacks a comprehensive safety protection system and has insufficient sensitivity to early warning of respiratory dysfunction, failing to respond promptly to danger signals such as decreased blood oxygenation and abnormal heart rate. Furthermore, some equipment is complex to operate, lacking intuitive breathing guidance and real-time feedback, resulting in low user compliance and difficulty in maintaining standardized training over the long term. These problems severely restrict the continuity, scientific rigor, and safety of respiratory rehabilitation, necessitating an integrated respiratory rehabilitation monitoring and training solution to address these deficiencies. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention discloses a respiratory rehabilitation monitoring and training method and system based on the synchronous acquisition of multi-dimensional physiological data, which achieves dynamic adaptation and safety protection for respiratory rehabilitation training through data cleaning, feature quantification, and personalized state determination.

[0006] This invention discloses a method for respiratory rehabilitation monitoring and training, which includes the following steps:

[0007] S1: Simultaneously collect respiratory airflow velocity data, chest and abdominal respiratory movement data, and blood oxygen saturation data of the subjects, and integrate them into the initial monitoring dataset;

[0008] S2: Clean the initial monitoring dataset to remove signal abrupt changes caused by motion artifacts or equipment interference, and generate a valid monitoring dataset.

[0009] S3: Based on the effective monitoring dataset, calculate at least one respiratory characteristic parameter that characterizes the stability of the subject's breathing pattern. The respiratory characteristic parameter includes respiratory rate fluctuation value, respiratory depth variation value, and respiratory rhythm stability coefficient.

[0010] S4: Based on the values ​​of respiratory characteristic parameters and the duration of their abnormality, and in conjunction with blood oxygen saturation data, comprehensively determine the subject's current respiratory rehabilitation status level; among which, the respiratory rehabilitation status level includes at least normal state, mild respiratory dysfunction state, and severe respiratory dysfunction state.

[0011] S5: Based on the determined respiratory rehabilitation status level, dynamically adjust the working parameters of the respiratory rehabilitation training device. The working parameters include respiratory resistance, the rhythm of visual or auditory guidance signals, and training guidance information displayed in the user interface.

[0012] Further, step S2 involves data cleaning of the initial monitoring dataset, specifically including:

[0013] S21: Divide the initial monitoring dataset into continuous data segments according to a preset time window;

[0014] S22: Perform smoothing filtering on the respiratory airflow velocity data and chest and abdominal respiratory movement data in each data segment, and calculate their local mean.

[0015] S23: Identify signal points in each data segment that deviate from their local mean by more than a first preset threshold as suspected anomalies;

[0016] S24: Physiological rationality verification for suspected outliers: If the heart rate data collected synchronously does not change significantly during the time period in which the suspected outlier appears, the suspected outlier is determined to be a non-physiological mutation and is removed. The data is then filled in using the local mean of the data segment or through an interpolation algorithm.

[0017] Further, step S3 calculates respiratory characteristic parameters, specifically including:

[0018] S31: Extract continuous respiratory cycle time series data from the effective monitoring dataset; the division of respiratory cycles is marked by the inspiratory peak of respiratory airflow velocity, and invalid cycles with significantly abnormal cycle duration due to missed or misjudged detection are automatically removed.

[0019] S32: Based on respiratory cycle time-series data, calculate the difference between the maximum and minimum respiratory rate within a predetermined time window as the respiratory rate fluctuation value; calculate the difference between the maximum and minimum respiratory depth as the respiratory depth variation value.

[0020] S33: Calculate the ratio of the standard deviation of the respiratory cycle duration to its mean, and use it as the respiratory rhythm stability coefficient.

[0021] Further, step S4 determines the respiratory rehabilitation status level, specifically as follows:

[0022] Preset baseline reference ranges corresponding to respiratory characteristic parameters. The baseline reference ranges are set based on the subject's initial measurement values ​​or personal historical data in a resting state.

[0023] When the respiratory rate fluctuation value exceeds the upper limit of its baseline reference range, and / or the respiratory rhythm stability coefficient is lower than the lower limit of its baseline reference range, and this state continues for the first time threshold, it is judged as a mild respiratory dysfunction state.

[0024] When the blood oxygen saturation data is detected to be lower than the second preset threshold, regardless of the respiratory characteristic parameters, it is immediately determined to be a severe respiratory dysfunction state.

[0025] Furthermore, step S4 also includes heart rate-assisted determination:

[0026] Simultaneously monitor the real-time heart rate data of the subjects;

[0027] When the heart rate data continuously exceeds the third preset threshold and reaches the second time threshold, and at the same time the blood oxygen saturation data is not lower than the second preset threshold, a heart rate increase warning message is pushed to the user interface, but the current training level is maintained.

[0028] When the heart rate data continuously exceeds the third preset threshold and reaches the second time threshold, and at the same time the blood oxygen saturation data is lower than the second preset threshold, the severe abnormal state is further confirmed and the alarm level is increased.

[0029] Furthermore, step S5 involves dynamically adjusting the operating parameters of the respiratory rehabilitation training device, specifically as follows:

[0030] When the condition is determined to be normal, maintain or adjust the breathing resistance to the first level and provide the first rhythm breathing guidance signal;

[0031] When a mild respiratory dysfunction is detected, the breathing resistance is lowered to the second level below the first level, and a breathing guidance signal of the second rhythm, which is slower than the first rhythm, is provided. At the same time, breathing technique guidance information is highlighted in the user interface.

[0032] When a severe respiratory dysfunction is detected, the respiratory resistance is reduced to zero, the respiratory guidance signal is suspended, and an emergency alarm and rest prompt are displayed in the user interface, while the event is recorded.

[0033] This invention discloses a respiratory rehabilitation monitoring and training system, applied to a respiratory rehabilitation monitoring and training method for achieving any of the above-mentioned claims, comprising:

[0034] The data acquisition module is used to simultaneously collect physiological data through an airflow sensor, a chest and abdominal respiratory motion sensor, and a blood oxygen sensor;

[0035] The data preprocessing module, which communicates with the data acquisition module, is used to receive physiological data and perform data cleaning.

[0036] The feature extraction module, which communicates with the data preprocessing module, is used to calculate respiratory feature parameters from the cleaned data;

[0037] The status decision module, which communicates with the feature extraction module, is used to assign respiratory rehabilitation status levels to subjects based on respiratory feature parameters, blood oxygen saturation, and duration.

[0038] The control execution module, which communicates with the status decision module, is used to generate control commands based on the respiratory rehabilitation status level to adjust the respiratory rehabilitation training device.

[0039] Furthermore, the data acquisition module also includes a signal quality detection unit, which is used to evaluate the signal-to-noise ratio of the airflow sensor and the chest and abdominal respiratory motion sensor in real time. When the signal-to-noise ratio is lower than the quality threshold, the user interface is triggered to issue a sensor wearing adjustment prompt.

