A breathing intensity data processing system, method and breathing trainer

Abstract

The application relates to the technical field of data processing, in particular to a breathing intensity data processing system and method and a breathing training device, which are applied to guiding a user to perform inhale and exhale training; the method comprises the following steps: collecting breathing signals of multiple breathing cycles; identifying a starting point and an ending point of each cycle; calculating breathing characteristics such as breathing air time ratio and waveform symmetry through feature extraction logic; and adjusting training feedback based on the breathing characteristics; adopting a topological modeling and derivative analysis mode to correct cycle boundaries, and improving boundary recognition accuracy and training individual adaptability; the breathing training device comprises an air pressure adjusting unit, an atomization unit and a data processing module, and can dynamically optimize training output according to the user state.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a respiratory intensity data processing system, method and respiratory trainer. Background Technology

[0002] Existing respiratory trainers typically issue inhalation and exhalation commands to users at a fixed rhythm, using signals such as respiratory pressure or flow rate as feedback indicators of user compliance. However, in practical applications, these trainers have significant errors in identifying the actual start and end times of a user's breathing, leading to inaccurate respiratory cycle segmentation and distorted respiratory feature extraction results, failing to accurately reflect the user's actual respiratory capacity. Furthermore, current systems lack effective modeling and correction methods for factors such as signal lag and nonlinear fluctuations, often resulting in errors in calculating inhalation and exhalation durations and failure to determine waveform asymmetry, ultimately affecting the accuracy of training feedback and the effectiveness of rehabilitation interventions.

[0003] To address the above issues, this application presents a respiratory intensity data processing system, method, and respiratory trainer. Summary of the Invention

[0004] The technical problem this application aims to solve is to address the shortcomings of existing technologies by providing a respiratory intensity data processing system, method, and respiratory trainer for guiding users in inhalation and exhalation training. The method includes: acquiring respiratory signals from multiple respiratory cycles; identifying the start and end points of each cycle; calculating respiratory features such as the breath duration ratio and waveform symmetry through feature extraction logic; and adjusting training feedback based on these respiratory features. Topological modeling and derivative analysis are used to correct cycle boundaries, improving boundary recognition accuracy and individual adaptability. The respiratory trainer includes a pressure regulation unit, a nebulization unit, and a data processing module, capable of dynamically optimizing training output based on user status.

[0005] To achieve the above objectives, this application provides the following technical solution:

[0006] A method for processing respiratory intensity data, applied to a breathing trainer, the breathing trainer being used to issue inhalation and exhalation commands to a user, each inhalation and exhalation command forming a breathing cycle, the method comprising:

[0007] Acquire respiratory signals from multiple respiratory cycles;

[0008] Based on the respiratory signals, the start and end points of each respiratory cycle are identified, and a feedback signal is obtained;

[0009] The feedback signal is processed by feature extraction logic to obtain respiratory features, wherein the respiratory features include at least one of the respiratory duration ratio and respiratory waveform symmetry.

[0010] The training feedback of the breathing trainer is adjusted based on the breathing characteristics.

[0011] The identification of the start and end points of each respiratory cycle includes:

[0012] For each respiratory cycle, the fluctuation point closest to the inspiratory command time point is taken as the original starting point, and the fluctuation point farthest from the expiratory command time point is taken as the original ending point.

[0013] Based on the signal fluctuations between the original starting point and the original ending point, the original starting point is shifted to the right to obtain the starting point, and the original ending point is shifted to the left to obtain the ending point.

[0014] Based on the signal fluctuations between the original starting point and the original ending point, the original starting point is shifted to the right to obtain the starting point, and the original ending point is shifted to the left to obtain the ending point, including:

[0015] Extract the signal segment between the original starting point and the original ending point, perform topological modeling on the signal segment, map the time and amplitude of the signal segment into a two-dimensional coordinate system, and construct the topological graph of the signal segment;

[0016] Taking the position of the original starting point in the topology graph as the starting point, calculate the first derivative of the nodes within multiple time windows after the starting point, calculate the topological change of the nodes based on the first derivative of the nodes, and determine the first node;

[0017] Taking the position of the original endpoint in the topology graph as the starting point, calculate the first derivative of the nodes within multiple time windows forward from the starting point, calculate the topological change of the nodes based on the first derivative of the nodes, and determine the second node;

[0018] Convergence, connectivity, and fluctuation analysis are performed on the adjacent curves of the first node and the second node to correct the first node and the second node, and to obtain the starting point and the ending point.

[0019] Calculate the topological changes of the nodes based on their first derivatives, including:

[0020] The topological change of a node is calculated based on the first derivative of each node, wherein the topological change is calculated based on the average difference of the first derivatives between the node and its neighboring nodes.

[0021] The topology changes are sorted in descending order, with the node corresponding to the first topology change being designated as the first node and the node corresponding to the last topology change being designated as the second node.

[0022] Convergence, connectivity, and fluctuation analysis are performed on the adjacent curves of the first node and the second node to correct the first node and the second node, obtaining the start point and the end point, including:

[0023] Obtain the right curve of the first node and the left curve of the second node, wherein the left port of the right curve is the first node and the right port is the node corresponding to the first peak value, and the left port of the left curve is the node corresponding to the first valley value and the right port is the second node.

[0024] The corresponding convergence is obtained based on the convergence rates of the left and right curves.

[0025] The connectivity is obtained based on the node connection strength of the left curve and the right curve, wherein the node connection strength is calculated based on the first derivative.

[0026] Based on the volatility of the left and right curves, the corresponding volatility changes are obtained;

[0027] Combining the convergence, connectivity, and fluctuation changes, a correction amount is calculated using a preset weighting function to correct the first node and the second node.

[0028] The feature extraction logic includes:

[0029] Based on the identified start and end points of each respiratory cycle, the inspiratory and expiratory segments within the respiratory cycle are segmented, wherein the left endpoint of the inspiratory segment is the start point and the right endpoint is the maximum point of the feedback signal, and the left endpoint of the expiratory segment is the maximum point of the feedback signal and the right endpoint is the end point.

[0030] Based on the segmentation results, calculate the ratio of the inspiratory segment duration to the expiratory segment duration to obtain the respiratory duration ratio;

[0031] Based on the segmentation results, the symmetry of the respiratory waveform between the inspiratory and expiratory segments is calculated using symmetry metrics, which include peak value, half-peak width, and waveform fit.

