Respiration triggering method and system based on multi-signal fusion and adaptive threshold

Abstract

The invention relates to the technical field of medical instruments, in particular to a respiration triggering method and system based on multi-signal fusion and a self-adaptive threshold value, specifically, a multi-signal fusion triggering scoring method is adopted, and inspiration and expiration phase state judgment is conducted in combination with net flow, pressure and derivatives of the net flow, the pressure and the derivatives of the net flow, the pressure and the derivatives of the pressure. The inspiration and expiration phase switching needs to meet triple conditions of a threshold value, a derivative and duration time; leakage estimation and baseline correction are carried out in an expiration stable section, and in addition, a confirmation mechanism and a non-response period are introduced, so that phase jitter is reduced; and meanwhile, the threshold value and the weight can be dynamically and adaptively updated along with the respiratory statistical characteristics, an anomaly detection and rollback strategy is executed, and the continuity and the safety of ventilation are guaranteed under the conditions of cough, swallowing or sensor failure. Furthermore, the system adopts a modular structure, so that embedded implementation and upgrading are facilitated.

Description

Technical Field

[0001] This invention relates to the field of medical device technology, specifically to a respiratory triggering method and system based on multi-signal fusion and adaptive threshold. Background Technology

[0002] As a core life support device in intensive care, surgical anesthesia, and home non-invasive ventilation, the triggering mechanism of a ventilator directly determines the synchronicity between the patient and the ventilator. The core task of triggering is to ensure that the ventilator responds with minimal delay and assists breathing when the patient makes a spontaneous inspiratory effort. Excessive triggering delay requires the patient to expend extra respiratory effort, leading to respiratory muscle fatigue; overly sensitive or false triggering can cause hyperventilation and respiratory discomfort. Studies have shown that trigger asynchrony is closely associated with poor patient outcomes, prolonged ventilator use, endotracheal intubation dependence, and increased mortality.

[0003] Currently, common triggering methods mainly include traffic triggering and stress triggering:

[0004] (1) Flow triggering: The ventilator is activated by detecting a drop in airway flow exceeding a set threshold during inspiration. The threshold is usually set at 1–3 L / min. This method is simple to implement, but it is highly susceptible to air leakage and baseline drift, and its performance is significantly reduced in non-invasive ventilation scenarios.

[0005] (2) Pressure triggering: Inspiration is determined by detecting a drop in airway pressure exceeding a threshold, which is usually set to -1 to -2 cmH2O. This method is more sensitive to low-flow inspiration, but performs poorly when there is air leakage, changes in PEEP (positive end-expiratory pressure), or unstable pressure baseline.

[0006] Both of the above methods have the following shortcomings:

[0007] (1) Fixed thresholds lack flexibility: Clinical patients vary greatly, and fixed thresholds cannot take into account multiple groups such as adults, children, and newborns;

[0008] (2) Highly sensitive to air leakage: Air leakage is common in non-invasive masks and pipeline systems, causing flow baseline drift, which often leads to false triggering or leakage triggering;

[0009] (3) Susceptible to noise interference: Ambient airflow, vibration, and electromagnetic interference may all cause signal pulses, which may be incorrectly identified as air intake triggering;

[0010] (4) Unstable phase switching: During the alternation of inhalation and exhalation, the signal exhibits complex fluctuations. The single threshold method causes state machine jitter, resulting in poor patient comfort.

[0011] (5) Lack of adaptability of the algorithm: The patient's condition changes dynamically, and traditional methods are difficult to automatically adjust parameters according to the recent respiratory status;

[0012] (6) Inadequate handling of abnormal working conditions: Short-term abnormal signals such as coughing, swallowing, and airway obstruction are not recognized, directly leading to erroneous triggering.