[0040] Furthermore, the state decision module stores a personalized baseline configuration unit, which is used to record the respiratory characteristic parameters of the subject in a quiet state as their personal baseline reference range during system initialization.

[0041] Furthermore, the control execution module includes a resistance control unit, a guide signal generation unit, and a user interface management unit;

[0042] The resistance control unit achieves stepless or stepped adjustment of breathing resistance levels through an electronically controlled valve; the guidance signal generation unit can generate visual light bar animations and auditory beat signals synchronized with the target breathing rhythm.

[0043] The user interface management unit is used to display real-time breathing curves, training progress, and dynamic graphic guidance information based on the current status level.

[0044] The beneficial effects of this invention are:

[0045] This invention achieves precise and personalized respiratory rehabilitation monitoring and training through simultaneous multi-dimensional physiological data acquisition and intelligent analysis, effectively solving the problems of one-sided monitoring, subjective judgment, and poor training adaptability in traditional respiratory rehabilitation programs. The system simultaneously collects respiratory airflow velocity, chest and abdominal respiratory movement, blood oxygen saturation, and heart rate data. Combined with a cleaning mechanism of moving average filtering and heart rate cross-validation, it accurately removes motion artifacts and equipment interference, ensuring the reliability of the raw data. Simultaneously, it constructs a personalized baseline based on initial data from the subject in a resting state, avoiding the incompatibility of fixed group thresholds with subjects of different physical conditions and illnesses. Furthermore, through quantitative calculations of respiratory rate fluctuation values, respiratory depth variation values, and respiratory rhythm stability coefficients, abstract breathing patterns are transformed into objective indicators, making the assessment of respiratory rehabilitation status more scientific and providing a precise basis for subsequent training adjustments.

[0046] This invention constructs a tiered intervention mechanism with safety as its core. It forms multiple safety barriers by prioritizing blood oxygen saturation assessment and using heart rate as an auxiliary verification method, reducing training risks. Simultaneously, it dynamically adjusts training parameters according to different rehabilitation states. Under normal conditions, it trains respiratory muscle groups with moderate resistance and rhythm; in cases of mild abnormalities, it lowers resistance and lengthens the respiratory cycle to guide stable breathing; and in cases of severe abnormalities, it immediately removes the load and suspends training, balancing training effectiveness and safety. The system also improves user convenience and training adherence by providing real-time signal quality detection prompts for sensor adjustment, visualizing respiratory curves, and offering dynamic guidance information. Attached Figure Description

[0047] Figure 1 This is a flowchart of a respiratory rehabilitation monitoring and training method according to an embodiment of this application.

[0048] Figure 2 This is an interface diagram of a respiratory rehabilitation monitoring and training method according to an embodiment of this application.

[0049] Figure 3 This is an interface diagram of a respiratory rehabilitation monitoring and training method according to an embodiment of this application. Detailed Implementation

[0050] To enable those skilled in the art to better understand the present invention, the technical solutions in the specific embodiments of the present invention will be clearly and completely described below.

[0051] This invention discloses a method for respiratory rehabilitation monitoring and training, which includes the following steps:

[0052] S1: Simultaneously collect respiratory airflow velocity data, chest and abdominal respiratory movement data, and blood oxygen saturation data of the subjects, and integrate them into the initial monitoring dataset;

[0053] S2: Clean the initial monitoring dataset to remove signal abrupt changes caused by motion artifacts or equipment interference, and generate a valid monitoring dataset.

[0054] S3: Based on the effective monitoring dataset, calculate at least one respiratory characteristic parameter that characterizes the stability of the subject's breathing pattern. The respiratory characteristic parameter includes respiratory rate fluctuation value, respiratory depth variation value, and respiratory rhythm stability coefficient.

[0055] S4: Based on the values ​​of respiratory characteristic parameters and the duration of their abnormality, and in conjunction with blood oxygen saturation data, comprehensively determine the subject's current respiratory rehabilitation status level; among which, the respiratory rehabilitation status level includes at least normal state, mild respiratory dysfunction state, and severe respiratory dysfunction state.

[0056] S5: Based on the determined respiratory rehabilitation status level, dynamically adjust the working parameters of the respiratory rehabilitation training device. The working parameters include respiratory resistance, the rhythm of visual or auditory guidance signals, and training guidance information displayed in the user interface.

[0057] Step S1: Simultaneously collect the subject's respiratory airflow velocity data, chest and abdominal respiratory movement data, and blood oxygen saturation data, and integrate them into an initial monitoring dataset. Respiratory airflow velocity data refers to the airflow velocity during the subject's inhalation or exhalation, measured by an airflow sensor. Its typical detection range is 0.2 L / s to 10 L / s, with a sampling frequency of 5 Hz to 20 Hz to ensure detailed capture of the respiratory waveform. Chest and abdominal respiratory movement data refers to the amplitude of chest and abdominal movement during respiration, measured by a flexible pressure sensor. Its range is divided into two levels depending on the subject group: for children or individuals with weak respiratory movement, the range is 0 kPa to 3 kPa; for adults or obese individuals, the range is 0 kPa to 8 kPa. The sensor sampling frequency is synchronized with the airflow sensor. Blood oxygen saturation data refers to the percentage of oxyhemoglobin in the blood, measured by a photoelectric sensor. Its detection range is 70% to 100%, with a sampling frequency of 1 Hz to 10 Hz. Synchronous acquisition refers to ensuring the time alignment of various data types through a unified clock signal or bus protocol, with time errors controlled within 200 milliseconds to guarantee data consistency. Multi-parameter synchronous acquisition can comprehensively capture respiratory physiological states, reduce misjudgments caused by data asynchrony, and provide a reliable foundation for subsequent analysis.

[0058] Step S2: Clean the initial monitoring dataset to remove signal abrupt changes caused by motion artifacts or equipment interference, generating a valid monitoring dataset. Data cleaning refers to identifying and processing abnormal data points using algorithms. Specifically, this includes: dividing the initial monitoring dataset into continuous data groups according to a preset time window (configurable from 5 to 20 seconds to accommodate different respiratory rates); applying a moving average algorithm to smooth each data group and calculating the local mean; initially identifying data points deviating from the local mean by 20% to 40% as suspected abrupt changes, and verifying them with synchronously collected heart rate data. If the heart rate fluctuation does not exceed 5 beats / minute, it is confirmed as a non-physiological abrupt change and replaced with the local mean or filled in by linear interpolation. This effectively removes noise and interference, improves data quality, ensures the accuracy of subsequent feature calculations, and avoids the erroneous removal of non-physiological fluctuations through heart rate-assisted verification.