[0032] The calculation of the breathing duration ratio includes:

[0033] Based on the inhalation phase, obtain the time difference from the start of inhalation to the end of inhalation, and then obtain the duration from the start of inhalation to the end of inhalation.

[0034] Based on the exhalation phase, obtain the time difference from the start to the end of exhalation to get the duration from the start to the end of exhalation;

[0035] The breathing duration ratio is obtained by comparing the duration from the start of inhalation to the end of inhalation with the duration from the start of exhalation to the end of exhalation.

[0036] The training feedback of the breathing trainer, adjusted according to the breathing characteristics, includes at least one of the following:

[0037] If the ratio of breathing duration to multiple breathing cycles does not meet the preset first threshold, the air pressure regulation unit of the breathing trainer is adjusted.

[0038] If the symmetry of the breathing waveforms corresponding to multiple breathing cycles does not meet the preset second threshold, the nebulization unit of the breathing trainer is adjusted.

[0039] A respiratory intensity data processing system, the system comprising:

[0040] The signal acquisition module is used to acquire respiratory signals from multiple respiratory cycles, identify the start and end points of each respiratory cycle based on the respiratory signals, and generate feedback signals.

[0041] The feature extraction module is used to receive the feedback signal, process the feedback signal through feature extraction logic, and obtain respiratory features, wherein the respiratory features include at least one of the respiratory duration ratio and respiratory waveform symmetry.

[0042] The training feedback adjustment module adjusts the training feedback of the breathing trainer according to the breathing characteristics, wherein the training feedback includes at least adjusting one of the breathing airflow, air pressure and nebulization output intensity.

[0043] A breathing trainer, the breathing trainer comprising:

[0044] A pressure regulating unit, comprising at least an air pump, a pressure sensor, and a valve, wherein the air pump is used to provide airflow, the pressure sensor is used to monitor the pressure of the airflow, and the valve adjusts the air pressure according to feedback requirements;

[0045] The nebulization unit includes a nebulizer and a nebulized drug storage container. The nebulizer is used to convert liquid drugs into atomized particles through airflow, and the atomized particles are delivered to the user's respiratory tract through a connecting tube.

[0046] A connecting tube connects the air pressure regulating unit to the nebulizing unit, so that airflow is delivered from the air pressure regulating unit to the nebulizing unit, and then the nebulizing unit outputs atomized particles to the user's respiratory tract.

[0047] The control unit, connected to the air pressure regulating unit and the nebulizing unit, is used to adjust the airflow, air pressure and nebulization output of the breathing trainer according to the breathing characteristics.

[0048] Compared with the prior art, the beneficial effects of this application are:

[0049] The respiratory intensity data processing method provided in this application can accurately identify the true inspiratory start point and expiratory end point in the respiratory signal through topological modeling and derivative analysis, thereby significantly improving the accuracy of respiratory cycle boundary determination. By introducing topological features such as convergence, connectivity, and fluctuation changes, the initial nodes are corrected in a secondary manner, effectively eliminating misjudgments caused by signal lag, noise disturbance, or physiological inertia, thus improving the stability and robustness of respiratory feature extraction. Furthermore, adjusting the air pressure output and nebulization intensity of the respiratory trainer based on the accurately calculated respiratory duration ratio and waveform symmetry helps to achieve dynamic feedback optimization tailored to the user's condition, enhancing the individualized adaptability and intervention effect of rehabilitation training. Attached Figure Description

[0050] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0051] Figure 1 This is a schematic diagram illustrating the signal hysteresis principle in an embodiment of this application;

[0052] Figure 2 This is a schematic diagram illustrating an exemplary application scenario of an embodiment of this application;

[0053] Figure 3 This is a flowchart illustrating a respiratory intensity data processing method according to an embodiment of this application;

[0054] Figure 4 This is a schematic diagram illustrating the principle of respiratory cycle point correction in an embodiment of this application;

[0055] Figure 5 This is a schematic diagram of the respiratory cycle point correction process in an embodiment of this application. Detailed Implementation

[0056] The technical solutions in 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, and not all embodiments.

[0057] The term "embodiment" as used herein means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0058] This application applies to intelligent respiratory training systems targeting patients in the recovery period and oriented towards assisted training scenarios, especially respiratory trainers that perform non-invasive respiratory feature acquisition and feedback regulation when the user is in a state of confusion, insufficient muscle strength or delayed nerve conduction.

[0059] Application scenarios include, but are not limited to:

[0060] In the scenario of delayed spontaneous breathing training under postoperative sedation, the respiratory response is slow and unstable.

[0061] Intermittent respiratory capacity recovery training in elderly patients or those with neurological impairment exhibits significant periodic fluctuations.

[0062] During the assessment period for weaning from assisted ventilation, it is necessary to extract respiratory characteristic parameters with high precision to determine the degree of recovery;

[0063] In scenarios where breathing intention is weak or coordinated movements are diminished, breathing curve signals are highly susceptible to interference from spurious features.

[0064] It is understandable that the raw signals collected by the breathing trainer are limited by factors such as fluctuations in the patient's active cooperation ability, delays in the response to training instructions, and low breathing amplitude. As a result, the feedback signals often exhibit problems such as unclear signal start and end, unclear period boundaries, and waveform asymmetry. These problems are more pronounced, especially during short-cycle training or shallow breathing.

[0065] It should be noted that the respiratory cycle recognition and feature extraction logic proposed in this application does not rely on a fixed physiological rhythm template or training pattern matching as a premise, nor does it pre-set a specific airway response model for the patient. Instead, it targets the nonlinear waveform signal that is naturally generated during respiratory training, and achieves stable recognition and feedback optimization of respiratory behavior through signal topology reconstruction and boundary correction.

[0066] Taking the ambiguity of signal start and end as an example, please refer to... Figure 1 To understand, Figure 1 This is a schematic diagram illustrating the signal hysteresis principle in an embodiment of this application.

[0067] Figure 1 The diagram shows the signal graph corresponding to the expiratory intensity data collected during a single respiratory cycle, where the signal graph corresponds to the user's postoperative recovery period after general anesthesia.

[0068] like Figure 1As shown, T0 represents the time point when the inhalation command is sent; T1 represents the first small fluctuation point closest to T0 in the collected expiratory intensity data, at which point the actual inhalation process has not yet begun; T2 represents the time point when the expiratory intensity data enters a clear upward trend and begins to fluctuate continuously, which is the actual inhalation start point; T3 represents the time point when the expiratory command is sent; T4 represents the time point when the expiratory intensity data has entered the rapid decline range and is close to the trough, which is the actual exhalation termination point, and effective exhalation has ended; T5 represents the time point when a weak fluctuation appears at the end and gradually flattens out, which can be regarded as the original termination point, reflecting residual exhalation or inertial signals after muscle relaxation.