[0013] In recent years, some studies have explored joint determination of flow and pressure, trend detection based on signal derivatives, and multi-parameter fusion triggering. However, most of these studies remain in the laboratory verification stage, lacking effective real-time leakage compensation mechanisms, robust adaptive threshold update strategies, and systematic anomaly detection and safety rollback functions. Summary of the Invention

[0014] The purpose of this invention is to provide a respiratory triggering method and system based on multi-signal fusion and adaptive threshold, which improves the sensitivity and specificity of inspiratory triggering in the presence of loop leakage, noise interference, baseline drift and weak patient inspiration; at the same time, it reduces jitter during the transition between inspiratory and expiratory phases, and enhances the synchronization and comfort between the patient and the ventilator.

[0015] To achieve the above objectives, the present invention provides a respiratory triggering method and system based on multi-signal fusion and adaptive thresholding, comprising the following steps:

[0016] Step 1: Signal acquisition and preprocessing, i.e., real-time acquisition of airway flow and pressure signals, followed by low-pass filtering and derivative calculation to obtain net flow, flow derivative and pressure derivative;

[0017] Step 2: Leakage and baseline estimation;

[0018] Step 3: Based on the net flow rate, pressure and their derivatives, a weighted combination is performed, and the score is calculated to obtain the trigger score S(t);

[0019] Step 4: When the trigger score S(t) continuously exceeds the threshold for a certain period of time, it is determined that the inspiratory phase has begun. In order to reduce the probability of false triggering, multiple signal judgment conditions are added to the inspiratory trigger to avoid repeated triggering in the same respiratory cycle.

[0020] Step 5: When the inspiratory phase lasts longer than the minimum inspiratory time, and the net flow rate turns from negative to positive and remains positive for a certain period of time, the system determines that inspiration has ended and switches to the expiratory phase. To improve the accuracy of respiratory phase switching, multiple conditions are introduced for expiratory triggering.

[0021] Step 6: Introduce confirmation sub-states and refractory periods to suppress misjudgments and phase jitter caused by short-term noise;

[0022] Step 7: Perform adaptive updates, dynamically adjusting the thresholds and weights based on recent respiratory statistical characteristics;

[0023] Step 8: Identify abnormal signals and trigger the shielding mechanism, and if necessary, revert to timed ventilation mode.

[0024] Optionally, during step 2, a leakage model is estimated during the stable expiratory phase: The net flow rate is then calculated to update the baseline net flow rate calculation formula: .

[0025] Optionally, in step 3, the net flow rate, flow derivative, pressure difference, and pressure derivative are weighted according to a set weight; the functional expression of the trigger score S(t) is defined as follows:

[0026]

[0027] in , , , As weight, This refers to the airway pressure difference.

[0028] Optionally, in step 4, the duration of entering the inspiratory phase is determined by the trigger score S(t) exceeding the threshold and lasting ≥50 ms. The multi-signal determination conditions for inspiratory triggering are as follows: ① Net flow rate It should be less than the set negative threshold. (Indicates flow in a clear inspiratory direction); ② Flow derivative condition: Flow derivative It should be greater than the set threshold. (A clear upward trend in flow rate indicates the start of inhalation); ③ Pressure conditions: pressure signal or pressure derivative A decrease occurs or the rate of decrease exceeds a set threshold. This indicates a decrease in airway pressure, consistent with the physiological characteristics of inspiratory breathing; ④ Time condition: The interval between the trigger signal and the previous trigger event should be greater than the refractory period (e.g., 200 ms) to avoid repeated triggering within the same respiratory cycle. When at least three of the above conditions are met simultaneously, the system determines it as a valid inspiratory trigger event.

[0029] Optionally, the minimum inhalation time in step 5 is 0.5s, and the multi-signal judgment conditions triggered by exhalation are as follows: ① Flow reversal condition: net flow rate The sign changes from negative to positive, and after the reversal, the average flow exceeds the threshold. ;② Flow derivative condition: Flow derivative A trend that changes from positive to negative indicates a decrease in inspiratory flow rate; ③ Pressure trend condition: airway pressure A rise or return to baseline at the end of inspiration indicates the end of lung filling; ④ Hysteresis protection condition: After the net flow rate turns positive, it needs to be maintained for 50-100 ms to prevent false switching caused by short-term signal fluctuations.