[0059] Step S3: Based on the effective monitoring dataset, calculate at least one respiratory characteristic parameter characterizing the stability of the subject's breathing pattern. These parameters include respiratory rate fluctuation, respiratory depth variation, and respiratory rhythm stability coefficient. The respiratory rate fluctuation refers to the difference between the maximum and minimum respiratory rate within a certain time window (e.g., 30 to 60 seconds), typically ranging from 0 to 6 breaths / minute. For children, this range can be configured to 0 to 4 breaths / minute. Standardization is used to eliminate the influence of dimensions. The respiratory depth variation refers to the difference between the maximum and minimum respiratory depth within the same time window, typically ranging from 0 to 4 centimeters. For children, this range can be configured to 0 to 3 centimeters. Standardization is also applied. The respiratory rhythm stability coefficient is the ratio of the standard deviation of the duration of a continuous respiratory cycle to the average cycle duration. It is calculated based on 5 to 10 effective respiratory cycles. Cycle extraction uses the inspiratory peak of the respiratory airflow velocity as a marker. If the cycle duration deviates from the average by more than 50%, it is considered invalid and discarded. The respiratory rhythm stability coefficient ranges from 0 to 1; a smaller value indicates a more stable rhythm, with an ideal value approaching 0. Quantifying complex breathing patterns into a set of characteristic parameters facilitates objective assessment of respiratory stability and provides calculable indicators for state determination.

[0060] Step S4: Based on the values ​​of respiratory characteristic parameters and their duration of abnormality, combined with blood oxygen saturation data, comprehensively determine the subject's current respiratory rehabilitation status level. The respiratory rehabilitation status level includes at least a normal state, a mild respiratory dysfunction state, and a severe respiratory dysfunction state. The duration of abnormality refers to the duration for which respiratory characteristic parameters exceed a preset baseline range. The baseline range is set based on the subject's historical data under resting conditions; for example, the upper limit of the baseline for respiratory rate fluctuation is 5 breaths / minute, and the lower limit of the baseline for respiratory depth variation is 2 cm. The abnormal threshold for blood oxygen saturation data is 90%; values ​​below this value are considered severe abnormalities. The determination rule is as follows: if the respiratory rate fluctuation value is greater than the upper limit of the baseline and / or the respiratory rhythm stability coefficient is lower than the lower limit of the baseline, and the duration exceeds 3-5 minutes, it is determined to be a mild respiratory dysfunction state; if blood oxygen saturation is below 90%, it is immediately determined to be a severe respiratory dysfunction state, regardless of the respiratory characteristic parameters. Through multi-dimensional data fusion determination, the sensitivity and specificity of status identification are improved, ensuring timely warning of respiratory function deterioration and protecting the subject's safety.

[0061] Step S5: Based on the determined respiratory rehabilitation status level, dynamically adjust the operating parameters of the respiratory rehabilitation training device. These parameters include respiratory resistance, the rhythm of visual or auditory guidance signals, and training guidance information displayed in the user interface. Dynamic adjustment refers to adjusting the training intensity in real time according to the status level: For a normal state, the respiratory resistance is set to a low level, such as 5-10 cmH2O, and the guidance signal cycle is 3-5 seconds; for a mild respiratory dysfunction state, the respiratory resistance is lowered to an even lower level or zero resistance, the guidance signal cycle is extended to 6-8 seconds, and guidance on techniques such as diaphragmatic breathing is displayed; for a severe respiratory dysfunction state, training is immediately paused, the resistance is reduced to zero, and an audible and visual alarm is triggered. This achieves personalized adaptive training, optimizing rehabilitation effects while ensuring safety, and guiding the subject to correctly perform training movements through an intuitive interface.

[0062] For setting respiratory resistance, in addition to using a fixed water column setting, an optional method based on a percentage of the patient's individual respiratory muscle strength can be adopted: first, the system collects the patient's maximum inspiratory pressure at rest during the initialization phase as a baseline respiratory muscle strength reference value, and then sets the respiratory resistance to 30%-50% of this baseline respiratory muscle strength reference value to adapt to differences in respiratory function among different ages and physical conditions; correspondingly, in cases of mild respiratory dysfunction, the respiratory resistance can be lowered to 10%-20% of the baseline respiratory muscle strength reference value, which avoids the problem of excessive respiratory load that a fixed resistance setting may cause to special groups, and ensures training safety and suitability through individualized muscle strength adaptation.

[0063] As one implementation method, step S2 cleans the initial monitoring dataset, specifically including:

[0064] S21: Divide the initial monitoring dataset into continuous data segments according to a preset time window;

[0065] S22: Perform smoothing filtering on the respiratory airflow velocity data and chest and abdominal respiratory movement data in each data segment, and calculate their local mean.

[0066] S23: Identify signal points in each data segment that deviate from their local mean by more than a first preset threshold as suspected anomalies;

[0067] S24: Physiological rationality verification for suspected outliers: If the heart rate data collected synchronously does not change significantly during the time period in which the suspected outlier appears, the suspected outlier is determined to be a non-physiological mutation and is removed. The data is then filled in using the local mean of the data segment or through an interpolation algorithm.

[0068] Step S21 divides the initial monitoring dataset into continuous data segments according to a preset time window. The preset time window can be configured based on the subject's respiratory rate, ranging from five to twenty seconds. A data segment refers to cutting a continuous time-series data stream into multiple data units of equal or approximately equal length; this division facilitates local feature analysis. By dividing massive amounts of data into short time segments, the efficiency and real-time performance of data processing can be significantly improved, while also laying the foundation for subsequent local anomaly detection.

[0069] Step S22 performs smoothing filtering on the respiratory airflow velocity data and chest and abdominal respiratory movement data within each data segment, and calculates their local means. The smoothing filtering uses a moving average algorithm to suppress fluctuations within the data segment; the length of the moving window is typically set to one-tenth to one-fifth of the data segment length. The local mean is the arithmetic mean of the smoothed data segment, representing the baseline level of the signal within that time period. Filtering effectively suppresses high-frequency random noise, preserves the macroscopic trend of the respiratory signal, and makes subsequent anomaly identification more accurate.

[0070] Step S23 identifies signal points in each data segment that deviate from their local mean by more than a first preset threshold as suspected anomalies. The first preset threshold is a configurable relative percentage threshold, set between 20% and 40%. A signal point refers to a single sampled data point in a data segment, while a suspected anomaly refers to data points that significantly deviate from normal physiological patterns due to motion artifacts or transient device interference. By setting a dynamic threshold, potential abnormal data can be automatically filtered out, providing targets for subsequent accurate judgment and avoiding the subjectivity of manual setting.