[0069] It is easy to understand that T1 is delayed relative to T0, and T2 is further delayed than T1, indicating that there is a lag in the user's response from receiving the instruction to the actual start of inhalation. T4 can be regarded as the completion of exhalation, but there is still a tail wave after T4. If the tail wave is used for judgment, taking T5 as the end point will misjudge the end of exhalation.

[0070] It should be noted that, in Figure 1 The respiratory intensity signals shown exhibit significant blurring or misalignment at the beginning and end, which is due to the following four reasons:

[0071] Firstly, during the postoperative recovery period or under sedation, the user's central nervous system exhibits delayed cognitive and neuromuscular responses to external respiratory commands. Especially when sedative drugs have not been fully metabolized or the user is confused, the user often cannot immediately coordinate actions after the command is given, resulting in a significant delay in inspiratory response.

[0072] Secondly, during the postoperative recovery period, users have not fully recovered their ability to control their breathing, especially before the nervous system has completed the coordinated scheduling of the main respiratory muscle groups such as the diaphragm and intercostal muscles. Some inspiratory behaviors exhibit uncontrolled passive inspiratory phenomena.

[0073] In other words, slight airflow fluctuations in the chest cavity caused by factors such as muscle tension imbalance or passive lung recoil, while appearing as a slight increase in the signal, do not signify the actual start of inspiration. Therefore, after the inspiration command is issued, the signal may initially show a small disturbance point driven by involuntary breathing, followed by a significant and sustained upward phase controlled by voluntary breathing. Mistaking this disturbance point for the start of inspiration can easily lead to a shift in cycle recognition.

[0074] Thirdly, the intensity sensing element of a breathing trainer is often deployed at the end of the mask, connecting tubing, or cavity. Due to factors such as structural length, tubing lag, and airflow inertia, there is a transmission delay in the signal along the sensing path. Especially at the end of expiration, the slow decay of airflow and the inertial backflow caused by muscle relaxation can easily manifest as a tail wave in the signal.

[0075] Fourthly, postoperative patients experience some degree of respiratory muscle fatigue or functional asymmetry, resulting in nonlinear execution force and rate of inhalation and exhalation. Especially in the early stages of training, breathing behavior exhibits irregular waveforms with significant differences between cycles, making it difficult to directly cut the original start and end points using conventional logic.

[0076] Please see Figure 2 This figure is a schematic diagram of an exemplary application scenario provided by an embodiment of this application.

[0077] Figure 2 The scenario shown includes a user terminal and a breathing trainer terminal, wherein the breathing trainer terminal issues inhalation and exhalation commands to the user in sequence, processes the breathing intensity data during the user's execution of the commands, and thereby adjusts the training feedback of the breathing trainer.

[0078] Figure 2 The user terminal is further shown to include a data acquisition module, which is used to acquire the breathing intensity signal generated by the user after receiving an inhalation or exhalation command, and transmit the signal to the breathing trainer terminal for analysis and processing.

[0079] Figure 2 The breathing trainer end is further shown to include:

[0080] The instruction issuing module is used to issue inhalation or exhalation instructions to the user according to a preset training rhythm in order to guide the user to complete breathing training.

[0081] The data processing module is used to receive and process respiratory intensity data from the data acquisition module, identify the start and end points of each respiratory cycle, and extract respiratory features, including but not limited to the respiratory duration ratio and waveform symmetry.

[0082] The training feedback module is used to determine whether the user has abnormal or asymmetrical breathing behavior based on the extracted breathing features, and to adjust the rhythm, difficulty or prompts of subsequent training instructions based on the judgment result.

[0083] Next, combined Figure 3 This application will introduce a respiratory intensity data processing method provided in its embodiments. Figure 3 The method shown is applied to a breathing trainer, which issues inhalation and exhalation commands to the user. Each inhalation and exhalation command forms a breathing cycle. The specific steps are as follows:

[0084] S1: Acquire respiratory signals from multiple respiratory cycles;

[0085] In this embodiment, the respiratory signal is acquired in real time by an installed data acquisition device. The respiratory signal can be in the form of a pressure signal, and it has the characteristics of temporal continuity and amplitude variation, reflecting the intensity of the user's inhalation and exhalation in each cycle.

[0086] To ensure the accuracy of subsequent processing, in an optional embodiment, a high-frequency data sampling method is used to capture the complete periodic segment, and the acquired data is preprocessed as necessary, including but not limited to noise reduction, normalization and filtering operations.

[0087] It is understood that the data acquisition device can be a pressure sensor, and its installation location can be in the breathing pathway, training mask, or training tubing. The specific structure and parameters of the data acquisition device can be adjusted according to the usage scenario, and this application does not limit this. As long as the data acquisition device can output a pressure change signal reflecting the user's breathing intensity in real time, it can be applied to the processing flow described in this application.

[0088] S2: Based on the respiratory signal, identify the start and end points of each respiratory cycle and obtain a feedback signal;

[0089] In this embodiment, the breathing trainer retrieves candidate start and end points from the signal in each cycle based on the time points corresponding to the inhalation and exhalation commands. Considering the potential for delayed respiratory response in postoperative patients or those in a state of confusion, this embodiment does not directly use the first fluctuation point of the signal as the starting point, nor the end point of the tail wave as the ending point. Instead, it analyzes the candidate signal segments by constructing a topology map and uses the first derivative and the topological change trend between adjacent nodes to identify boundaries within local regions.

[0090] Specifically, the original starting point is slid backward through multiple time windows to identify the first node with the greatest degree of topological mutation; similarly, the original ending point is slid forward to identify the second node, and the connectivity, volatility and convergence indices of its left and right curves are used for correction and adjustment, and finally the starting point and ending point are determined.

[0091] S3: The feedback signal is processed by feature extraction logic to obtain respiratory features, wherein the respiratory features include at least one of the respiratory duration ratio and respiratory waveform symmetry;

[0092] In this embodiment, each respiratory cycle in the feedback signal is divided into an inspiratory segment and an expiratory segment. The inspiratory segment is defined as the curve interval between the starting point and the signal maximum point, and the expiratory segment is the interval between the maximum point and the ending point. Based on this, the following two dimensions of respiratory characteristics are calculated:

[0093] The respiratory duration ratio is calculated by separately counting the durations of the inspiratory and expiratory phases and then calculating the ratio between the two.