[0030] When the above primary criterion (① or ②) and any auxiliary criterion (③ to ④) are satisfied simultaneously, the system confirms the end of inhalation and switches to the expiratory phase.

[0031] Optionally, the adaptive update process in step 7 includes the following steps:

[0032] Step 7.1: Record the peak flow rate over the last 10 respiratory cycles. Valley value and baseline fluctuation range;

[0033] Step 7.2: Dynamically adjust the trigger threshold and hysteresis using an exponentially weighted moving average; the defined function expression is as follows:

[0034]

[0035] in Can be the most recent effective period Or based on statistics of peak-to-valley differences, Take 0.05;

[0036] Step 7.3: Adjust using an unsupervised / weakly supervised adaptive strategy Small compensation adjustment ,in It is the i-th feature. The learning rate; the adjustment range each time. ,in To calculate the weight of the i-th signal in the fusion scoring, This is the update amount for that weight. This is the preset maximum adjustment step size;

[0037] Step 7.4: When there is chronic air leakage or a shift in the patient's respiratory characteristics, and The settings will be adjusted gradually to maintain stable trigger sensitivity and specificity.

[0038] Optionally, the abnormal signals identified in step 8 include: coughing, sensor failure, and prolonged lack of triggering;

[0039] Cough is characterized by a short-duration, high-amplitude, high-frequency energy surge, with a short duration. The formula for calculating the short-duration energy is as follows: ,like A cough event is defined as lasting less than 300ms.

[0040] Sensor failures include saturation failure and flatness failure. Saturation failure is defined as the signal remaining close to the maximum or minimum value of the ADC for a long time, while flatness failure is defined as the signal having almost no fluctuation over a long period of time.

[0041] The method for determining if a device has not been triggered for an extended period is as follows:

[0042] Define a trigger timeout timer If the duration since the last effective inhalation trigger meets the following condition... If it is determined that there has been no trigger for a long time, then it is considered that there has been no trigger for a long time. Take 3-5 seconds.

[0043] Furthermore, this invention also proposes a breathing triggering system based on multi-signal fusion and adaptive threshold, used in the breathing triggering method based on multi-signal fusion and adaptive threshold. The system includes the following modules: a signal acquisition module, a preprocessing module, a leakage estimation module, a fusion scoring module, a state machine judgment module, an adaptive update module, an anomaly detection module, and a safety rollback mechanism judgment module. Flow and pressure data are preprocessed after flowing into the preprocessing module via the signal acquisition module. After leakage estimation and fusion scoring, the data enters the state machine judgment module for inspiratory triggering and expiratory switching state judgment. During execution, adaptive updates and anomaly detection adjustments are performed.

[0044] This invention provides a respiratory triggering method and system based on multi-signal fusion and adaptive thresholding. Specifically, it employs a multi-signal fusion triggering scoring method, combining net flow rate, pressure, and their derivatives to determine the inspiratory and expiratory phases. The transition between inspiratory and expiratory phases must satisfy three conditions: threshold, derivative, and duration. Leakage estimation and baseline correction are performed during the stable expiratory phase. Furthermore, a confirmation mechanism and refractory period are introduced to reduce phase jitter. Simultaneously, the threshold and weights can be dynamically and adaptively updated based on respiratory statistical characteristics, implementing anomaly detection and fallback strategies to ensure ventilation continuity and safety in the event of coughing, swallowing, or sensor failure. Moreover, the system adopts a modular structure, facilitating embedded implementation and upgrades. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 This is a schematic diagram of a breathing triggering method based on multi-signal fusion and adaptive threshold according to the present invention.

[0047] Figure 2 This is a schematic diagram of the adaptive adjustment process of a breathing triggering method based on multi-signal fusion and adaptive threshold according to the present invention.

[0048] Figure 3This is a schematic diagram of the abnormal detection process of a breathing triggering method based on multi-signal fusion and adaptive threshold according to the present invention.