[0071] Step S24 verifies the physiological rationality of suspected anomalies. Specifically, if the synchronously collected heart rate data does not change significantly within the time period of the suspected anomaly, the suspected anomaly is determined to be a non-physiological mutation and is removed. Here, a significant change in heart rate data is defined as a fluctuation of more than five beats per minute in adjacent sampling periods. Adjacent sampling periods refer to consecutive 0.2-1 second sampling intervals of the heart rate sensor. A heart rate fluctuation of more than five beats per minute means that the difference between the instantaneous heart rate values ​​calculated from two consecutive 0.2-1 second sampling periods is greater than 5 beats / minute. For the removed data points, the local mean of the data segment or a linear interpolation algorithm is used to fill in the data. The linear interpolation algorithm refers to the method of calculating the filler value using adjacent normal data points before and after the anomaly point according to a linear relationship. Introducing heart rate as an independent physiological parameter for cross-validation can effectively distinguish between real physiological anomalies and equipment noise, improve the scientific rigor and reliability of data cleaning, and ensure the quality of the data used for subsequent analysis.

[0072] As one implementation method, step S3 calculates respiratory characteristic parameters, specifically including:

[0073] S31: Extract continuous respiratory cycle time series data from the effective monitoring dataset; the division of respiratory cycles is marked by the inspiratory peak of respiratory airflow velocity, and invalid cycles with significantly abnormal cycle duration due to missed or misjudged detection are automatically removed.

[0074] S32: Based on respiratory cycle time-series data, calculate the difference between the maximum and minimum respiratory rate within a predetermined time window as the respiratory rate fluctuation value; calculate the difference between the maximum and minimum respiratory depth as the respiratory depth variation value.

[0075] S33: Calculate the ratio of the standard deviation of the respiratory cycle duration to its mean, and use it as the respiratory rhythm stability coefficient.

[0076] Step S31 extracts continuous respiratory cycle time-series data from the valid monitoring dataset. A respiratory cycle refers to the time taken to complete one full inhalation and exhalation, defined by the inspiratory peak in the respiratory airflow velocity waveform. The inspiratory peak is the maximum airflow velocity reached within a single respiratory cycle, marking the end of the inspiratory phase and the beginning of the expiratory phase. The system automatically removes invalid cycles with significantly abnormal cycle lengths due to signal acquisition anomalies or misjudgments by the analysis algorithm. Invalid cycles are defined as those whose duration deviates from the configurable range of 50% to 70% of the average duration of adjacent cycles. To ensure statistical reliability, the system requires the continuous extraction of 5 to 10 valid respiratory cycles to form a time-series dataset. Through cycle division and automatic verification mechanisms, the continuity and physiological rationality of the data used for subsequent feature parameter calculations are ensured, laying a solid foundation for objectively assessing respiratory patterns.

[0077] To ensure the system can maintain basic operation even with poor signal quality, if at least 5 valid respiratory cycles cannot be obtained within a continuous 60-second time window according to the conventional judgment criteria, the system automatically activates a degraded mode, temporarily relaxing the invalid cycle judgment criteria to 80% deviation. In this mode, once the system successfully obtains at least 3 valid cycles, it uses these as the basis for subsequent feature calculations. However, at the same time, a message indicating reduced data reliability is displayed on the user interface to alert the user that the current results may be affected by data quality.

[0078] Step S32, based on the respiratory cycle time-series data obtained in step S31, calculates two key variability indicators within a predetermined analysis time window. The predetermined time window is typically set to 30 to 60 seconds to cover a sufficient number of respiratory cycles for calculation. The difference between the maximum and minimum respiratory rate within this time window is calculated as the respiratory rate fluctuation value, measured in breaths per minute (BPM), with a typical range of 0-6 BPM for normal adults at rest. Respiratory rate refers to the number of complete respiratory cycles completed per minute. The difference between the maximum and minimum respiratory depth within this time window is calculated as the respiratory depth variability value, measured in centimeters (cm), with a typical range of 0-4 cm for normal adults at rest. Respiratory depth is a physical quantity obtained by calibrating displacement data collected from a chest and abdominal respiratory motion sensor, characterizing the amplitude of a single breath. Quantifying the dynamic changes in respiratory rate and depth over time into numerical indicators with clear physiological significance can intuitively reflect the degree of variability in the subject's breathing pattern, providing a basis for assessing the stability of respiratory control function.

[0079] Step S33 calculates the respiratory rhythm stability coefficient. The respiratory rhythm stability coefficient is defined as the ratio of the standard deviation of the duration of a continuous effective respiratory cycle to its arithmetic mean, i.e., the coefficient of variation. The duration of a respiratory cycle refers to the time interval between two adjacent inspiratory peak markers, measured in seconds. This coefficient is a dimensionless statistic, ranging from 0 to 1. The closer the value is to 0, the more regular and stable the respiratory rhythm; the closer the value is to 1, the more disordered and unstable the respiratory rhythm. By using this statistical indicator, the coefficient of variation, the influence of individual differences in baseline respiratory rate on the assessment results is effectively eliminated, purely characterizing the stability of the respiratory rhythm from a time dimension, thus providing a sensitive and objective quantitative basis for assessing respiratory neural regulation function.

[0080] As one implementation method, step S4 determines the respiratory rehabilitation status level, specifically as follows:

[0081] Preset baseline reference ranges corresponding to respiratory characteristic parameters. The baseline reference ranges are set based on the subject's initial measurement values ​​or personal historical data in a resting state.

[0082] When the respiratory rate fluctuation value exceeds the upper limit of its baseline reference range, and / or the respiratory rhythm stability coefficient is lower than the lower limit of its baseline reference range, and this state continues for the first time threshold, it is judged as a mild respiratory dysfunction state.

[0083] When the blood oxygen saturation data is detected to be lower than the second preset threshold, regardless of the respiratory characteristic parameters, it is immediately determined to be a severe respiratory dysfunction state.

[0084] First, a baseline reference range corresponding to the respiratory characteristic parameters needs to be preset. The baseline reference range refers to the normal fluctuation range of various respiratory characteristic parameters of the subject in a resting state. It is set based on the subject's initial measurements under quiet and relaxed conditions or their personal historical training data. For example, during the system initialization phase, the respiratory rate fluctuation value, respiratory depth variation value, and respiratory rhythm stability coefficient are recorded during 3 to 5 minutes of quiet breathing. The average value plus or minus two standard deviations is taken as the initial baseline reference range. This fully considers the physiological differences between different subjects, avoids the maladaptive effects that may arise from using fixed thresholds, and makes the state assessment more individualized and accurate.