[0094] The symmetry of the respiratory waveform is measured by structural fitting of parameters such as waveform peak position, half-peak width, rise or fall rate, and left and right area distribution, to measure the morphological differences between the inspiratory and expiratory phases.

[0095] S4: Adjust the training feedback of the breathing trainer according to the breathing characteristics;

[0096] In this embodiment, the presence of significant asymmetry in the respiratory rhythm is determined based on the extracted respiratory duration ratio and respiratory waveform symmetry parameters. When an abnormal rhythm is identified, the breathing trainer automatically adjusts the subsequent training rhythm or difficulty, including but not limited to:

[0097] Delaying the next command transmission time gives users more time to prepare and recover;

[0098] Reduce training load, such as by reducing inspiratory resistance or shortening the duration of each inhalation;

[0099] Enable real-time voice / image prompts to guide users to strengthen control over the inhalation or exhalation process;

[0100] It outputs trend forecast information and provides suggestions for the target range for the next period.

[0101] Before detailing the specific technical aspects corresponding to the steps, this application's embodiments need to reiterate:

[0102] In practical applications, the respiratory intensity signal collected by a breathing trainer does not always respond synchronously with the moment the command is issued. This is especially true in typical scenarios such as rehabilitation training, postoperative recovery, or use by the elderly, where the user's physiological response exhibits significant time lag and behavioral fluctuations. This lag is not a simple time delay, but rather a multi-stage, discontinuous response caused by the asynchronous control of the nervous system and respiratory muscles. Its signal curve often resembles the following:

[0103] In the initial stage after the command is issued, the signal changes very little, followed by a short period of weak fluctuations, before entering a stable rising or falling phase, eventually reaching the peak of inhalation or the trough of exhalation. Such fluctuations not only blur the start and end points, but also significantly reduce the accuracy of traditional algorithms that rely solely on threshold judgments or fixed time windows to divide the cycle.

[0104] The method proposed in this application does not rely on abrupt changes at fixed time points or with set amplitudes as the boundaries of the respiratory cycle. Instead, it extracts signal nodes that truly represent the user's effective breathing behavior by continuously modeling signal change trends, spatial mapping of topological maps, and analyzing the local evolutionary behavior of fluctuation clusters. Especially at the two critical moments of inhalation initiation and exhalation end, considering that signal fluctuations are often in an unstable state, this application uses the directional inflection point caused by the change in signal derivative as the first identification dimension, and further uses the convergence, connectivity, and volatility of adjacent curve segments as correction references to finally dynamically determine the start and end points of each cycle.

[0105] It should be noted that this application does not use the maximum or minimum values ​​of respiratory signals, significant peaks and troughs, or predefined time windows as the basis for judgment. Instead, it adopts an extension mapping model based on behavioral trends, dividing the period based on small but continuous changes in the fluctuation curve. This effectively avoids misjudgment of the period caused by delayed command response or tailing of terminal signals, ensuring the stability and personalized adaptability of training feedback parameters.

[0106] Next, we will further elaborate on the part of the method in this application regarding the identification of the start and end points.

[0107] refer to Figure 4 , Figure 4 This is a schematic diagram illustrating the principle of respiratory cycle point correction in an embodiment of this application, where the arrows indicate the correction process.

[0108] In one example, identifying the start and end points of each respiratory cycle includes:

[0109] S2.1: For the respiratory signal corresponding to each respiratory cycle, the fluctuation point closest to the inhalation command time point is taken as the original starting point, and the fluctuation point farthest from the exhalation command time point is taken as the original ending point.

[0110] Specifically, since users do not immediately produce significant breathing actions after receiving inhalation or exhalation commands, and the actual signal fluctuations have a certain delay, directly using the command time point as the boundary of period segmentation can easily lead to inaccurate period recognition, thereby affecting the effectiveness of subsequent feature extraction and the stability of feedback adjustment.

[0111] In this embodiment, based on the time point of each inhalation command, the first non-stationary fluctuation point in the respiratory signal is sequentially searched from the subsequent sequence, and this point is taken as the original starting point. The original starting point is usually the turning point where the signal tends to change slightly from a static state. The identification of fluctuation points is based on the calculation of the local change trend of continuously sampled data. The judgment criteria may include the signal difference between adjacent time windows exceeding the static threshold or the gradient exceeding the reference background level.

[0112] Furthermore, after each exhalation command, considering the multiple weak fluctuations that appear at the end of the actual signal curve, the point of maximum fluctuation that is farthest from the exhalation command time but still within the cycle is extracted by reverse traversal using local fluctuation energy or delayed time windows, and this point is used as the original endpoint. The original endpoint is usually located at the disturbance point before the signal in the tail end of exhalation tends to stabilize, which may be due to residual airflow or inertial feedback caused by relaxation of respiratory muscles. If it is directly used as the cycle endpoint, it will interfere with the accurate judgment of the exhalation duration, so further correction is required.

[0113] S2.2: Based on the signal fluctuation between the original starting point and the original ending point, the original starting point is shifted to the right to obtain the starting point, and the original ending point is shifted to the left to obtain the ending point;

[0114] Specifically, this step is used to further correct the original start point and original end point extracted by S2.1. Since the original start point usually reflects the initial disturbance when the user begins to react but has not yet entered true inhalation, and the original end point often contains non-functional wake waves, if they are not corrected a second time, invalid segments will be included before and after the period division, affecting the accuracy of the characteristic quantity calculation.

[0115] In this embodiment, the correction process uses the original starting point and the original ending point as reference windows, extracts the signal segment between them, and performs local topological modeling on this segment. By mapping this segment to a two-dimensional coordinate graph, local waveform structure changes are identified. Multiple windows are slid backward from the original starting point on the time axis, and the first derivative of each node is calculated to characterize the signal change rate. Based on this, the key node where the signal curve transitions from an inactive state to a continuously rising state is identified, and this node is defined as the actual starting point. Similarly, by sliding a window forward to the left of the original ending point, the first derivative of the descending segment signal is dynamically analyzed. Combining features such as derivative continuity, curve convergence trend, and slope change rate, the functional termination point of the exhalation process, i.e., the termination point, is located.

[0116] refer to Figure 5 , Figure 5 This is a schematic diagram of the respiratory cycle point correction process in an embodiment of this application.