[0049] Figure 4 This is a schematic diagram of the structural composition of a respiratory triggering system based on multi-signal fusion and adaptive threshold according to the present invention. Detailed Implementation

[0050] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0051] This invention provides a respiratory triggering method and system based on multi-signal fusion and adaptive threshold, comprising the following steps:

[0052] Step 1: Signal acquisition and preprocessing, i.e., real-time acquisition of airway flow and pressure signals, followed by low-pass filtering and derivative calculation to obtain net flow, flow derivative and pressure derivative;

[0053] Step 2: Leakage and baseline estimation;

[0054] Step 3: Based on the net flow rate, pressure and their derivatives, a weighted combination is performed, and the score is calculated to obtain the trigger score S(t);

[0055] Step 4: When the trigger score S(t) continuously exceeds the threshold for a certain period of time, it is determined that the inspiratory phase has begun. In order to reduce the probability of false triggering, multiple signal judgment conditions are added to the inspiratory trigger to avoid repeated triggering in the same respiratory cycle.

[0056] Step 5: When the inspiratory phase lasts longer than the minimum inspiratory time, and the net flow rate turns from negative to positive and remains positive for a certain period of time, the system determines that inspiration has ended and switches to the expiratory phase. To improve the accuracy of respiratory phase switching, multiple conditions are introduced for expiratory triggering.

[0057] Step 6: Introduce confirmation sub-states and refractory periods to suppress misjudgments and phase jitter caused by short-term noise;

[0058] Step 7: Perform adaptive updates, dynamically adjusting the thresholds and weights based on recent respiratory statistical characteristics;

[0059] Step 8: Identify abnormal signals and trigger the shielding mechanism, and if necessary, revert to timed ventilation mode.

[0060] Please see Figures 1 to 4 The following provides further explanation with reference to specific embodiments and execution steps:

[0061] 1. Respiratory trigger phase determination

[0062] (1) Sampling and preprocessing: The flow signal was acquired at 100 Hz. and pressure signals After filtering out noise above 10Hz using a second-order IIR filter, the filtered signal can be expressed as:

[0063]

[0064] (2) Leakage estimation: A linear leakage model was fitted using the least squares method during the stable expiratory phase. ,in and Calculated using the least squares method: Online updates are robust, updating every respiratory cycle or every 5 seconds. , Using the root mean square of the residuals To determine the fit quality for N samples, if... If the threshold is exceeded, no update will be performed. This is to prevent abnormal samples from contaminating the system.

[0065] (3) Net flow calculation: through Obtain net flow;

[0066] (4) Integrated scoring calculation: Net flow, flow derivative, pressure difference, and pressure derivative are weighted according to set weights; scoring function is defined:

[0067] in , , , As weight, This refers to the airway pressure difference.

[0068] (5) Inspiratory Triggering Condition: When the score S(t) exceeds the threshold and lasts for ≥50 ms (i.e., 5 consecutive sampling points at a sampling rate of 100 Hz) and the system is not currently in the refractory period, it is determined to be an inspiratory triggering event, and the system enters the inspiratory phase. To reduce the probability of false triggering, this embodiment adds multiple signal determination conditions for inspiratory triggering, as follows: ① Net flow rate It should be less than the set negative threshold. (Indicates flow in a clear inspiratory direction); ② Flow derivative condition: Flow derivative It should be greater than the set threshold. (A clear upward trend in flow rate indicates the start of inhalation); ③ Pressure conditions: pressure signal or pressure derivative A decrease occurs or the rate of decrease exceeds a set threshold. This indicates a decrease in airway pressure, consistent with the physiological characteristics of inspiratory breathing; ④ Time condition: The interval between the trigger signal and the previous trigger event should be greater than the refractory period (e.g., 200 ms) to avoid repeated triggering within the same respiratory cycle. When at least three of the above conditions are met simultaneously, the system determines it as a valid inspiratory trigger event. By using multi-condition fusion for determination, false triggering caused by single signal fluctuations, noise, or short-term coughing can be significantly reduced, improving trigger accuracy and stability.