[0085] A mild respiratory dysfunction is diagnosed when the respiratory rate fluctuation exceeds the upper limit of its baseline reference range, and / or the respiratory rhythm stability coefficient falls below the lower limit of its baseline reference range, and this state persists for a first time threshold. Here, "and / or" indicates that either or both conditions must be met to trigger the diagnosis. The first time threshold is a configurable time parameter, typically set between 3 and 5 minutes. "Persistent" means that the aforementioned abnormal characteristics need to be detected continuously or frequently within this time window, rather than as transient fluctuations. By introducing the duration dimension, temporary physiological fluctuations can be effectively distinguished from persistent respiratory dysfunction, ensuring the system's sensitivity to genuine abnormalities while avoiding overreaction to transient disturbances.

[0086] When blood oxygen saturation is detected to be below a second preset threshold, the system will immediately classify it as a state of severe respiratory dysfunction, regardless of the level of respiratory characteristic parameters. The second preset threshold is set at 90% based on the clinical diagnostic criteria for hypoxemia. "Immediate judgment" means that once this condition is met, the system will switch the status level within the next processing cycle, typically in less than three seconds. This establishes the highest priority safety barrier, ensuring the fastest possible emergency response when a significant drop in blood oxygen saturation occurs, maximizing the safety of the training subjects.

[0087] As one implementation method, step S4 further includes heart rate-assisted determination:

[0088] Simultaneously monitor the real-time heart rate data of the subjects;

[0089] When the heart rate data continuously exceeds the third preset threshold and reaches the second time threshold, and at the same time the blood oxygen saturation data is not lower than the second preset threshold, a heart rate increase warning message is pushed to the user interface, but the current training level is maintained.

[0090] When the heart rate data continuously exceeds the third preset threshold and reaches the second time threshold, and at the same time the blood oxygen saturation data is lower than the second preset threshold, the severe abnormal state is further confirmed and the alarm level is increased.

[0091] The system synchronously monitors the subject's real-time heart rate data using photoplethysmography (PPG) technology. Real-time heart rate data refers to the heart rate calculated from the heartbeat cycle signals continuously collected by a fingertip or wrist sensor. The sampling frequency is 1 to 5 Hz, and the detection range covers 40 to 200 beats per minute to meet the monitoring needs from resting to light activity.

[0092] When the monitored heart rate data consistently exceeds the third preset threshold and reaches the second time threshold, while the blood oxygen saturation data is not lower than the second preset threshold, the system will push a heart rate elevation warning message to the user interface, but maintain the current training level. The third preset threshold is set to 110 beats per minute, based on the safe range of adult resting heart rate. The second time threshold is set to 30 seconds to ensure the stability of the judgment. "Consistently exceeding" means that within this 30-second time window, the heart rate data must remain above the threshold rather than fluctuating momentarily. When isolated heart rate elevation occurs while the blood oxygen level is normal, the system can identify that this may be due to emotional stress or non-respiratory factors. By issuing a gentle warning rather than interrupting training, the system alerts the user to changes in physiological state while avoiding unnecessary training interference, thus improving the user experience.

[0093] When the heart rate data continuously exceeds the third preset threshold and reaches the second time threshold, and at the same time the blood oxygen saturation data is lower than the second preset threshold, the severe abnormal state is further confirmed and the alarm level is increased.

[0094] This invention constructs a more comprehensive respiratory status assessment system. By cross-referencing heart rate and blood oxygen data, it significantly improves the accuracy and reliability of status judgment and provides a higher level of safety assurance for subjects.

[0095] As one implementation method, step S5 involves dynamically adjusting the operating parameters of the respiratory rehabilitation training device, specifically as follows:

[0096] When the condition is determined to be normal, maintain or adjust the breathing resistance to the first level and provide the first rhythm breathing guidance signal;

[0097] When a mild respiratory dysfunction is detected, the breathing resistance is lowered to the second level below the first level, and a breathing guidance signal of the second rhythm, which is slower than the first rhythm, is provided. At the same time, breathing technique guidance information is highlighted in the user interface.

[0098] When a severe respiratory dysfunction is diagnosed, respiratory resistance is reduced to zero, respiratory guidance signals are paused, and an emergency alarm and rest prompt are displayed in the user interface. Simultaneously, complete physiological data, including respiratory airflow, chest and abdominal movement, blood oxygenation, and heart rate, are recorded for subsequent medical analysis within two minutes before and one minute after the event. In addition to the basic intervention of reducing respiratory resistance to zero, a low respiratory resistance level of 1-2 cmH2O can be maintained to sustain slight respiratory muscle activity, depending on the patient's real-time respiratory tolerance. Simultaneously, a slow abdominal breathing guidance mode is initiated to assist the patient in improving ventilation efficiency through standardized breathing movements, alleviating hypoxia, and providing a buffer time for subsequent medical treatment. Both modes can be set as needed.

[0099] When the system determines that the subject is in a normal state, it maintains or adjusts the breathing resistance to the first level. The first level corresponds to a moderate training load, with a resistance value typically set between 5 and 10 cmH2O, designed to maintain and moderately exercise the respiratory muscles. Simultaneously, the system provides a breathing guidance signal for the first rhythm. This signal typically consists of a visual light bar animation and a synchronized auditory beat. A complete breathing cycle, i.e., one inhalation and one exhalation, lasts for 3 to 5 seconds, with inhalation accounting for approximately 35% and exhalation approximately 65%. This provides a highly efficient and comfortable rehabilitation training rhythm for subjects in a good state, optimizing the user experience while ensuring training effectiveness and helping to maintain training adherence.

[0100] When the system determines that the subject is in a state of mild respiratory dysfunction, to ensure safety and promote the recovery of the breathing pattern, the respiratory resistance is lowered to a second level, below the first level. The resistance value of the second level is typically a low-load level of 0 cmH2O to 5 cmH2O. The provided breathing guidance signal is adjusted to a second rhythm that is slower than the first rhythm, with its complete cycle extended to 6 to 8 seconds, aiming to guide the subject to take deeper and smoother breaths. By reducing physical load, extending the respiratory cycle, and providing clear technique guidance, this multi-pronged approach helps the subject correct disordered breathing patterns and rebuild a normal breathing rhythm, demonstrating the auxiliary and corrective functions of rehabilitation training.

[0101] When the system determines that a subject is in a state of severe respiratory dysfunction, it will execute the highest level of safety intervention. First, respiratory resistance will be immediately reduced to zero centimeter-hydraulic column, completely removing the training load. Simultaneously, all respiratory guidance signals will be suspended to eliminate any external factors that might interfere with the subject's spontaneous breathing recovery. The system will automatically record a detailed timestamp of this abnormal event and associated physiological data. Prioritizing subject safety, this system establishes a crucial safety baseline by rapidly removing all training stimuli and providing complete data records for subsequent medical assessments.