[0117] In one example, the specific steps of S2.1 are as follows:

[0118] S2.1.1: Extract the signal segment between the original starting point and the original ending point, perform topological modeling on the signal segment, map the time and amplitude of the signal segment into a two-dimensional coordinate system, and construct the topological graph of the signal segment;

[0119] Specifically, analyzing respiratory signals directly based on time-amplitude sequences, i.e., traditional signal maps, often only yields linear or one-dimensional information corresponding to local trends. It fails to capture the spatial continuity, dynamic evolution, and hidden transitional structures of the signal near its start and end points from an overall structural perspective. Furthermore, due to the inherent nonlinear and multi-scale characteristics of respiratory signals, traditional one-dimensional signal processing suffers from ambiguity in boundary identification.

[0120] It is understandable that differences in users' physiological states mean that the inhalation or exhalation process does not occur instantaneously at a specific point in time, but rather unfolds gradually, and may even exhibit passive fluctuations after the command is issued. Therefore, relying solely on the amplitude or local slope on the signal graph can easily lead to misjudgment of the starting point or over-reliance on threshold settings, lacking a global understanding of complex boundary structures.

[0121] In this embodiment, the construction of the topology map is based on the time interval determined by the original start and end points in the respiratory signal, and the signal segments are extracted and re-encoded into a topology structure.

[0122] In an optional embodiment, each sampling point in the signal segment is first represented as a topological node. Each node is mapped to the horizontal and vertical coordinates of the topological graph based on its corresponding timestamp and signal amplitude. Unlike traditional signal graphs, in the topological graph, each node not only forms a time series chain with the nodes before and after it, but also establishes extended connections through nonlinear characteristic relationships, thus forming a topological graph structure.

[0123] For example, the derivative between adjacent time points can be used to measure the rate of change, while the sequence of derivatives between multiple points can be further used to construct a change direction map, forming a local geometric curvature subgraph of the fluctuation path.

[0124] S2.1.2: Taking the position of the original starting point in the topology graph as the starting point, calculate the first derivative of the nodes within multiple time windows after the starting point, calculate the topological change of the nodes based on the first derivative of the nodes, and determine the first node;

[0125] Specifically, as described above, the original starting point is usually not the actual start point of inhalation, but rather the initial signal position caused by a delayed response or muscle inertia. The effective initiation phase of the breathing process only exists after the original starting point.

[0126] In this embodiment, the first derivative is used as a criterion to identify signal change trends. The first derivative reflects the rate at which signal strength changes over time; its positive and negative values ​​represent rising and falling trends, respectively, and the magnitude of the value represents the degree of fluctuation. In respiratory trainer applications, the actual inhalation start point often appears in the region where the derivative begins to be consistently positive and gradually increases. Therefore, sliding the window backward from the original starting point and calculating the derivative value point by point can help identify where the signal trend changes significantly.

[0127] Furthermore, to prevent misjudgments caused by accidental noise interference, this embodiment does not use single-point derivative values ​​as the basis for judgment. Instead, it calculates the average derivative change through multiple time windows. Specifically, each time window contains multiple consecutive nodes, and the average value of the derivative within that time window and the difference in derivative change between that time window and the previous time window are calculated. When multiple consecutive time windows show derivative increments in the same direction, it can be determined that there is a significant trend transition in this region, serving as an indicator of topological change.

[0128] In one example, the topological change of a node is calculated based on the first derivative of each node, wherein the topological change is calculated based on the average difference of the first derivatives between the node and its neighboring nodes.

[0129] Furthermore, the topology changes are sorted in descending order, and the node corresponding to the first topology change is taken as the first node.

[0130] S2.1.3: Taking the position of the original endpoint corresponding to the position in the topology graph as the starting point, calculate the first derivative of the nodes within multiple time windows forward from the starting point, calculate the topological change of the nodes based on the first derivative of the nodes, and determine the second node;

[0131] Similarly, the original endpoint is usually not the actual end point of exhalation, but rather the location corresponding to the tail wave, descent phase, or residual signal fluctuation point after muscle relaxation that occurs during the acquisition process. Since the end of a user's exhalation may not mean the signal stops immediately, but rather that physiological or structural factors such as delayed release in the respiratory tract, chest recoil, or delayed exhaust in the expiratory tubing can cause slight fluctuations in the signal even after the physical end of exhalation, using only the last fluctuation point of the signal curve as the cycle endpoint often leads to misjudgment, failing to accurately identify the boundary of the truly effective expiratory segment, and affecting the subsequent calculation of expiratory duration and waveform symmetry.

[0132] In this embodiment, to eliminate the interference of the aforementioned lag factors, a method similar to the intake segment processing method is adopted. That is, by sliding the analysis forward from the original endpoint in the topology graph, the key change points before the signal ends are identified as the second nodes. Unlike the analysis backward from the original starting point in S2.1.2, the analysis direction here is forward. That is, starting from the original endpoint, multiple time windows are divided forward, and the first derivative of the node is calculated window by window, and the change characteristics of the derivative behavior are extracted.

[0133] Understandably, the physical meaning of the first derivative in this stage still reflects the rate of change in respiratory signal intensity. During a typical exhalation, the signal exhibits a clear downward trend, at which point the derivative is negative. After effective exhalation ends, it enters a phase of weak fluctuations or a flat signal, where the derivative approaches zero but no longer undergoes significant and drastic changes. Therefore, the true end point of exhalation often does not correspond to a jump point of drastic change, but rather falls in the region of minimal change in the derivative—that is, a smooth ending phase where the derivative value is relatively stable, fluctuations are gradual, and the trend of change weakens.

[0134] Furthermore, the topological change for each node is calculated. This topological change reflects the intensity of the disturbance in the derivative trend, i.e., the degree of abrupt change in the derivative. Unlike the inhalation phase, the goal of the exhalation phase is not to find abrupt change points, but rather to find the final segment where the derivative change is minimal and tends to stabilize.

[0135] It's easy to understand that among all the topological changes of nodes, the smaller the value, the closer the derivative of the node is to a steady state, and the more likely it is to be in the flat range at the end of exhalation, thus better matching the termination characteristics of the actual exhalation action. Therefore, after sorting the topological changes in descending order, the node at the end has the smallest topological change, reflecting that the node is in the position with the least drastic signal fluctuation, which is a reasonable candidate for an effective exhalation termination point.