[0069] (6) Expiratory phase switching condition: When the inspiratory time lasts ≥0.5s and the net flow rate changes from negative to positive and remains positive for a certain period of time (e.g., 50-100 ms), the system determines that the inspiratory phase has ended and switches to the expiratory phase. To improve switching accuracy, this embodiment introduces the following multi-condition determination: ① Flow reversal condition: Net flow rate The sign changes from negative to positive, and after the reversal, the average flow exceeds the threshold. ;② Flow derivative condition: Flow derivative A trend that changes from positive to negative indicates a decrease in inspiratory flow rate; ③ Pressure trend condition: airway pressure A rise or return to baseline at the end of inspiration indicates the end of lung filling; ④ Hysteresis protection condition: After the net flow rate turns positive, it needs to be maintained for 50-100 ms to prevent false switching caused by short-term signal fluctuations.

[0070] When the primary criterion (① or ②) and any of the secondary criters (③ to ④) are simultaneously satisfied, the system confirms the end of inspiration and switches to the expiratory phase. This design ensures a smooth and accurate transition between respiratory phases, avoiding the risks of premature switching or overinspiration.

[0071] 2. Adaptive Update

[0072] (1) Data statistics: Record the peak flow rate over the past 10 respiratory cycles. Valley value And baseline fluctuation range.

[0073] (2) Threshold Update: The trigger threshold and hysteresis are dynamically adjusted using an exponentially weighted moving average (EWMA); the defined function expression is as follows:

[0074]

[0075] in Can be the most recent effective period (Peak score) or a statistic based on the peak-to-valley difference; Take 0.05.

[0076] Weight optimization: Adaptive unsupervised / weakly supervised strategy for weight adjustment. (Correct / Incorrect), adjust with minor compensation. ,in It is the i-th feature. This is the learning rate. To prevent over-adjustment within a single cycle, the magnitude of each adjustment is limited. ,in To calculate the weight of the i-th signal in the fusion scoring, This is the update amount for that weight. This is the preset maximum adjustment step size; The value range is 0.01 to 0.05, and it can be calibrated according to the system noise characteristics and response requirements to achieve smooth convergence of weights and stability of trigger judgment. At the same time, a protection range is set for the initial weights to avoid being skewed by abnormal data. The weights are adjusted based on recent judgment accuracy. , , , Fine-tuning.

[0077] Achieved effect: Through the above adaptation, when chronic air leakage or drift in patient breathing characteristics (e.g., flow rate / force), and The settings will be adjusted gradually to maintain stable trigger sensitivity and specificity.

[0078] 3. Anomaly detection and rollback mechanism

[0079] An anomaly detection mechanism is introduced to improve security.

[0080] Cough detection: Cough is characterized by a short-duration, high-amplitude, high-frequency energy surge, with a short duration (typically <300ms). The short-duration energy is calculated on the signal after bandpass at 5–30 Hz (window size is...). ≈ 50–150 ms). The formula for calculating short-time energy is as follows: ,like Furthermore, a coughing event can be identified as lasting less than 300ms, after which the trigger is disabled. And mark the waveform segment in the log.

[0081] (2) Sensor failure: Saturation failure is defined as the signal remaining close to the maximum or minimum value of the ADC for an extended period, possibly due to sensor short circuit, blockage, or circuit distortion. Flatness failure is defined as the signal remaining almost unchanged for a considerable period, possibly due to sensor detachment, conduit blockage, or hardware freezing. The method for determining saturation failure is as follows: And satisfy the duration It can be determined to be saturated. Among them... This is the maximum value that the ADC can measure. For safety margin, The time limit is set to 0.5 seconds. The flatness detection and determination method is as follows:

[0082]

[0083] in The standard deviation within the window. The preset threshold is N, where N is the number of samples collected within 1 second. The fallback mechanism is as follows: the system immediately triggers an alarm (audio-visual + interface prompt). Next, it will enter a timed ventilation mode (e.g., at a fixed frequency of 12–16 breaths / minute, with a fixed tidal volume output) to ensure patient safety.