[0102] Through the aforementioned graded adjustment mechanism, this invention achieves comprehensive intelligent adaptation of respiratory rehabilitation training, from load intensity and guidance rhythm to information prompts, ensuring the safety, effectiveness, and individualization of the training process.

[0103] This invention discloses a respiratory rehabilitation monitoring and training system, comprising:

[0104] The data acquisition module is used to simultaneously collect physiological data through an airflow sensor, a chest and abdominal respiratory motion sensor, and a blood oxygen sensor;

[0105] The data preprocessing module, which communicates with the data acquisition module, is used to receive physiological data and perform data cleaning.

[0106] The feature extraction module, which communicates with the data preprocessing module, is used to calculate respiratory feature parameters from the cleaned data;

[0107] The status decision module, which communicates with the feature extraction module, is used to assign respiratory rehabilitation status levels to subjects based on respiratory feature parameters, blood oxygen saturation, and duration.

[0108] The control execution module, which communicates with the status decision module, is used to generate control commands based on the respiratory rehabilitation status level to adjust the respiratory rehabilitation training device.

[0109] The data acquisition module is used to simultaneously collect physiological data through an airflow sensor, a chest and abdominal respiratory motion sensor, and a blood oxygen sensor. The airflow sensor employs a thermal or differential pressure sensing principle, with a sampling frequency of 5 Hz to 20 Hz and a detection range of 0.2 L / s to 10 L / s. The chest and abdominal respiratory motion sensor uses a flexible thin-film pressure sensor, offering two selectable ranges: 0-3 kPa and 0-8 kPa, suitable for subjects of different body types. The blood oxygen sensor is based on the photoplethysmography (PPG) principle, detecting blood oxygen saturation from 70% to 100%. The data acquisition module also includes a heart rate sensor to collect heart rate data. All sensors achieve synchronous data transmission via a controller area network (LAN) bus, with synchronization time errors controlled within 200 milliseconds. Through synchronous acquisition by multiple sensors, a comprehensive respiratory physiological data acquisition system is constructed, providing a rich and reliable raw data source for subsequent analysis.

[0110] The data preprocessing module communicates with the data acquisition module via a serial peripheral interface or integrated circuit bus to receive raw physiological data and perform data cleaning. This module includes a data grouping unit, a mean calculation unit, and a mutation value processing unit. It can group data into configurable time windows of five to twenty seconds and smooth the data using a moving average algorithm. The moving average window length is one-tenth to one-fifth of the data segment length. Through a professional data cleaning process, motion artifacts and environmental interference are effectively removed, improving data quality and preparing a clean dataset for feature extraction. If the signal quality remains below the threshold for more than one minute, the system will automatically pause training and display a message on the user interface indicating that signal instability has been detected, requesting the user to check the sensor wear until the signal recovers.

[0111] The feature extraction module communicates with the data preprocessing module to calculate respiratory characteristic parameters from the cleaned data. This module integrates a two-dimensional sorting unit and a feature solving subunit, capable of calculating respiratory rate fluctuations, respiratory depth variations, and respiratory rhythm stability coefficients based on a time window of 30 to 60 seconds. All feature parameters are standardized to eliminate the influence of dimensions. This transforms complex respiratory waveform signals into quantitative features with clear clinical significance, enabling an objective assessment of respiratory pattern stability.

[0112] The status decision module communicates with the feature extraction module to assign respiratory rehabilitation status levels to subjects based on respiratory characteristic parameters, blood oxygen saturation, and duration. This module includes a multi-parameter timing unit and a level determination subunit, capable of continuously monitoring abnormal statuses of characteristic parameters for 3 to 5 minutes and determining the status based on a personalized baseline reference range. Through multi-dimensional data fusion analysis and duration verification, the accuracy and reliability of status determination are ensured, effectively distinguishing between physiological fluctuations and pathological abnormalities. The multi-parameter timing unit contains one or more status timers to accumulate the duration of specific abnormal conditions. When an abnormal condition is determined to be met based on the latest calculated respiratory characteristic parameters, the corresponding status timer starts or continues to accumulate; when the abnormal condition is not met, the timer is reset. The level determination subunit can continuously monitor abnormal statuses of characteristic parameters for 3 to 5 minutes and determine the status based on a personalized baseline reference range. Its determination logic is as follows: when the accumulated value of the status timer corresponding to an abnormal condition reaches a preset first time threshold, a corresponding status level change is triggered, such as determining a mild respiratory dysfunction status. Status determination follows the highest safety priority principle. A blood oxygen saturation level below 90% is the highest priority and is unconditionally classified as severely abnormal. The next highest priority is the combined result of heart rate-assisted assessment and respiratory characteristic assessment. When the results from different assessment pathways are inconsistent, the system adopts the assessment result with the higher risk level.

[0113] The control execution module communicates with the state decision module to generate control commands based on the respiratory rehabilitation state level, thereby adjusting the respiratory rehabilitation training device. This module includes a resistance control unit, an audio-visual guidance unit, and a prompting push unit. It can precisely control the solenoid valves via pulse width modulation signals to achieve resistance adjustment from zero to four levels, while simultaneously generating corresponding rhythmic audio-visual guidance signals. This achieves closed-loop control of training parameters, automatically adjusting the training intensity based on the subject's real-time physiological state, ensuring both training safety and optimized rehabilitation effects.

[0114] Through the organic coordination of the above modules, this invention constructs a complete respiratory rehabilitation monitoring and training platform, realizing an automated process from data collection, processing, analysis to execution, and providing scientific and reliable technical support for respiratory rehabilitation training.

[0115] As one implementation, the data acquisition module also includes a signal quality detection unit, which is used to evaluate the signal-to-noise ratio of the airflow sensor and the chest and abdominal respiratory motion sensor in real time. When the signal-to-noise ratio is lower than the quality threshold, the user interface is triggered to issue a sensor wearing adjustment prompt.

[0116] The signal quality detection unit quantifies signal quality by calculating the signal-to-noise ratio (SNR). SNR is the ratio of effective signal power to noise power, measured in decibels (dB). Its calculation is based on the power spectral density estimation of the sensor's output signal. Signal power is calculated within the main frequency range of the breathing signal from 0.1 Hz to 0.5 Hz, and noise power is calculated in the high-frequency band to obtain the final SNR value. The quality threshold is set to a configurable range of 10 dB to 15 dB. When the calculated SNR is lower than this threshold, the signal quality is deemed unacceptable.

[0117] When a signal quality defect is detected, the system immediately triggers a sensor wearing adjustment prompt on the user interface. This prompt includes specific adjustment instructions, such as checking if the airflow sensor is aligned with the mouth and nose or adjusting the tightness of the chest strap. The prompt is displayed prominently in the user interface and accompanied by a soft audible alert. The system continuously monitors the signal quality, and the prompt will automatically disappear after the signal-to-noise ratio recovers to above the quality threshold and remains above it for ten to thirty seconds.