[0136] S2.1.4: Perform convergence, connectivity, and fluctuation analysis on the adjacent curves of the first node and the second node to correct the first node and the second node, and obtain the starting point and the correction point;

[0137] Specifically, the first and second nodes are key points initially identified based on the trend of signal derivatives and topological changes. While they provide reference boundaries, in actual respiratory data, signals often exhibit non-ideal fluctuations, such as sudden spikes, plateaus, or brief rebounds, which may affect the accuracy of single-node determination. Therefore, to further improve the stability and adaptability of boundary identification, this step introduces topological feature analysis of signal segments. Taking the curve segments adjacent to the first and second nodes as the object, the boundary nodes are further optimized by comprehensively considering three indicators: convergence, connectivity, and fluctuation changes, thereby obtaining cycle start and end points that more closely match the actual physiological process.

[0138] In this embodiment, the first node and its adjacent curve segment to the right, and the second node and its adjacent curve segment to the left, are respectively used as analysis targets. The curve segments are defined as the signal interval between the first node and its first subsequent maximum, and the signal interval between the second node and its first preceding minimum. This can cover the complete process segments of the inspiratory ascent and expiratory descent.

[0139] Furthermore, based on the aforementioned curve segment, the following three topological features are calculated:

[0140] The first topological characteristic is convergence. In a curve segment, the degree to which the signal trend tends towards a fixed direction is its convergence. For example, in an intake segment, if the signal continues to rise after the first node and the fluctuation amplitude gradually weakens, it indicates that the signal is converging towards the intake peak and has a good intake trend; conversely, it may still be in the early stage of fluctuation and cannot be considered a true starting point. By calculating the consistency of the derivatives within this segment, the continuity of the derivative sign, and whether the amplitude of change converges to a certain trend, the trend certainty of this segment can be evaluated.

[0141] Specifically, the convergence rate can be calculated by statistically analyzing the proportion of continuous positive derivatives in the segment, the variance of derivative changes, and the slope of the first derivative of the signal curve. The specific convergence rate can be calculated by weighted summation of the aforementioned factors, which will not be elaborated upon here.

[0142] The second topological feature is connectivity, which reflects the continuity and trend consistency of curves between a node and its adjacent signal segments. In real signals, if a node and its subsequent curves exhibit abrupt changes or discontinuities in the derivative direction or numerical trend, it is likely an outlier or a misjudged point and is unsuitable as a periodic boundary. Therefore, when determining connectivity, it is necessary not only to evaluate the amplitude difference between the node value and its adjacent nodes but also to assess whether the trend change of its derivative is smooth.

[0143] Specifically, the derivative connectivity between a node and its neighboring curve segments is used as the node's connectivity strength. If the derivative changes between a node and its next (or previous) n points are continuous and the trend direction is consistent, then the connectivity can be considered good; otherwise, the boundary position needs to be adjusted.

[0144] Understandably, the node connection strength is a Boolean parameter. When, within a preset sliding window, the first derivative of a node and its n neighboring points are consistent in numerical direction, and the continuously changing derivative values ​​do not undergo significant reversals or drastic jumps within the allowable disturbance tolerance range, the node connection strength is True, indicating that the node has good connectivity with its neighboring segments. Conversely, if there are multiple derivative sign jumps or derivative increment fluctuations exceeding a set threshold, the trend is considered discontinuous or the structure is broken, and the node connection strength is False, indicating that the node has weak connectivity.

[0145] Those skilled in the art will understand that connectivity, as part of the node correction weight calculation, helps determine whether to adjust the positions of the first and second nodes, thereby improving the stability and rationality of the determination of the start and end points of the cycle. Correction is only required when the node connection strength corresponding to connectivity is true.

[0146] The third topological feature is fluctuation variation, which is used to assess whether the signal contains high-frequency jitter or unstable components within that segment. Respiratory signals typically exhibit greater stability during the initial stages of inspiration and expiration, while areas with uncertain boundaries may show severe noise fluctuations. Therefore, if the signal fluctuation rate is high in this segment, the original boundary point may mistakenly contain non-target signals, requiring appropriate translation.

[0147] Specifically, the calculation method includes short-time energy analysis, variance analysis, and sliding window range comparison of the curve segment to obtain the volatility, which will not be elaborated here.

[0148] Combining the convergence, connectivity, and fluctuation changes, a correction amount is calculated using a preset weighting function to correct the first node and the second node;

[0149] Specifically, each feature index is converted into a standardized numerical form. For example, convergence is mapped to a value between 0 and 1 using a derivative continuity score; fluctuation is processed by the reciprocal of the normalized signal variance; and connectivity Boolean values ​​directly determine whether correction is needed. When correction is required, the first two are linearly weighted using a set of empirically determined weights or weights learned from training samples. These weights can be flexibly configured according to different application scenarios. For example, if the system prioritizes signal trend stability, the convergence weight can be appropriately increased; if the target population has significant respiratory delays, the influence weight of the fluctuation score can be increased.

[0150] Furthermore, the calculation results are used as correction values, and the nodes are offset according to these correction values.

[0151] In one example, the specific steps of S3 are as follows:

[0152] S3.1: Based on the start and end points of each identified respiratory cycle, the inspiratory and expiratory segments within the respiratory cycle are segmented, wherein the left endpoint of the inspiratory segment is the start point and the right endpoint is the maximum point of the feedback signal, and the left endpoint of the expiratory segment is the maximum point of the feedback signal and the right endpoint is the end point.

[0153] Specifically, the goal of this step is to further refine the structural features within the breathing process based on the identification results of the cycle boundaries, subdividing a complete breathing cycle into two functional segments: inhalation and exhalation. This provides an accurate structural foundation for subsequent duration ratio analysis and waveform feature extraction. In actual breathing, the inhalation segment is usually accompanied by an increase in air pressure, while the exhalation segment is accompanied by a decrease in air pressure. Therefore, the entire feedback signal exhibits a waveform structure of first rising and then falling. The start and end points define the cycle boundaries, and the maximum point usually corresponds to the peak value reached during inhalation, possessing clear physical representativeness.

[0154] In this embodiment, by traversing the signal data between the start and end points, the global maximum value is found as the end point of the inhalation process. The end point corresponding to the global maximum value not only has a stable shape in the signal curve but also exhibits good repeatability across multiple cycles. Based on this structure, the inhalation segment is defined from the start point to the maximum value point, and the exhalation segment is defined from the maximum value point to the end point, thereby achieving accurate segmentation of the breathing process into different stages.