[0084] (3) Prolonged lack of triggering: In cases where the patient's breathing is extremely weak or the trigger sensitivity is set too high, a valid inspiratory trigger may not be detected for an extended period. If supplemental ventilation is not provided promptly, there is a risk of hypoxia. Judgment method: Define a trigger timeout timer. If the duration since the last effective inhalation trigger meets the following condition... If it is determined that there has been no trigger for a long time, then it is considered that there has been no trigger for a long time. Set the time to 3-5 seconds. The time can be adjusted according to the actual clinical application. If there is no effective trigger after 3 seconds, the system will be forced to enter the backup breathing mode.

[0085] Furthermore, this invention also proposes a respiratory triggering system based on multi-signal fusion and adaptive threshold, such as... Figure 4 As shown, the breathing triggering method based on multi-signal fusion and adaptive threshold includes the following modules: a signal acquisition module, a preprocessing module, a leakage estimation module, a fusion scoring module, a state machine judgment module, an adaptive update module, an anomaly detection module, and a safety rollback mechanism judgment module. Flow and pressure data are preprocessed after flowing into the preprocessing module through the signal acquisition module, and after leakage estimation and fusion scoring, they enter the state machine judgment module for inspiratory triggering and expiratory switching state judgment. During execution, adaptive updates and anomaly detection adjustments are performed.

[0086] By introducing sensor failure detection and long-term no-trigger detection, the system can automatically enter a safety mode under abnormal circumstances, ensuring that the patient receives minimum ventilation support and avoiding the risk of suffocation caused by sensor malfunction or lack of respiratory triggering by the patient, thereby improving the robustness and safety of the system.

[0087] The above description discloses only one or more preferred embodiments of the present invention, and should not be construed as limiting the scope of the present invention. Those skilled in the art will understand that all or part of the processes of the above embodiments can be implemented, and equivalent changes made in accordance with the claims of the present invention are still within the scope of the invention.

Claims

1. A respiratory triggering method based on multi-signal fusion and adaptive threshold, characterized in that, Includes the following steps: Step 1: Signal acquisition and preprocessing, i.e., real-time acquisition of airway flow and pressure signals, followed by low-pass filtering and derivative calculation to obtain net flow, flow derivative and pressure derivative; Step 2: Leakage and baseline estimation; Step 3: Based on the net flow rate, pressure and their derivatives, a weighted combination is performed, and the score is calculated to obtain the trigger score S(t); Step 4: When the trigger score S(t) continuously exceeds the threshold for a certain period of time, it is determined that the inspiratory phase has begun; in order to reduce the probability of false triggering, multiple signal determination conditions are added to the inspiratory trigger to avoid repeated triggering in the same respiratory cycle; Step 5: When the inspiratory phase lasts longer than the minimum inspiratory time and the net flow rate turns from negative to positive and remains so for a certain period of time, the system determines that the inspiratory phase has ended and switches to the expiratory phase; to improve the accuracy of respiratory phase switching, multiple conditions are introduced for expiratory triggering. Step 6: Introduce confirmation sub-states and refractory periods to suppress misjudgments and phase jitter caused by short-term noise; Step 7: Perform adaptive updates, dynamically adjusting the thresholds and weights based on recent respiratory statistical characteristics; Step 8: Identify abnormal signals and trigger the shielding mechanism, and if necessary, revert to timed ventilation mode.

2. The breathing triggering method based on multi-signal fusion and adaptive threshold as described in claim 1, characterized in that, During step 2, the leakage model is estimated during the stable expiratory phase: The net flow rate is then calculated to update the baseline net flow rate calculation formula: .

3. The breathing triggering method based on multi-signal fusion and adaptive threshold as described in claim 2, characterized in that, In step 3, the net flow rate, flow derivative, pressure difference, and pressure derivative are weighted according to the set weights; the functional expression of the trigger score S(t) is defined as follows: ； in , , , As weight, This refers to the airway pressure difference.