[0118] By monitoring signal quality in real time and promptly alerting users to adjustments when quality deteriorates, the reliability and accuracy of the collected data are fundamentally guaranteed. This proactive quality assurance mechanism effectively avoids data distortion caused by poor sensor contact or displacement, ensuring that all subsequent analysis and decisions are based on high-quality raw data, thereby improving the overall system reliability and user experience. The signal quality detection unit can also evaluate the signal-to-noise ratio (SNR) of the blood oxygen and heart rate sensors in real time. When the SNR falls below a quality threshold, it triggers a sensor adjustment prompt on the user interface.

[0119] As one implementation, the state decision module stores a personalized baseline configuration unit, which is used to record the respiratory characteristic parameters of the subject in a quiet state as their personal baseline reference range during system initialization.

[0120] During the system initialization phase, when the subject is in a quiet resting state, the personalized baseline configuration unit automatically initiates the data acquisition and analysis process. The initialization phase requires the subject to maintain calm breathing for three to five minutes, during which time the system continuously collects data on respiratory airflow rate, chest and abdominal respiratory movement, and blood oxygen saturation. A quiet state refers to a physiological state where the subject is relaxed in a sitting or lying position, has not undergone active breathing training, and is emotionally stable. The system supports both manual and automatic baseline updates. Automatic updates are triggered when the subject has maintained a normal state for three consecutive training sessions and their respiratory stability score is higher than the preset target. A new baseline reference range is calculated based on the data from these three training sessions. Healthcare professionals can also manually confirm or adjust the baseline after assessment. The respiratory stability score is based on a 100-point scale, calculated as: Score = (1 - Respiratory Rhythm Stability Coefficient) × 100.

[0121] The system extracts and calculates respiratory characteristic parameters from collected resting state data, including respiratory rate fluctuation, respiratory depth variability, and respiratory rhythm stability coefficient. The ranges of these parameters are based on extensive clinical data: respiratory rate fluctuation is typically 0.5 to 3 breaths per minute, and respiratory depth variability is generally 0 to 4 centimeters. The system uses the average value of these parameters during initialization as the baseline value, and determines the individual baseline reference range within a range of plus or minus two standard deviations of the baseline value. Respiratory depth data is obtained from pressure data collected by a chest and abdominal respiratory motion sensor, which is converted into physical displacement through an individualized calibration procedure. The individualized calibration procedure establishes a linear relationship model between chest and abdominal pressure changes and estimated vital capacity or tidal volume by having the subject perform several deep breaths of standard amplitude, thereby converting the pressure data into equivalent displacement in centimeters.

[0122] The individual baseline reference range represents the fluctuation range of the subject's respiratory characteristic parameters under normal conditions. All data is stored in the module's non-volatile memory and can be retrieved in subsequent training sessions. After initialization, the system allows subjects or medical personnel to fine-tune the baseline reference range by 5% to 10% according to actual conditions to better adapt to individual needs.

[0123] By establishing individualized assessment benchmarks, the accuracy and relevance of condition determination have been significantly improved. Compared with using fixed population thresholds, personalized baselines can more sensitively identify subtle changes in the respiratory function of subjects, effectively reducing misjudgments and missed diagnoses, providing a scientific basis for personalized respiratory rehabilitation training, and enhancing the system's adaptability to subjects with different physical conditions.

[0124] In one implementation, the control execution module includes a resistance control unit, a guide signal generation unit, and a user interface management unit;

[0125] The resistance control unit achieves stepless or stepped adjustment of breathing resistance levels through an electronically controlled valve;

[0126] The guidance signal generation unit can generate visual light bar animations and auditory beat signals synchronized with the target's breathing rhythm;

[0127] The user interface management unit is used to display real-time breathing curves, training progress, and dynamic graphic guidance information based on the current status level.

[0128] The resistance control unit adjusts the breathing resistance level via a high-precision electrically controlled valve. This valve is driven by a stepper motor or proportional electromagnet and controlled by pulse-width modulation signals, with a duty cycle adjustment range of 5% to 95%. The unit supports two adjustment modes: a stepless adjustment mode that allows for continuous and smooth adjustment of the resistance value within the range of 0 to 20 cmH2O; and a graded adjustment mode that provides four to six preset levels, each corresponding to a specific resistance range. The calibration accuracy of the resistance value is controlled within ±0.5 cmH2O. This allows for precise and adjustable respiratory load based on the subject's real-time physiological state, ensuring both the effectiveness and safety of the training process.

[0129] The guidance signal generation unit is responsible for generating multi-sensory guidance signals synchronized with the target breathing rhythm. The visual guidance signal is presented as a dynamic light bar animation, with the bar length smoothly changing with the breathing rhythm: during inhalation, the light bar fills uniformly from bottom to top, lasting 1.5 to 3 seconds; during exhalation, the light bar recedes from top to bottom, lasting 2 to 4 seconds. The auditory guidance signal consists of beats of different frequencies: an ascending tone of 800 Hz to 1000 Hz is used during inhalation, and a descending tone of 500 Hz to 700 Hz is used during exhalation, with a smooth transition between the two tones within 50 milliseconds. Through the coordinated guidance of both visual and auditory channels, the system effectively helps subjects establish and maintain a standardized breathing rhythm, improving training compliance and effectiveness.

[0130] The user interface management unit integrates and presents various training information. Real-time breathing curves are displayed as waveforms, with a horizontal time axis ranging from 30 to 60 seconds and a vertical amplitude axis automatically adapting to signal strength. The refresh rate reaches 10 to 20 Hz to ensure smoothness. Training progress displays include the number of training sessions completed, the current breathing stability score, and the remaining time to the target. The stability score is converted to a percentage based on the breathing rhythm stability coefficient. Dynamic graphic guidance information intelligently switches according to the status level: a prompt to maintain the current rhythm is displayed in a normal state; a diagram of diaphragmatic breathing techniques is highlighted in a mild abnormal state; and a prominent red rest prompt is displayed in a severe abnormal state. This provides subjects and medical staff with intuitive and comprehensive visualization of training status, significantly improving the safety and effectiveness of training through immediate feedback and clear guidance.

[0131] The preset thresholds, time windows, and measurement ranges described in this invention, such as the first preset threshold, the first time threshold, and the sensor range, are all configurable parameters and are stored in the system's configuration file.