[0155] S3.2: Based on the segmentation results, calculate the ratio of the inspiratory segment duration to the expiratory segment duration to obtain the respiratory duration ratio;

[0156] Specifically, the core of this step lies in extracting quantitative indicators that reflect the user's respiratory rhythm and functional status from the segmented structure, namely the respiratory duration ratio. The respiratory duration ratio is the ratio between the duration of the inspiratory phase and the duration of the expiratory phase, and it is often used to assess medical and sports indicators such as respiratory coordination, lung recovery, or training response intensity.

[0157] In this embodiment, based on the inspiratory and expiratory segments obtained in step S3.1, their start and end times are recorded, their respective durations are calculated, and then their ratio is determined. For example, the duration of the inspiratory segment is the time difference from the starting point to the maximum point, and the duration of the expiratory segment is the time difference from the maximum point to the end point; the ratio of the two yields the respiratory duration ratio. Since the speed, depth, and rhythm of inhalation and exhalation may fluctuate in each cycle, this ratio exhibits periodic variation characteristics, making it suitable as a parameter for long-term trend tracking or real-time training adjustment.

[0158] In one example, the calculation of the breathing duration ratio includes:

[0159] Based on the inhalation phase, obtain the time difference from the start of inhalation to the end of inhalation, and then obtain the duration from the start of inhalation to the end of inhalation.

[0160] Based on the exhalation phase, obtain the time difference from the start to the end of exhalation to get the duration from the start to the end of exhalation;

[0161] The breathing duration ratio is obtained by comparing the duration from the start of inhalation to the end of inhalation with the duration from the start of exhalation to the end of exhalation.

[0162] S3.3: Based on the segmentation results, calculate the symmetry of the respiratory waveform between the inspiratory and expiratory segments using symmetry metrics, wherein the symmetry metrics include peak value, half-peak width, and waveform fit.

[0163] Specifically, the symmetry in the signal morphology of the inspiratory and expiratory phases reflects the coordination and balance of respiratory movements, and is an important indicator for measuring the user's neuromuscular control ability, the progress of lung function recovery, and respiratory fatigue. An ideal respiratory waveform should exhibit a certain degree of mirror symmetry in shape, meaning that the peaks and troughs formed by inhalation and exhalation are relatively evenly distributed, and the steepness, duration, slope, and peak position of the curve should be symmetrical within a certain range.

[0164] In this embodiment, to quantify this symmetry, the following metrics are used as evaluation dimensions:

[0165] First, extract the maximum amplitude of each of the inhalation and exhalation phases, and assess whether their relative amplitudes are close.

[0166] Secondly, calculate the half-peak width, which is the time width corresponding to the curve reaching half of its maximum amplitude. If the difference between the half-peak widths of the inhalation and exhalation segments is too large, it indicates that there is an inconsistency in the rate of change between the two segments.

[0167] Finally, by fitting the two segments with polynomial or Gaussian functions respectively, and then calculating their fitting error and overlap area, we can evaluate whether the overall waveform shape is approximately mirror-symmetric in space.

[0168] Furthermore, to improve the stability of symmetry measurement, a moving average calculation can be performed after each respiratory cycle to construct a time-stable symmetry trend curve. This curve is then used by the upper-level module to determine whether the breathing pattern has entered a steady state or deviated from normal. If the symmetry score in a certain cycle is below the threshold and persists for multiple cycles, it can serve as a trigger condition for adjusting the feedback strategy, prompting the trainer to extend the expiratory guidance time or reduce the inspiratory load, thereby achieving real-time assisted intervention.

[0169] The training feedback of the breathing trainer, adjusted according to the breathing characteristics, includes at least one of the following:

[0170] If the ratio of breathing duration to multiple breathing cycles does not meet the preset first threshold, the air pressure regulation unit of the breathing trainer is adjusted.

[0171] If the symmetry of the breathing waveforms corresponding to multiple breathing cycles does not meet the preset second threshold, the nebulization unit of the breathing trainer is adjusted.

[0172] It is understood that the first and second thresholds can be set based on experience, and how to adjust them specifically is a technical matter that can be understood by those skilled in the art, including but not limited to:

[0173] By increasing the assistive air pressure intensity during the inspiratory phase, extending the duration of inspiratory guidance, and increasing the inspiratory guidance airflow response delay, users can be stimulated to actively inhale or their pressure sensitivity during the inspiratory phase can be improved, thereby guiding them to form a more balanced inspiratory and exhalation rhythm in subsequent training.

[0174] By changing the nebulizer particle size, adjusting the nebulization time window to align with the inspiratory phase, and increasing the delayed output of nebulized volume at the end of expiration, nebulization can be used to enhance the sense of respiratory flow or provide rhythmic guidance signals to help users perceive and adjust their own inhalation and exhalation rhythm, thereby gradually optimizing its waveform morphology and achieving a more natural respiratory coordination.

[0175] In one example, this application provides a respiratory intensity data processing system, the system comprising:

[0176] The signal acquisition module is used to acquire respiratory signals from multiple respiratory cycles, identify the start and end points of each respiratory cycle based on the respiratory signals, and generate feedback signals.

[0177] The feature extraction module is used to receive the feedback signal, process the feedback signal through feature extraction logic, and obtain respiratory features, wherein the respiratory features include at least one of the respiratory duration ratio and respiratory waveform symmetry.

[0178] The training feedback adjustment module adjusts the training feedback of the breathing trainer according to the breathing characteristics, wherein the training feedback includes at least adjusting one of the breathing airflow, air pressure and nebulization output intensity.

[0179] In one example, this application provides a breathing trainer, the breathing trainer comprising:

[0180] A pressure regulating unit, comprising at least an air pump, a pressure sensor, and a valve, wherein the air pump is used to provide airflow, the pressure sensor is used to monitor the pressure of the airflow, and the valve adjusts the air pressure according to feedback requirements;

[0181] The nebulization unit includes a nebulizer and a nebulized drug storage container. The nebulizer is used to convert liquid drugs into atomized particles through airflow, and the atomized particles are delivered to the user's respiratory tract through a connecting tube.

[0182] A connecting tube connects the air pressure regulating unit to the nebulizing unit, so that airflow is delivered from the air pressure regulating unit to the nebulizing unit, and then the nebulizing unit outputs atomized particles to the user's respiratory tract.

[0183] The control unit, connected to the air pressure regulating unit and the nebulizing unit, is used to adjust the airflow, air pressure and nebulization output of the breathing trainer according to the breathing characteristics.