4. The breathing triggering method based on multi-signal fusion and adaptive threshold as described in claim 3, characterized in that, In step 4, the duration of entering the inspiratory phase is determined by the trigger score S(t) exceeding the threshold and lasting ≥50 ms. Furthermore, the multi-signal determination conditions for inspiratory triggering in step 4 are as follows: ①Net flow It should be less than the set negative threshold. (Indicates a clear inhalation direction of airflow); ② Flow derivative condition: Flow derivative It should be greater than the set threshold. (A clear upward trend in airflow indicates the start of inhalation); ③ Pressure conditions: pressure signal or pressure derivative A decrease occurs or the rate of decrease exceeds a set threshold. This indicates a decrease in airway pressure, which is consistent with the physiological characteristics of inhalation. ④ Time condition: The interval between the trigger signal and the previous trigger event should be greater than the refractory period (e.g., 200 ms) to avoid repeated triggering in the same respiratory cycle; When at least three of the above conditions are met simultaneously, the system determines it as a valid inhalation trigger event; The minimum inhalation time in step 5 is 0.5 seconds. Furthermore, the multi-signal determination conditions triggered by exhalation in step 5 are as follows: ① Condition for reversal of flow direction: net flow The sign changes from negative to positive, and after the reversal, the average flow exceeds the threshold. ; ② Flow derivative condition: Flow derivative A trend of turning from positive to negative indicates a decrease in inhalation speed; ③ Pressure trend condition: airway pressure A rise or return to baseline at the end of inspiration indicates the end of lung filling. ④ Hysteresis protection conditions: After the net flow turns positive, it needs to be maintained for 50-100 ms to prevent false switching caused by short-term signal fluctuations; When the primary criterion ① or ② and any secondary criterion ③ or ④ are satisfied simultaneously, the system confirms the end of inhalation and switches to the expiratory phase.

5. The breathing triggering method based on multi-signal fusion and adaptive threshold as described in claim 4, characterized in that, The adaptive update process in step 7 includes the following steps: Step 7.1: Record the peak flow rate over the last 10 respiratory cycles. Valley value and baseline fluctuation range; Step 7.2: Dynamically adjust the trigger threshold and hysteresis using an exponentially weighted moving average; the defined function expression is as follows: ； in Can be the most recent effective period Or based on statistics of peak-to-valley differences; Take 0.05; Step 7.3: Adjust using an unsupervised / weakly supervised adaptive strategy Small compensation adjustment ,in It is the i-th feature. The learning rate; the adjustment range each time. ,in To calculate the weight of the i-th signal in the fusion scoring, This is the update amount for that weight. This is the preset maximum adjustment step size; Step 7.4: When there is chronic air leakage or a shift in the patient's respiratory characteristics, and The settings will be adjusted gradually to maintain stable trigger sensitivity and specificity.

6. The breathing triggering method based on multi-signal fusion and adaptive threshold as described in claim 5, characterized in that, Abnormal signals identified in step 8 include: coughing, sensor failure, and prolonged periods without triggering; Cough is characterized by a short-duration, high-amplitude, high-frequency energy surge, with a short duration. The formula for calculating the short-duration energy is as follows: ,like A cough event is defined as lasting less than 300ms. Sensor failures include saturation failure and flatness failure. Saturation failure is defined as the signal remaining close to the maximum or minimum value of the ADC for a long time, while flatness failure is defined as the signal having almost no fluctuation over a long period of time. The method for determining if a device has not been triggered for an extended period is as follows: Define a trigger timeout timer If the duration since the last effective inhalation trigger meets the following condition... If it is determined that there has been no trigger for a long time, then it is considered that there has been no trigger for a long time. Take 3-5 seconds.

7. A respiratory triggering system based on multi-signal fusion and adaptive threshold, used in the respiratory triggering method based on multi-signal fusion and adaptive threshold as described in any one of claims 1 to 6, characterized in that, Includes the following modules: The system includes a signal acquisition module, a preprocessing module, a leakage estimation module, a fusion scoring module, a state machine judgment module, an adaptive update module, an anomaly detection module, and a safety rollback mechanism judgment module. Flow and pressure data are preprocessed after entering the preprocessing module through the signal acquisition module. After leakage estimation and fusion scoring, the data enters the state machine judgment module for inhalation triggering and exhalation switching state judgment. During execution, adaptive updates and anomaly detection adjustments are performed.

Images