[0132] It should be understood that those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A method for respiratory rehabilitation monitoring and training, characterized in that, Includes the following steps: S1: Simultaneously collect respiratory airflow velocity data, chest and abdominal respiratory movement data, and blood oxygen saturation data of the subjects, and integrate them into the initial monitoring dataset; S2: Clean the initial monitoring dataset to remove signal abrupt changes caused by motion artifacts or equipment interference, and generate a valid monitoring dataset. S3: Based on the effective monitoring dataset, calculate at least one respiratory characteristic parameter that characterizes the stability of the subject's breathing pattern. The respiratory characteristic parameter includes respiratory rate fluctuation value, respiratory depth variation value, and respiratory rhythm stability coefficient. S4: Based on the values ​​of respiratory characteristic parameters and the duration of their abnormality, and in conjunction with blood oxygen saturation data, comprehensively determine the subject's current respiratory rehabilitation status level; among which, the respiratory rehabilitation status level includes at least normal state, mild respiratory dysfunction state, and severe respiratory dysfunction state. S5: Based on the determined respiratory rehabilitation status level, dynamically adjust the working parameters of the respiratory rehabilitation training device. The working parameters include respiratory resistance, the rhythm of visual or auditory guidance signals, and training guidance information displayed in the user interface.

2. The respiratory rehabilitation monitoring and training method according to claim 1, characterized in that, Step S2 involves cleaning the initial monitoring dataset, specifically including: S21: Divide the initial monitoring dataset into continuous data segments according to a preset time window; S22: Perform smoothing filtering on the respiratory airflow velocity data and chest and abdominal respiratory movement data in each data segment, and calculate their local mean. S23: Identify signal points in each data segment that deviate from their local mean by more than a first preset threshold as suspected anomalies; S24: Physiological rationality verification for suspected outliers: If the heart rate data collected synchronously does not change significantly during the time period in which the suspected outlier appears, the suspected outlier is determined to be a non-physiological mutation and is removed. The data is then filled in using the local mean of the data segment or through an interpolation algorithm.

3. The respiratory rehabilitation monitoring and training method according to claim 1, characterized in that, Step S3 calculates respiratory characteristic parameters, specifically including: S31: Extract continuous respiratory cycle time series data from the effective monitoring dataset; the division of respiratory cycles is marked by the inspiratory peak of respiratory airflow velocity, and invalid cycles with significantly abnormal cycle duration due to missed or misjudged detection are automatically removed. S32: Based on respiratory cycle time-series data, calculate the difference between the maximum and minimum respiratory rate within a predetermined time window as the respiratory rate fluctuation value; calculate the difference between the maximum and minimum respiratory depth as the respiratory depth variation value. S33: Calculate the ratio of the standard deviation of the respiratory cycle duration to its mean, and use it as the respiratory rhythm stability coefficient.

4. The respiratory rehabilitation monitoring and training method according to claim 1, characterized in that, Step S4 determines the respiratory rehabilitation status level, specifically as follows: Preset baseline reference ranges corresponding to respiratory characteristic parameters. The baseline reference ranges are set based on the subject's initial measurement values ​​or personal historical data in a resting state. When the respiratory rate fluctuation value exceeds the upper limit of its baseline reference range, and / or the respiratory rhythm stability coefficient is lower than the lower limit of its baseline reference range, and this state continues for the first time threshold, it is judged as a mild respiratory dysfunction state. When the blood oxygen saturation data is detected to be lower than the second preset threshold, regardless of the respiratory characteristic parameters, it is immediately determined to be a severe respiratory dysfunction state.

5. The respiratory rehabilitation monitoring and training method according to claim 4, characterized in that, Step S4 also includes heart rate-assisted determination: Simultaneously monitor the real-time heart rate data of the subjects; When the heart rate data continuously exceeds the third preset threshold and reaches the second time threshold, and at the same time the blood oxygen saturation data is not lower than the second preset threshold, a heart rate increase warning message is pushed to the user interface, but the current training level is maintained. When the heart rate data continuously exceeds the third preset threshold and reaches the second time threshold, and at the same time the blood oxygen saturation data is lower than the second preset threshold, the severe abnormal state is further confirmed and the alarm level is increased.

6. The respiratory rehabilitation monitoring and training method according to claim 1, characterized in that, Step S5 involves dynamically adjusting the operating parameters of the respiratory rehabilitation training device, specifically as follows: When the condition is determined to be normal, maintain or adjust the breathing resistance to the first level and provide the first rhythm breathing guidance signal; When a mild respiratory dysfunction is detected, the breathing resistance is lowered to the second level below the first level, and a breathing guidance signal of the second rhythm, which is slower than the first rhythm, is provided. At the same time, breathing technique guidance information is highlighted in the user interface. When a severe respiratory dysfunction is detected, the respiratory resistance is reduced to zero, the respiratory guidance signal is suspended, and an emergency alarm and rest prompt are displayed in the user interface, while the event is recorded.

7. A respiratory rehabilitation monitoring and training system, applied to implement a respiratory rehabilitation monitoring and training method according to any one of claims 1-6, characterized in that, include: The data acquisition module is used to simultaneously collect physiological data through an airflow sensor, a chest and abdominal respiratory motion sensor, and a blood oxygen sensor; The data preprocessing module, which communicates with the data acquisition module, is used to receive physiological data and perform data cleaning. The feature extraction module, which communicates with the data preprocessing module, is used to calculate respiratory characteristic parameters from the cleaned data; The status decision module, which communicates with the feature extraction module, is used to assign respiratory rehabilitation status levels to subjects based on respiratory feature parameters, blood oxygen saturation, and duration. The control execution module, which communicates with the status decision module, is used to generate control commands based on the respiratory rehabilitation status level to adjust the respiratory rehabilitation training device.

8. The respiratory rehabilitation monitoring and training system according to claim 7, characterized in that... ； The data acquisition module also includes a signal quality detection unit, which is used to evaluate the signal-to-noise ratio of the airflow sensor and the chest and abdominal respiratory motion sensor in real time. When the signal-to-noise ratio is lower than the quality threshold, the user interface is triggered to issue a sensor wearing adjustment prompt.

9. The respiratory rehabilitation monitoring and training system according to claim 7, characterized in that: The state decision module stores a personalized baseline configuration unit, which is used to record the respiratory characteristic parameters of the subject in a quiet state as their personal baseline reference range during system initialization.

10. The respiratory rehabilitation monitoring and training system according to claim 7, characterized in that: The control execution module includes a resistance control unit, a guide signal generation unit, and a user interface management unit; The resistance control unit achieves stepless or stepped adjustment of breathing resistance levels through an electronically controlled valve; the guidance signal generation unit can generate visual light bar animations and auditory beat signals synchronized with the target breathing rhythm. The user interface management unit is used to display real-time breathing curves, training progress, and dynamic graphic guidance information based on the current status level.

Images