[0184] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A method for processing respiratory intensity data, applied to a breathing trainer, characterized in that, The breathing trainer is used to issue inhalation and exhalation commands to the user, each inhalation and exhalation command forming a breathing cycle, and the method includes: Acquire respiratory signals from multiple respiratory cycles; Based on the respiratory signals, the start and end points of each respiratory cycle are identified, and a feedback signal is obtained; The feedback signal is processed by feature extraction logic to obtain respiratory features, wherein the respiratory features include at least one of the respiratory duration ratio and respiratory waveform symmetry. The training feedback of the breathing trainer is adjusted according to the breathing characteristics. The identification of the start and end points of each respiratory cycle includes: For each respiratory cycle, the fluctuation point closest to the inspiratory command time point is taken as the original starting point, and the fluctuation point farthest from the expiratory command time point is taken as the original ending point. Based on the signal fluctuations between the original starting point and the original ending point, the original starting point is shifted to the right to obtain the starting point, and the original ending point is shifted to the left to obtain the ending point. Based on the signal fluctuations between the original starting point and the original ending point, the original starting point is shifted to the right to obtain the starting point, and the original ending point is shifted to the left to obtain the ending point, including: Extract the signal segment between the original starting point and the original ending point, perform topological modeling on the signal segment, map the time and amplitude of the signal segment into a two-dimensional coordinate system, and construct the topological graph of the signal segment; Taking the position of the original starting point in the topology graph as the starting point, calculate the first derivative of the nodes within multiple time windows after the starting point, calculate the topological change of the nodes based on the first derivative of the nodes, and determine the first node; Taking the position of the original endpoint in the topology graph as the starting point, calculate the first derivative of the nodes within multiple time windows forward from the starting point, calculate the topological change of the nodes based on the first derivative of the nodes, and determine the second node; Convergence, connectivity, and fluctuation analysis are performed on the adjacent curves of the first node and the second node to correct the first node and the second node, and to obtain the starting point and the ending point. The training feedback of the breathing trainer, adjusted according to the breathing characteristics, includes at least one of the following: If the ratio of breathing duration to multiple breathing cycles does not meet the preset first threshold, the air pressure regulation unit of the breathing trainer is adjusted. If the symmetry of the breathing waveforms corresponding to multiple breathing cycles does not meet the preset second threshold, the nebulization unit of the breathing trainer is adjusted.

2. The respiratory intensity data processing method according to claim 1, characterized in that, Calculate the topological changes of the nodes based on their first derivatives, including: The topological change of a node is calculated based on the first derivative of each node, wherein the topological change is calculated based on the average difference of the first derivatives between the node and its neighboring nodes. The topology changes are sorted in descending order, with the node corresponding to the first topology change being designated as the first node and the node corresponding to the last topology change being designated as the second node.

3. The respiratory intensity data processing method according to claim 1, characterized in that, Convergence, connectivity, and fluctuation analysis are performed on the adjacent curves of the first node and the second node to correct the first node and the second node, obtaining the start point and the end point, including: Obtain the right curve of the first node and the left curve of the second node, wherein the left port of the right curve is the first node and the right port is the node corresponding to the first peak value, and the left port of the left curve is the node corresponding to the first valley value and the right port is the second node. The corresponding convergence is obtained based on the convergence rates of the left and right curves. The connectivity is obtained based on the node connection strength of the left curve and the right curve, wherein the node connection strength is calculated based on the first derivative. Based on the volatility of the left and right curves, the corresponding volatility changes are obtained; Combining the convergence, connectivity, and fluctuation changes, a correction amount is calculated using a preset weighting function to correct the first node and the second node.

4. The respiratory intensity data processing method according to claim 1, characterized in that, The feature extraction logic includes: Based on the identified start and end points of each respiratory cycle, the inspiratory and expiratory segments within the respiratory cycle are segmented, wherein the left endpoint of the inspiratory segment is the start point and the right endpoint is the maximum point of the feedback signal, and the left endpoint of the expiratory segment is the maximum point of the feedback signal and the right endpoint is the end point. Based on the segmentation results, calculate the ratio of the inspiratory segment duration to the expiratory segment duration to obtain the respiratory duration ratio; Based on the segmentation results, the symmetry of the respiratory waveform between the inspiratory and expiratory segments is calculated using symmetry metrics, which include peak value, half-peak width, and waveform fit.

5. The respiratory intensity data processing method according to claim 4, characterized in that, The calculation of the breathing duration ratio includes: Based on the inhalation phase, obtain the time difference from the start of inhalation to the end of inhalation, and then obtain the duration from the start of inhalation to the end of inhalation. Based on the exhalation phase, obtain the time difference from the start to the end of exhalation to get the duration from the start to the end of exhalation; The breathing duration ratio is obtained by comparing the duration from the start of inhalation to the end of inhalation with the duration from the start of exhalation to the end of exhalation.

6. A respiratory intensity data processing system, used to implement the respiratory intensity data processing method as described in any one of claims 1-5, characterized in that, The system includes: The signal acquisition module is used to acquire respiratory signals from multiple respiratory cycles, identify the start and end points of each respiratory cycle based on the respiratory signals, and generate feedback signals. The feature extraction module is used to receive the feedback signal, process the feedback signal through feature extraction logic, and obtain respiratory features, wherein the respiratory features include at least one of the respiratory duration ratio and respiratory waveform symmetry. The training feedback adjustment module adjusts the training feedback of the breathing trainer according to the breathing characteristics, wherein the training feedback includes at least adjusting one of the breathing airflow, air pressure and nebulization output intensity.

7. A breathing trainer for implementing a breathing intensity data processing method as described in any one of claims 1-5, characterized in that, The breathing trainer includes: A pressure regulating unit, comprising at least an air pump, a pressure sensor, and a valve, wherein the air pump is used to provide airflow, the pressure sensor is used to monitor the pressure of the airflow, and the valve adjusts the air pressure according to feedback requirements; The nebulization unit includes a nebulizer and a nebulized drug storage container. The nebulizer is used to convert liquid drugs into atomized particles through airflow, and the atomized particles are delivered to the user's respiratory tract through a connecting tube. A connecting tube connects the air pressure regulating unit to the nebulizing unit, so that airflow is delivered from the air pressure regulating unit to the nebulizing unit, and then the nebulizing unit outputs atomized particles to the user's respiratory tract. The control unit, connected to the air pressure regulating unit and the nebulizing unit, is used to adjust the airflow, air pressure and nebulization output of the breathing trainer according to the breathing characteristics.

Images